Five decisions, put through the guide.

1. Turn 1. Operator agent, answering from the fact sheet

   Use Frame Free in this PDF. Ask me one question at a time.

   Our warehouse management system contract runs out on 30 November and the vendor says the version we’re on is losing support, so I need to pick between two replacement platforms by the end of October. Picking errors are costing us money and I want this sorted before the Christmas peak.

2. Turn 2. Method agent, following the guide

   Thanks. Let me confirm what you’ve given me before we go further.

   Decision as stated: choose one of two replacement warehouse management platforms by the end of October.

   Trigger as stated: your current WMS contract ends on 30 November, and the vendor says your version is losing support. You also want this settled before the Christmas peak.

   One thing to sharpen before we move on. “Picking errors are costing us money” is the problem you want fixed, but as it stands it’s a category, not a measurable scenario. So, first question:

   Roughly how often do picking errors happen now, what does each one cost you, and how do you know? Is that a figure someone has measured, or a sense of the pain?

3. Turn 3. Operator agent, answering from the fact sheet

   We ship about 10,000 order lines a week. Over the last quarter the picking error rate was 2.1%, so roughly 210 wrong lines a week. On cost, the hard number I have is credit notes: we issued $14,200 in credit notes for picking errors last quarter. I don’t have a cost-per-error figure beyond that, and I don’t know what else each error costs us on top of the credit notes. So the rate and the credit note total are figures from the quarter, not a gut feel, but the full cost per error isn’t something I’ve measured.

4. Turn 4. Method agent, following the guide

   That’s useful, and it’s more measured than most people have at this stage. Here is what I’m holding:

   • GIVEN: about 10,000 order lines a week; a 2.1% picking error rate last quarter; $14,200 in credit notes for picking errors last quarter.
   • DERIVED: about 210 wrong lines a week. If the quarter was 13 weeks, that is roughly 2,730 wrong lines and about $5 in credit notes per wrong line. That figure is a floor, because it counts only credit notes.
   • UNKNOWN: what else each error costs you beyond the credit note.

   So the decision and trigger are clear: two platforms, chosen by the end of October, because the contract ends on 30 November and your version is losing support.

   Second question:

   Suppose the platform choice disappeared tomorrow, for example because your current vendor agreed to keep supporting your version for another two years. What larger question would still be sitting there? Put another way, would those 210 wrong lines a week go away with the decision, or would they still need solving?

5. Turn 5. Operator agent, answering from the fact sheet

   No, the 210 wrong lines a week wouldn’t go away just because the platform decision went away. If the current vendor kept supporting us, I’d still be sitting on a 2.1% error rate and the credit notes that come with it, and I’d still want it fixed before the December peak. So the bigger question is how we get picking errors down. I’ve been treating a new platform as the way to do that, because both vendors showed in their demonstrations that their systems would cut errors. But that comes from the demos only. I haven’t checked it any further than that.

6. Turn 6. Method agent, following the guide

   That changes the shape of it. The platform choice is the surface. The decision underneath is how you get picking errors down from 2.1% before the December peak, and a new platform is one means to that, not the outcome itself.

   I’m noting one thing for later rather than testing it now. The belief that a new platform would cut your errors comes from the two vendor demonstrations, and you’ve said you haven’t checked it further. I’ll hold that as:

   • GIVEN: both vendors showed in their demonstrations that their systems would cut errors.
   • UNKNOWN: whether either system would cut errors in your warehouse, and by how much.

   We’ll come back to that. First, the option set. Right now it’s framed as platform A or platform B.

   Third question:

   What genuine third path have you not been treating as real? That could be keeping the current position for a while, for example asking the current vendor about an extension or extended support past 30 November. It could be working on the picking errors directly before or without changing platform. Or it could be a smaller version of one of the two. I’m not asking you to judge it yet, only to name one you’d be willing to take seriously.

7. Turn 7. Operator agent, answering from the fact sheet

   The one I’d take seriously is staying with our current vendor on their newer version. Their account manager sent us a quote in September to upgrade to it. I haven’t opened it, so I don’t know what it costs or what it includes. I’d been treating it as a choice between A and B and hadn’t counted that quote as a real option, but I’d be willing to.

8. Turn 8. Method agent, following the guide

   Good. That’s a genuine third path, and it was already sitting in your inbox. The option set is now three, not two:

   • Platform A
   • Platform B
   • Staying with your current vendor and moving to their newer version

   I’m not weighing it against the other two. I’m only recording what we know about it:

   • GIVEN: your current vendor’s account manager sent a quote in September to upgrade to their newer version.
   • UNKNOWN: what the quote costs, what it includes, and whether the newer version would do anything about picking errors.

   Now the belief that carries the decision. So far everything has pointed at the platform as the way to bring errors down. Fourth question:

   Finish this sentence in your own words: “Changing or upgrading the platform makes sense provided that …” Then tell me what you have actually seen, measured or checked that supports it, apart from the vendor demonstrations.

9. Turn 9. Operator agent, answering from the fact sheet

   “Changing or upgrading the platform makes sense provided that the new system actually cuts our picking errors, and it’s in and working before the December peak.”

   As for what I’ve actually seen, measured or checked that supports the first part: nothing beyond the vendor demonstrations. I haven’t trialled either platform, spoken to their customers or tested them against our own orders. The peak timing is the part I have checked, from last year’s sales data, but the claim that a new platform would bring our errors down rests on the demos alone.

10. Turn 10. Method agent, following the guide

    That sentence has two conditions in it, and you’ve told me they rest on different evidence. So I’ll split them.

    The load-bearing one is the first: “the new system actually cuts our picking errors.” If that’s wrong, the timing doesn’t matter, because a platform that’s in and working before the peak would still leave you at 2.1%. Here is how I’m holding it:

    • GIVEN: both vendors showed in their demonstrations that their systems would cut errors.
    • GIVEN: you haven’t trialled either platform, spoken to their customers or tested them against your own orders.
    • UNKNOWN: whether any of the three options would cut errors in your warehouse, and by how much. That includes the current vendor’s newer version, which nobody has looked at yet.

    The second condition, being in and working before the December peak, is one I’ll treat as a constraint. You’ve checked the peak timing from last year’s sales data, so that part has evidence behind it. How long each option would take to get in and working hasn’t come up yet.

    Fifth question. You’ve been working to three dates or positions that act as fixed:

    1. The contract ends on 30 November.
    2. Your current version is losing support.
    3. The choice has to be made by the end of October.

    For each one, is it verified or assumed? Specifically: have you read the 30 November date in the contract itself, do you know the actual date support stops and what “losing support” means in practice, and who set the end-of-October deadline? If there’s a budget limit you’ve been working within, tell me where that came from too.

11. Turn 11. Operator agent, answering from the fact sheet

    Taking them in order.

    1. The 30 November date: assumed. Nobody has read the renewal or termination clauses in the contract, me included. I’ve been working on the basis that we can’t use the current system after 30 November, but I haven’t checked that against the contract.

    2. Losing support: assumed. It comes from a call and a follow-up email from the current vendor’s account manager, saying our installed version “moves to limited support”. I don’t know what limited support means in practice, and I don’t know the actual date it starts.

    3. The end-of-October deadline: that’s my own target. I want a contract signed by 31 October. No one else set it.

    On budget, we have $90,000 for systems projects this year. That one is checked: the managing director approved it.

12. Turn 12. Method agent, following the guide

    That’s a clear split, and it moves more than it might look like. Here is how I’m marking the constraints:

    • 30 November contract end: ASSUMED. Nobody has read the renewal or termination clauses, so what happens to your use of the current system after that date is not yet established.
    • Losing support: ASSUMED. The source is a call and a follow-up email from the account manager saying your version “moves to limited support”. What that means in practice, and when it starts, is not known.
    • End-of-October deadline: VERIFIED as your own target, not an external constraint. You set it, and no one else did. As far as I can tell it rests on the 30 November date, which is itself assumed.
    • $90,000 systems budget for this year: VERIFIED. The managing director approved it.
    • December peak timing: VERIFIED. You checked it against last year’s sales data.

    So two of the three dates you were treating as fixed haven’t been checked, and the third is one you set yourself.

    Sixth and last question: the evidence map. I’ve drafted it from what you’ve told me so far. Each claim carries one tag: GIVEN, DERIVED, INFERRED or UNKNOWN.

    • GIVEN: about 10,000 order lines a week; a 2.1% picking error rate last quarter; $14,200 in credit notes for picking errors last quarter; both vendors showed in demonstrations that their systems would cut errors; no trial, customer reference or test against your own orders for either platform; a September upgrade quote from your current vendor, not yet opened; the account manager’s “moves to limited support” call and email; the end-of-October target is yours; the $90,000 budget is approved; the peak timing checked from last year’s sales data.
    • DERIVED: about 210 wrong lines a week; roughly 2,730 wrong lines last quarter if it ran 13 weeks; about $5 in credit notes per wrong line, which is a floor because it counts credit notes only.
    • INFERRED: nothing yet. I haven’t put the platform-cuts-errors belief here, because the only support for it is the demos.
    • UNKNOWN: the full cost of each error beyond the credit note; whether any of the three options would cut errors in your warehouse, and by how much; what the September quote costs and includes; what the contract says about renewal, termination or use after 30 November; what “limited support” means and when it starts; how long each option would take to get in and working; what platforms A and B cost against the $90,000, which hasn’t come up.

    Is anything on that map wrongly tagged, and is anything material missing? In particular, do you know anything measured about what causes the picking errors, for example particular products, shifts, pickers or locations? Or is that unknown too?

13. Turn 13. Operator agent, answering from the fact sheet

    On the tagging, nothing looks wrong to me. A few things are missing, though.

    On what causes the picking errors, here is what I know:

    • About 60% of picking errors happen on the night shift. The night shift moved to mostly casual pickers in March 2026.
    • The racking was reconfigured in February 2026, and some bin labels were not updated.
    • Error rates were below 1% for the two years before March 2026, on the same software we’re running now.

    I can’t tell you how the 60% figure was worked out beyond that, and I don’t have a breakdown by product or by individual picker.

    On the platform costs, which you’ve marked as not having come up: Platform A is $68,000 to implement plus $2,900 a month, and Platform B is $41,000 to implement plus $3,600 a month, both on three-year terms. Over three years that’s $172,400 for A and $170,600 for B.

    One more unknown that isn’t on the map: I don’t know whether either new platform integrates with our accounting system without custom work.

14. Turn 14. Method agent, following the guide

    Thanks. Those three facts about the errors matter a lot, so I’ve added them, along with the platform costs and the integration question. That was the last of the six questions. Here is the framing handoff.

    1. Presented decision and trigger

    Choose between two replacement warehouse management platforms, with a contract signed by 31 October. The trigger is that your current WMS contract runs out on 30 November and the vendor says your version is losing support. You want it settled before the Christmas peak.

    2. Decision underneath

    How you get picking errors down from 2.1% before the December peak. A new platform is one way to do that. It is not the outcome itself. If the platform choice disappeared, the 210 wrong lines a week would still be there.

    3. Missing third option

    Staying with your current vendor and moving to their newer version. The account manager sent a quote for it in September. Nobody has opened it, so its cost and contents are not known. It was not being treated as a real option until now. The option set is three, not two: Platform A, Platform B, or the current vendor’s newer version.

    4. Load-bearing assumption

    In your words: “Changing or upgrading the platform makes sense provided that the new system actually cuts our picking errors.” The only support for this is the two vendor demonstrations. You have not trialled either platform, spoken to their customers or tested them against your own orders. What you told me in your last message bears directly on it. Error rates were below 1% for two years on the software you run now, and the rise came after the racking change in February 2026 and the move to mostly casual night-shift pickers in March 2026. That timing doesn’t prove what caused the errors, but it is the main evidence that bears on this assumption.

    5. Untested constraints

    • 30 November contract end: ASSUMED. Nobody has read the renewal or termination clauses, so what happens to your use of the current system after that date is not established.
    • Losing support: ASSUMED. The source is a call and a follow-up email from the account manager saying your version “moves to limited support”. What that means in practice, and when it starts, is not known.
    • 31 October signing deadline: VERIFIED as your own target, not an external constraint. It appears to rest on the 30 November date, which is assumed.
    • $90,000 systems budget for this year: VERIFIED. The managing director approved it.
    • December peak timing: VERIFIED. Checked against last year’s sales data.

    6. Evidence map

    • GIVEN: about 10,000 order lines a week; a 2.1% picking error rate last quarter; $14,200 in credit notes for picking errors last quarter; error rates below 1% for the two years before March 2026, on the current software; about 60% of picking errors happen on the night shift, though you don’t know how that figure was worked out; the night shift moved to mostly casual pickers in March 2026; the racking was reconfigured in February 2026 and some bin labels were not updated; both vendors showed in demonstrations that their systems would cut errors; no trial, customer reference or test against your own orders for either platform; a September upgrade quote from your current vendor, not yet opened; the account manager’s “moves to limited support” call and email; Platform A at $68,000 to implement plus $2,900 a month, and Platform B at $41,000 to implement plus $3,600 a month, both on three-year terms; the 31 October target is yours; the $90,000 budget is approved; the peak timing is checked against last year’s sales data.
    • DERIVED: about 210 wrong lines a week now. Below 1% would be fewer than 100 a week. Roughly 2,730 wrong lines last quarter if it ran 13 weeks. About $5 in credit notes per wrong line, which is a floor because it counts credit notes only. Three-year cost of $172,400 for A and $170,600 for B, a difference of $1,800. Implementation plus the first twelve months comes to $102,800 for A and $84,200 for B. How that lines up against this year’s $90,000 depends on when spending would fall, which hasn’t come up.
    • INFERRED: the rise from below 1% to 2.1% began around the racking change and the night-shift staffing change, on the same software. So at least part of the error problem may come from bin labelling and night-shift picking rather than the software. This is reasoned from the timing and the 60% figure. It has not been tested.
    • UNKNOWN: what is actually driving the errors, including whether they cluster on bins with labels that weren’t updated, and how the 60% night-shift figure was produced; the full cost of each error beyond the credit note; whether any of the three options would cut errors in your warehouse, and by how much; what the September quote costs and includes; what the contract says about renewal, termination or use after 30 November; what “limited support” means and when it starts; how long each option would take to get in and working; whether either new platform integrates with your accounting system without custom work.

    7. Reframed decision and the single evidence gap to verify next

    Reframed decision: how to bring picking errors back from 2.1% toward the below-1% level you had on the same software, before the December peak, and what part a platform change plays in that, if any. The timing should come from what the contract and the support terms actually say, not from the 30 November date as assumed.

    Evidence gap to verify next: what is driving the rise in errors since early 2026. Specifically, check whether the errors cluster on the bins whose labels weren’t updated after the February reconfiguration and on the night shift, and find out how the 60% figure was produced. If the errors trace mainly to those two changes, the role a new platform plays in fixing them changes, and so does what you need from any of the three options.

    That’s where Frame Free stops. I haven’t compared the three options, scored them or recommended one. The question is now framed, but the decision itself is still yours to make.

Written before the run. The operator agent was told never to volunteer its last section.

1. Turn 1. Operator agent, answering from the fact sheet

   Use Frame Free in this PDF. Ask me one question at a time.

   We keep delivering drawing packages late and two clients have complained this quarter, so I’m deciding whether to hire another senior structural engineer. I’d like to advertise the role before the end of the month.

2. Turn 2. Method agent, following the guide

   Let’s use Frame Free. Here is how I’ve read what you said, so we start from the same place:

   • Decision: whether to hire another senior structural engineer.
   • Trigger: drawing packages keep going out late, and two clients complained this quarter.
   • Timing you want: advertise the role before the end of the month.

   Question 1: the decision and the trigger.

   “Delivering late” covers a lot, so I’d like to pin it down before we go further. Over this quarter, roughly how many drawing packages went out late, out of how many in total, and by how much were they late? And what changed this quarter to make this live now, rather than last quarter? For example, was it the two complaints themselves, a change in workload, or someone leaving?

   Give me whatever figures you have, even rough ones. Just tell me which ones are estimates.

3. Turn 3. Operator agent, answering from the fact sheet

   I don’t have the figures broken out for this quarter on its own. What I do have is the last 12 months: 41 of the 63 projects we delivered were late, about 65%, and on average they were 9 working days late. Those come from our records, not estimates.

   As for what made it live now, it was the complaints. Two repeat clients wrote to us in the September quarter about late packages. I don’t know of a change in workload or anyone leaving that set it off.

4. Turn 4. Method agent, following the guide

   Thanks. That gives us a firmer starting point. Here is what I’m noting, and how I’m tagging it:

   • From your records: 41 of 63 projects delivered in the last 12 months were late, an average of 9 working days late.
   • Calculated from that: 41 out of 63 is about 65%, which matches your figure.
   • The trigger: two repeat clients wrote to you in the September quarter about late packages. You don’t know of a workload change or a departure that set it off.

   One thing stands out. Your figures cover 12 months, but the trigger is this quarter’s complaints. So the late delivery may not be new. What may be new is that clients have started saying so. I’m not drawing a conclusion from that. I’m keeping it visible because it may matter later.

   Question 2: the decision underneath.

   Suppose hiring a senior structural engineer were off the table, with no budget and no candidate. What problem would still be there? Put another way, if the hire is the means, what outcome is it supposed to change? Fewer late packages, keeping those two clients, taking on more work, or something else?

   Name the outcome as specifically as you can, even if you’re not sure the hire is the right way to get there.

5. Turn 5. Operator agent, answering from the fact sheet

   The late packages would still be there. That’s the problem. The outcome I want the hire to change is fewer packages going out late, because that’s what the two repeat clients complained about. I haven’t set a specific target for how many fewer, and taking on more work isn’t what’s driving this.

6. Turn 6. Method agent, following the guide

   That’s clear. Here is how I’m recording it:

   • The problem that would remain without the hire is late drawing packages.
   • The outcome you want the hire to change is fewer packages going out late, because that is what the two repeat clients complained about.
   • You haven’t set a target for how many fewer. I’m noting that as a gap, not a fault. It means “fewer late packages” can’t yet be measured against anything.
   • Taking on more work is not what’s driving this.

   So the decision underneath reads as: how do we cut the number of drawing packages that go out late? Hiring a senior structural engineer is one way to do that. It only works if a shortage of senior engineering time is what makes packages late. I’m not judging whether it is. We’ll come back to that when we get to assumptions.

   Question 3: the missing option.

   Right now the choice looks like “hire another senior structural engineer, or don’t.” Is there a third path you haven’t been treating as a real option? It has to be credible enough to take seriously later. Keeping things as they are for a set period counts. So does a smaller version of the hire, or anything that goes after late packages some other way.

   Name the one that feels most real to you, even if you’d currently rule it out. We won’t weigh it yet.

7. Turn 7. Operator agent, answering from the fact sheet

   Honestly, I haven’t been treating a third path as a real option. The two options I’ve had in front of me are hiring one senior structural engineer at about $145,000 a year plus super, or keeping the current team and accepting the late deliveries. I don’t have a third one in mind that I’d call credible, so I can’t name one for you.

8. Turn 8. Method agent, following the guide

   That’s a fair answer, and it tells us something useful. Here is how I’m recording it:

   • Option A: hire one senior structural engineer at about $145,000 a year plus super.
   • Option B: keep the current team and accept the late deliveries.
   • Third path: none named yet.

   Option B is already the “do nothing” path, so it is in the set. What is missing is anything between “hire a full senior engineer” and “accept the lateness as it is.”

   I’ll ask one follow-up and then we move on, whatever the answer.

   Follow-up on Question 3. Two shapes count as a genuine third path, and I’m not suggesting either is right:

   • A smaller version of the hire. Something less than one full-time senior engineer that still adds senior time.
   • A held position for a set period. Keep the current team for a fixed time, but not on the open-ended “accept it” terms of Option B.

   Does either of those, or any other route that goes after late packages without a full senior hire, seem credible enough to take seriously later? If nothing comes to mind, say so. I’ll record “no third path named” and we’ll go on to the assumption the decision rests on.

9. Turn 9. Operator agent, answering from the fact sheet

   No, nothing comes to mind. I haven’t looked at a part-time or smaller version of the hire, and I haven’t thought about holding the team as it is for a set period. I don’t have another route that I’d call credible, so please record “no third path named” and we can move on.

10. Turn 10. Method agent, following the guide

    Recorded. Here is where the option set stands:

    • Option A: hire one senior structural engineer at about $145,000 a year plus super.
    • Option B: keep the current team and accept the late deliveries.
    • Third path: none named. You haven’t looked at a part-time or smaller version of the hire, or at holding the team as it is for a set period.

    I’m leaving that gap in the frame as it is. I won’t fill it in for you.

    Question 4: the load-bearing assumption.

    Most decisions rest on one belief. If that belief turned out to be wrong, the decision would change. Try finishing this sentence in your own words:

    “Hiring another senior structural engineer makes sense provided that …”

    Then tell me what you have actually seen, measured or checked that supports it. For example, do your records show what held up the 41 late projects? I mean things like senior engineers’ time, checking and sign-off, waiting on information from clients or other consultants, drafting, or something else. If you haven’t looked at why they were late, say so. That is a useful answer too.

11. Turn 11. Operator agent, answering from the fact sheet

    “Hiring another senior structural engineer makes sense provided that the packages are late because we don’t have enough senior engineering time to get them done.”

    As for what I’ve actually checked, not much. I haven’t done a proper analysis of why the 41 projects were late. What I can tell you from what I know:

    • Engineer billable utilisation averaged 78% last financial year.
    • The project system shows packages sitting in “awaiting principal review” for an average of 6 working days. I review every package personally, and I also run business development.
    • Drafters regularly report waiting for markups before they can continue.
    • About 40% of the late projects had client-requested design changes after the fee was agreed. Most of those changes were absorbed without a variation.
    • Two engineers have said privately that they have spare time some weeks.

    I don’t know how many hours a week a new hire would actually free up.

12. Turn 12. Method agent, following the guide

    Thanks. That sentence is the load-bearing assumption, in your words:

    “Hiring another senior structural engineer makes sense provided that the packages are late because we don’t have enough senior engineering time to get them done.”

    You said you haven’t analysed why the 41 projects were late, so I’m recording that assumption as untested. Here is what you’ve told me that bears on it, kept as you stated it:

    • Engineer billable utilisation averaged 78% last financial year.
    • The project system shows packages in “awaiting principal review” for an average of 6 working days. You review every package yourself, and you also run business development.
    • Drafters regularly report waiting for markups before they can continue.
    • About 40% of the late projects had client-requested design changes after the fee was agreed, and most were absorbed without a variation.
    • Two engineers have said privately that they have spare time some weeks.
    • You don’t know how many hours a week a new hire would free up.

    I’ll be direct about one thing without drawing a conclusion. Several of these point somewhere other than a shortage of senior engineers’ time: the wait at principal review, the wait for markups, the scope changes and the reported spare time. Some may support your assumption too. None of them has been tied to specific late projects yet, so I’m not weighing them. I’m noting that the assumption is carrying the decision while the evidence you hold doesn’t clearly point one way.

    Follow-up on Question 4. One fact would show whether a hire could touch the 6-day wait at all. If you hired a senior structural engineer, would they be allowed to review and sign off packages, or would every package still come to you for principal review? Tell me whether that is a settled rule, such as a registration, insurance or client requirement, or simply how you have always worked.

13. Turn 13. Operator agent, answering from the fact sheet

    It’s partly a settled rule and partly an assumption. The settled part is that every package has to be signed off by a principal before it’s issued. Our professional indemnity insurer requires that for this practice, and I have checked it. The part I haven’t checked is that only I can do that sign-off. I’ve assumed it. Nobody has asked the insurer whether another registered engineer could do it. So as things stand, I don’t know whether a new senior engineer would be allowed to review and sign off packages, and on the way we work now, every package would still come to me.

14. Turn 14. Method agent, following the guide

    Thanks. That separates two things that were sitting together. Here is how I’m recording them:

    • Verified: every package must be signed off by a principal before it is issued. Your professional indemnity insurer requires this, and you have checked it.
    • Assumed: only you can do that sign-off. Nobody has asked the insurer whether another registered engineer could.
    • Consequence on current practice: every package would still come to you, hire or no hire. So on the way you work now, a new senior engineer would not touch the 6-day wait at principal review. That follows from what you’ve said. I’m not judging whether the practice should change.

    That closes Question 4. The load-bearing assumption stays as you worded it, and it remains untested.

    Question 5: untested constraints.

    Some things in this decision are being treated as fixed. I want to mark each one verified or assumed. Two have come up so far, besides the sign-off rule:

    • Timing: you want to advertise before the end of the month.
    • Cost: about $145,000 a year plus super for the hire.

    For each one, who set it, and has anyone tested it? For the timing, what happens if the ad goes out next month instead? Is something forcing the end-of-month date, such as a client commitment, a budget cycle or a promise to the two clients? For the $145,000, is it a quote from a recruiter or salary data, or your own estimate? And is there an approved budget behind it, or is that still to be decided?

    If there is another constraint you’ve been treating as fixed, such as a promise you’ve made to the two clients, add it. Tell me whether it’s verified or assumed.

15. Turn 15. Operator agent, answering from the fact sheet

    On timing: I set it myself. I want to decide by 30 October 2026 so the role can be advertised. I don’t know of anything outside that forcing the end-of-month date. I can’t point to a client commitment, a budget cycle or a promise to the two clients behind it, and I don’t know what would happen if the ad went out next month instead.

    On cost: the $145,000 is my own figure, not a recruiter quote or salary data. I’ve been treating it as the minimum needed to attract anyone, but I haven’t gathered any market data, so that’s assumed. I don’t know whether there’s an approved budget behind it.

    One other constraint I’ve been treating as fixed: I believe a senior engineer would take about three months to recruit. That’s assumed too. It’s based on our last hire, in 2023, and I haven’t checked it since.

16. Turn 16. Method agent, following the guide

    Thanks. That’s a clear answer on each one. Here is how I’m marking the constraints:

    • Principal sign-off before issue: VERIFIED. Your professional indemnity insurer requires it, and you have checked.
    • Only you can do that sign-off: ASSUMED. Nobody has asked the insurer.
    • Decide by 30 October 2026 so the role can be advertised: ASSUMED. You set the date yourself. You don’t know of anything outside forcing it, or what would happen if the ad went out next month.
    • About $145,000 a year plus super: ASSUMED. It’s your own figure, with no market data behind it. Whether there is an approved budget is not known.
    • About three months to recruit: ASSUMED. It’s based on your last hire in 2023 and hasn’t been checked since.

    So one of the five things being treated as fixed has been checked. The other four have not. I’m not saying any of them is wrong. They just haven’t been tested yet.

    Question 6: the evidence map.

    This is the last question. I’ve drafted the map from what you’ve told me, using four tags only. Please correct anything I’ve put in the wrong place.

    • GIVEN (you supplied it or checked it):
      • 41 of 63 projects delivered in the last 12 months were late, by an average of 9 working days. This is from your records.
      • Two repeat clients wrote to you in the September quarter about late packages.
      • Engineer billable utilisation averaged 78% last financial year.
      • The project system shows packages in “awaiting principal review” for an average of 6 working days.
      • You review every package yourself, and you also run business development.
      • Your insurer requires principal sign-off before issue.
      • Drafters regularly report waiting for markups. This is what they report, not a measured figure.
      • About 40% of the late projects had client-requested design changes after the fee was agreed, and most were absorbed without a variation.
      • Two engineers have said privately that they have spare time some weeks. This is what they told you, not a measured figure.
    • DERIVED (calculated from the given figures):
      • 41 of 63 is about 65% late.
      • About 40% of 41 late projects is roughly 16 projects with late scope changes.
    • INFERRED (reasoned from what is known):
      • The lateness may not be new. What may be new is that clients have started saying so. This rests on 12-month figures and a quarter-only trigger, with no quarter figures to compare.
      • On the way you work now, a new senior engineer would not shorten the 6-day wait at principal review, because every package would still come to you.
    • UNKNOWN (material but not established):
      • Why each of the 41 projects was late.
      • How many hours a week a new hire would free up.
      • Whether another registered engineer could sign off packages under your insurance.
      • The market salary for the role, and whether a budget is approved.
      • How long recruiting would take now.
      • What would happen if the decision or the ad moved past 30 October.
      • How many fewer late packages would count as success.

    One focused follow-up on this. Where does the “about 40%” figure come from? Did you count it from project records, or is it your estimate? If it’s an estimate, I’ll move it out of GIVEN and keep the “about” in the claim. And tell me if any other item above is tagged wrongly, or if something material is missing.

17. Turn 17. Operator agent, answering from the fact sheet

    On the “about 40%”: I can’t tell you it was counted from project records. It’s a figure I’ve been carrying as “about 40%”, and I don’t know that anyone counted it project by project, so treat it as an estimate and keep the “about”.

    The rest looks right to me. Two things I’d add because they’re missing:

    • Under GIVEN: the team is 6 engineers, 5 drafters, 1 principal reviewer and 6 support and admin staff, 18 people in all.
    • Under UNKNOWN: whether clients would pay more for faster turnaround.

18. Turn 18. Method agent, following the guide

    Thanks. I’ve moved the “about 40%” out of GIVEN, because nobody counted it project by project. I’ve also dropped the “roughly 16 projects” line from DERIVED, since a calculation can’t be firmer than the figure it comes from. I’ve added the team size and the question about clients paying more. That completes the six questions, so here is the framing handoff.

    1. Presented decision and trigger

    Whether to hire another senior structural engineer, at about $145,000 a year plus super, and advertise the role before the end of the month. The trigger is that two repeat clients wrote to you in the September quarter about late drawing packages.

    2. Decision underneath

    How do we cut the number of drawing packages that go out late? The hire is one way to do that. No target has been set for how many fewer late packages would count as success. Taking on more work is not driving this.

    3. Missing third option

    No third path was named. The option set is still: A, hire one senior structural engineer; or B, keep the current team and accept the late deliveries. Two shapes came up in the conversation: a smaller version of the hire, and holding the current team for a set period. You haven’t looked at either, and neither was adopted. That gap stays open.

    4. Load-bearing assumption

    In your words: “Hiring another senior structural engineer makes sense provided that the packages are late because we don’t have enough senior engineering time to get them done.” This has not been tested. Why the 41 late projects were late has not been analysed.

    5. Untested constraints

    • Every package must be signed off by a principal before it is issued: VERIFIED. Your professional indemnity insurer requires it, and you have checked.
    • Only you can do that sign-off: ASSUMED. Nobody has asked the insurer whether another registered engineer could.
    • Decide by 30 October 2026 so the role can be advertised: ASSUMED. You set the date yourself. You don’t know of anything outside forcing it, or what would happen if the ad went out next month.
    • About $145,000 a year plus super, treated as the minimum needed to attract anyone: ASSUMED. It’s your own figure, with no market data behind it.
    • About three months to recruit: ASSUMED. It’s based on your last hire in 2023 and hasn’t been checked since.

    6. Evidence map

    • GIVEN
      • 41 of 63 projects delivered in the last 12 months were late, by an average of 9 working days. This is from your records.
      • Two repeat clients wrote to you in the September quarter about late packages.
      • The team is 6 engineers, 5 drafters, 1 principal reviewer and 6 support and admin staff, 18 people in all.
      • Engineer billable utilisation averaged 78% last financial year.
      • The project system shows packages in “awaiting principal review” for an average of 6 working days.
      • You review every package yourself, and you also run business development.
      • Your insurer requires principal sign-off before issue.
      • Drafters regularly report waiting for markups. This is what they report, not a measured figure.
      • Two engineers have said privately that they have spare time some weeks. This is what they told you, not a measured figure.
    • DERIVED
      • 41 of 63 is about 65% late.
    • INFERRED
      • The lateness may not be new. What may be new is that clients have started saying so. This rests on 12-month figures and a quarter-only trigger, with no quarter figures to compare.
      • On the way you work now, a new senior engineer would not shorten the 6-day wait at principal review, because every package would still come to you.
    • UNKNOWN
      • Why each of the 41 projects was late.
      • The share of late projects with client-requested design changes after the fee was agreed, most absorbed without a variation. You carry it as about 40%, but it has not been counted.
      • How many hours a week a new hire would free up.
      • Whether another registered engineer could sign off packages under your insurance.
      • The market salary for the role, and whether a budget is approved.
      • How long recruiting would take now.
      • What would happen if the decision or the ad moved past 30 October.
      • How many fewer late packages would count as success.
      • Whether clients would pay more for faster turnaround.

    7. Reframed decision and the single evidence gap to verify next

    Reframed decision: how do we cut the number of drawing packages that go out late, when the cause of the lateness has not been established and every package currently passes through one principal review? Whether to hire a senior structural engineer is one answer to that question, not the question itself.

    Evidence gap to verify next: why each of the 41 late projects was late, and in particular whether a shortage of senior engineering time was what held them up. This is the test of the load-bearing assumption. Most of what you’ve told me so far points to other possible causes, and none of it has yet been tied to specific late projects.

    That’s where Frame Free stops. I haven’t ranked the options or recommended a course of action. Framing prepares the decision; it doesn’t make it.

Written before the run. The operator agent was told never to volunteer its last section.

1. Turn 1. Operator agent, answering from the fact sheet

   Use Frame Free in this PDF. Ask me one question at a time.

   I’m deciding whether to buy a $420,000 automated bun line for the bakery, because we expect a big lift in wholesale demand next year. The supplier’s price only holds until 15 November.

2. Turn 2. Method agent, following the guide

   Happy to. Let me confirm what I have before we start.

   The decision: whether to buy a $420,000 automated bun line for the bakery. The reason: you expect a big lift in wholesale demand next year. The timing: the supplier’s price holds until 15 November.

   You’ve given me two separate things that could be the trigger, and they aren’t the same kind of thing. One is a demand expectation and the other is a price deadline. “A big lift” is also a category, not a number yet.

   So here is the first question: What made this decision live now, the demand signal or the supplier’s 15 November date? And what does “a big lift in wholesale demand” mean in concrete terms, for example how many extra buns a week, from which customers, and starting when?

3. Turn 3. Operator agent, answering from the fact sheet

   Honestly, both. The supplier’s quote has an exchange-rate price hold that ends on 15 November, and that’s written into the quote, so the price goes up after that. But the reason I’m looking at the line at all is a meeting in August with a regional cafe chain about wholesale supply.

   On the numbers: the supplier’s case assumes about 10,500 extra buns a day. Their buyer told me they “could take up to 9,000 a day”, and the other 1,500 a day is general growth. They mentioned wanting supply from early 2027, but that was said in the meeting and there’s nothing in writing. For context, we make about 7,500 buns a day now, over 300 production days a year.

4. Turn 4. Method agent, following the guide

   Thanks, that separates the two cleanly. The August meeting with the regional cafe chain is the reason you are looking at the line at all. The 15 November exchange-rate price hold, which is written into the quote, is what sets the timing.

   I’ll note a few things for later and not weigh them yet:

   • The supplier’s case rests on about 10,500 extra buns a day. That is more than your current 7,500 a day, so it works out to roughly 18,000 a day in total, about 2.4 times today’s output.
   • Up to 9,000 of those 10,500 come from one buyer, who said they “could take up to 9,000 a day”. The early 2027 start was said in the meeting, and nothing is in writing.
   • The other 1,500 a day is described as general growth, with no source given yet.

   Here is the second question. Suppose the bun line were off the table tomorrow: the supplier withdrew the quote and no other machine was available. What larger question about the bakery would still be sitting there, and what outcome are you trying to change?

5. Turn 5. Operator agent, answering from the fact sheet

   I suppose the question still sitting there would be the cafe chain. Can we supply them if they come on, and how? The current semi-manual line does about 7,500 buns a day, and they talked about up to 9,000 a day on their own. What I’m trying to change is being able to take on that wholesale growth. Beyond that, I haven’t really thought about it as a bigger question than the line, so I don’t know how to put it more broadly.

6. Turn 6. Method agent, following the guide

   That helps. Without the line, the cafe chain is the question that’s left. That tells us the line is a means, not the thing being decided.

   “Being able to take on that wholesale growth” is still a means, though. Here is what you’ve told me so far. Up to 9,000 a day from one chain would on its own be more than your current 7,500 a day. So taking them on would more than double the bakery’s output, and most of that new volume would depend on a single customer whose start date isn’t in writing.

   One follow-up, and then we’ll move on. If you did take on the chain, what would change for the bakery that matters to you, for example profit, owner income, keeping the team busy, or the long-term direction of the business? And would it be accurate to say the larger question is “should we become a bakery that depends substantially on one wholesale customer, and on what terms?” If that’s wrong, correct it in your own words.

7. Turn 7. Operator agent, answering from the fact sheet

   The only figure I’ve had in front of me is the supplier’s: about $164,000 a year, made up of the extra buns at $0.04 contribution each plus $38,000 a year in labour savings. So profit, I suppose. Beyond that I honestly haven’t worked out what it would mean for owner income or the direction of the business. On the team, the two staff on the current line would be redeployed, not made redundant. I’ve agreed that with both of them.

   On your wording: I hadn’t thought of it as “becoming a bakery that depends on one wholesale customer”. I’ve been thinking about it as whether to buy the line. I can’t tell you the terms, because I don’t know whether the chain will sign a supply agreement or on what terms. So I can’t say your version is wrong, but it isn’t how I’d been putting it.

8. Turn 8. Method agent, following the guide

   That’s a fair correction, and I’ll keep your wording rather than mine. You have been framing this as “whether to buy the line”. With the line removed, the question left is “can we supply the cafe chain if they come on, and how?” The outcome you named is profit. The one profit figure you have seen is the supplier’s.

   I’ll note two things for the evidence map and not weigh them yet:

   • The supplier’s $164,000 a year does tie to their own inputs. 10,500 extra buns a day over 300 days at $0.04 each comes to $126,000, and adding $38,000 in labour savings gives $164,000. So all of it rests on the full 10,500 a day, and up to 9,000 of those depend on the chain.
   • Whether the chain will sign a supply agreement, and on what terms, is unknown. You’ve also agreed redeployment with both staff on the current line, so that one is settled.

   Here is the third question. So far the choice on the table is buy the line by 15 November or don’t. What is a genuine third path you haven’t been treating as real? It could be keeping things as they are for a set period, a smaller version of the same move, or a different way of supplying the chain if they come on. It only needs to be credible enough to take seriously later. We won’t judge it now.

9. Turn 9. Operator agent, answering from the fact sheet

   I don’t have one. The only two options I’ve had in front of me are buying the line at $420,000 installed, financed over five years, or carrying on with the existing semi-manual line. I haven’t looked at a smaller version, a set waiting period, or another way of supplying the chain, so I don’t know whether any of those would be credible.

10. Turn 10. Method agent, following the guide

    That’s an honest answer, and “I don’t have one” is useful to record. It also corrects how I put the choice. Your two options are buying the line at $420,000 installed, financed over five years, or carrying on with the existing semi-manual line. So keeping things as they are is already one of the two. The third path has to be something other than those.

    I won’t make one up for you. The guide gives two forms a third path often takes, and I’ll ask about each using only what you’ve told me. This is my one follow-up on this question.

    First, a timing version. You have nothing in writing from the chain, and the price hold ends on 15 November. Is there any version of “get the chain’s intent in writing first, then decide on the line” that you could take seriously, even if it meant missing the price hold? Second, a smaller version. Do you know whether the supplier, or anyone else, offers a smaller or staged line sized nearer the chain’s volume than the full 10,500 extra a day? If you don’t know, say so and I’ll record both as unknown. We won’t judge either of them now.

11. Turn 11. Operator agent, answering from the fact sheet

    On the timing version: I haven’t considered it, so I don’t know whether I’d take it seriously. What I do know is that the price goes up after 15 November, because that’s written into the quote. I don’t know whether I could get anything in writing from the chain before then, because nothing has been put in writing so far and I don’t know whether they’ll sign a supply agreement at all.

    On the smaller version: I don’t know. The only line I’ve been quoted is the one that can make up to 24,000 a day, and I haven’t asked the supplier or anyone else about a smaller or staged option.

12. Turn 12. Method agent, following the guide

    Thanks. I’ll record both as you gave them and not judge either one.

    • The timing version, getting the chain’s intent in writing before deciding, has not been considered. Whether the chain would put anything in writing before 15 November is unknown.
    • The smaller or staged version has not been asked about. The only line quoted is one that can make up to 24,000 buns a day. For comparison, your current 7,500 a day plus the supplier’s 10,500 extra comes to about 18,000 a day.

    So the missing third path stays open for now. Neither version has been tested, and that gap will go into the handoff.

    Here is the fourth question. Nearly everything so far traces back to one point. The supplier’s 10,500 extra buns a day, and so the $164,000 a year, depends mostly on the cafe chain taking up to 9,000 a day from early 2027. Nothing about that is in writing.

    Finish this sentence in your own words: “Buying the line makes sense provided that …” If one belief turned out to be wrong and that would change your answer, what is it? And what have you actually seen, measured or had confirmed in support of it, apart from what the buyer said in the August meeting?

13. Turn 13. Operator agent, answering from the fact sheet

    “Buying the line makes sense provided that the cafe chain actually comes on and takes the volume they talked about.” That’s the belief. If they don’t sign, I’m not sure the case holds, though I can’t tell you by how much.

    What I’ve seen in support of it, apart from what the buyer said in August: nothing. There’s no letter of intent and nothing in writing. The buyer said they “could take up to 9,000 a day”, and that’s all I have.

14. Turn 14. Method agent, following the guide

    Thanks. That’s clear, and I’ll use your words as the load-bearing assumption: “Buying the line makes sense provided that the cafe chain actually comes on and takes the volume they talked about.” What has been seen in support of it, apart from the buyer’s words in August, is nothing. There’s no letter of intent and nothing in writing. You’ve also said you can’t tell how far the case would move if they don’t sign. I’ll record that as unknown and not work it out now.

    Here is the fifth question. Several things in this decision are being treated as fixed. I’ll list them as you’ve given them, and I need you to tell me which ones have actually been checked:

    • The 15 November date. The price hold is written into the quote, so the hold itself is on paper. Has anyone asked the supplier whether it can be extended, or how much the price rises after that date?
    • The $420,000 installed price, financed over five years. Is that finance arranged or approved, or is it the structure in the supplier’s proposal?
    • The chain’s early 2027 start, which you’ve said was said in the meeting and is not in writing.
    • The redeployment of the two staff on the current line, which you’ve said you’ve agreed with both of them.

    For each one, who set it, when was it last tested, and would you call it verified or assumed? If you haven’t asked, “nobody has asked” is a fine answer, and I’ll mark it that way.

15. Turn 15. Operator agent, answering from the fact sheet

    Taking them in order:

    • The 15 November date: the supplier set it, and the price hold ending then is written into the quote, so I’d call the hold itself verified. Nobody has asked the supplier whether it can be extended, and I don’t know how much the price rises after that date.
    • The $420,000 installed, financed over five years: that’s the option as I’ve had it in front of me. I don’t know whether the finance is arranged or approved, so I can’t call it verified.
    • The chain’s early 2027 start: the chain’s buyer said it in the August meeting. Nothing is in writing and nobody has tested it since, so it’s assumed.
    • The redeployment of the two staff: I set that, and I’ve agreed it with both of them, so I’d call it verified. I can’t give you a date for when it was last confirmed.

16. Turn 16. Method agent, following the guide

    Thanks. I’ll mark the constraints as you’ve given them:

    • The 15 November price hold: VERIFIED, because it is written into the quote. Whether it can be extended, and how much the price rises afterwards, has not been asked.
    • The $420,000 installed, financed over five years: ASSUMED. You don’t know whether the finance is arranged or approved.
    • The chain’s early 2027 start: ASSUMED. It was said in the August meeting, nothing is in writing, and nobody has tested it since.
    • The redeployment of the two staff: VERIFIED, agreed with both of them. When it was last confirmed isn’t known.

    Here is the sixth and last question. It asks what you actually know. I’ve drafted an evidence map from what you’ve told me, using only four tags. Please correct anything I’ve put in the wrong place.

    • GIVEN: the quote is $420,000 installed, and the price hold ends on 15 November. You make about 7,500 buns a day over 300 production days a year. The quoted line can make up to 24,000 a day. The chain’s buyer said they “could take up to 9,000 a day”. Both staff on the current line have agreed to be redeployed.
    • GIVEN as the supplier’s figures, with their accuracy not checked by you: 10,500 extra buns a day, $0.04 contribution per bun, and $38,000 a year in labour savings.
    • DERIVED: today’s 7,500 plus the 10,500 extra comes to about 18,000 a day, roughly 2.4 times today’s output. The supplier’s inputs give $126,000 from extra buns plus $38,000 in savings, which is $164,000 a year. On those same inputs, the chain’s 9,000 a day is about $108,000 of that $164,000.
    • INFERRED: the line is a means of supplying the chain rather than the decision itself, because without the line the chain is the question still left.
    • UNKNOWN: whether the chain signs a supply agreement, on what terms, and when it would start. The source of the other 1,500 a day of general growth. Whether the finance is arranged. Whether the price hold can be extended, and how much the price rises after 15 November. Whether a smaller or staged line exists. How far the case moves if the chain doesn’t sign. What any of this means for owner income or the direction of the business.

    Two things to check. First, is anything here in the wrong category? In particular, have you tested the $0.04 contribution or the $38,000 labour saving against your own costs, or are both only the supplier’s numbers? Second, is there a material claim missing, something you’re relying on that isn’t in this list?

17. Turn 17. Operator agent, answering from the fact sheet

    On the categories: the $0.04 contribution and the $38,000 labour saving are both from the supplier’s case. I can’t tell you I’ve tested either against our own costs, so treat them as the supplier’s numbers only. The same goes for the 18,000 a day: the supplier filled that figure in during the sales meeting.

    On what’s missing, there are two things I’m relying on that aren’t in your list:

    • I’ve been assuming the building can take the line without electrical upgrades. That hasn’t been checked. The electrician hasn’t been asked.
    • The supplier’s payback calculator shows 2.6 years, based on 18,000 units a day. That’s the payback figure I’ve had in front of me.

18. Turn 18. Method agent, following the guide

    Thanks. Both corrections go in. The $0.04 contribution, the $38,000 labour saving and the 18,000 a day are all the supplier’s numbers, and you haven’t tested any of them against your own costs. The electrical capacity and the 2.6-year payback are both new to the map. One note on the payback: $420,000 divided by the supplier’s $164,000 a year comes to about 2.6 years. So it rests on the same inputs as the $164,000, including the full 10,500 extra buns a day.

    That completes the six questions. Here is the framing handoff.

    1. Presented decision and trigger

    Whether to buy a $420,000 automated bun line, installed and financed over five years, or carry on with the existing semi-manual line. The reason it is being looked at is an August meeting with a regional cafe chain about wholesale supply. The timing comes from the exchange-rate price hold in the supplier’s quote, which ends on 15 November.

    2. Decision underneath

    In your words: “Can we supply them if they come on, and how?” The line is a means of supplying the cafe chain, not the decision itself. Taking on the chain’s up to 9,000 a day would more than double the bakery’s current output of about 7,500 a day, and most of the new volume would come from one customer. You named profit as the outcome you are trying to change. You haven’t yet worked out what it would mean for owner income or the direction of the business.

    3. Missing third option

    None identified. You said “I don’t have one”, and that is recorded as it stands. Two forms were raised and neither has been tested. One is a timing version: get the chain’s intent in writing first, then decide on the line, even if that means missing the price hold. You haven’t considered it. The other is a smaller or staged line sized nearer the chain’s volume. Nobody has asked the supplier or anyone else whether one exists. The only line quoted can make up to 24,000 a day. Both stay open, and neither has been judged.

    4. Load-bearing assumption

    In your words: “Buying the line makes sense provided that the cafe chain actually comes on and takes the volume they talked about.” What supports it, apart from the buyer saying in August that they “could take up to 9,000 a day”, is nothing. There is no letter of intent and nothing in writing. How far the case moves if they don’t sign is unknown.

    5. Untested constraints

    • The 15 November price hold: VERIFIED. It is written into the quote. Nobody has asked whether it can be extended, and how much the price rises afterwards is not known.
    • The $420,000 installed, financed over five years: ASSUMED. Whether the finance is arranged or approved is not known.
    • The chain’s early 2027 start: ASSUMED. The buyer said it in the August meeting, nothing is in writing, and nobody has tested it since.
    • The building can take the line without electrical upgrades: ASSUMED. The electrician hasn’t been asked.
    • The redeployment of the two staff on the current line: VERIFIED. You set it and agreed it with both of them. When it was last confirmed is not known.

    6. Evidence map

    • GIVEN: the quote is $420,000 installed, financed over five years, with a price hold ending 15 November. The bakery makes about 7,500 buns a day over 300 production days a year. The quoted line can make up to 24,000 a day. The chain’s buyer said they “could take up to 9,000 a day” and mentioned early 2027, in the August meeting only. Both staff on the current line have agreed to be redeployed.
    • GIVEN as the supplier’s figures, none tested against your own costs: 10,500 extra buns a day, of which 1,500 is described as general growth. $0.04 contribution per bun. $38,000 a year in labour savings. 18,000 a day in total, filled in by the supplier during the sales meeting. A 2.6-year payback, based on 18,000 units a day.
    • DERIVED: 7,500 plus 10,500 is about 18,000 a day, roughly 2.4 times today’s output. 10,500 a day over 300 days at $0.04 is $126,000, plus $38,000 in savings gives $164,000 a year. On the same inputs, the chain’s 9,000 a day accounts for about $108,000 of that. $420,000 divided by $164,000 is about 2.6 years, so the supplier’s payback ties to their own inputs.
    • INFERRED: the line is a means of supplying the chain rather than the decision itself, because without the line the chain is the question still left. Because the $164,000 and the 2.6-year payback both assume the full 10,500 extra a day, both depend mostly on the chain.
    • UNKNOWN: whether the chain signs a supply agreement, on what terms, and when it would start. The source of the 1,500 a day of general growth. Whether the $0.04 and the $38,000 hold against your own costs. Whether the finance is arranged. Whether the price hold can be extended, and how much the price rises after 15 November. Whether a smaller or staged line exists. Whether the building needs electrical upgrades. How far the case moves if the chain doesn’t sign. What any of this means for owner income or the direction of the business.

    7. Reframed decision and the single evidence gap to verify next

    Reframed decision: whether, and on what terms, to build capacity to supply the cafe chain’s possible volume of up to 9,000 a day, given that nothing from the chain is in writing, and whether the supplier’s 15 November price hold should set the timing of that decision.

    The single evidence gap to verify next: whether the cafe chain will commit in writing to volume, terms and a start date. That is the load-bearing assumption, and the supplier’s $164,000 a year and 2.6-year payback both depend mostly on it. What you find out may also change how much the 15 November date matters.

    That is where the framing stops. No options have been compared or scored, and nothing has been recommended. Framing prepares the decision. It does not make it.

Written before the run. The operator agent was told never to volunteer its last section.

1. Turn 1. Operator agent, answering from the fact sheet

   Use Frame Free in this PDF. Ask me one question at a time.

   Our client portal programme is over budget and behind, and I have to tell the board on 22 October whether we finish it or pause it. We’ve put $310,000 into it so far.

2. Turn 2. Method agent, following the guide

   Thanks. Here is what I have so far:

   • Decision: whether to finish or pause the client portal programme.
   • Trigger: you report to the board on 22 October, and the programme is over budget and behind schedule.
   • Spent to date: $310,000.

   We’ll go through six questions, one at a time. This is framing only. I won’t score the options or recommend one.

   Question 1: The decision and trigger

   Can you state the decision in one sentence, in specific terms? Two things would make it concrete:

   1. What do “finish” and “pause” mean in practice? For example, how much more money and time would finishing take, and would a pause stop all spending or keep a reduced team going?
   2. What made this live now rather than last month? Is it the board date itself, or did something happen recently, such as a revised estimate, a missed milestone or a budget threshold being crossed?

3. Turn 3. Operator agent, answering from the fact sheet

   In one sentence: I need to recommend to the board on 22 October whether we put an estimated $190,000 more into the client portal to finish it by mid-2027, or pause it.

   1. Finishing means funding that extra $190,000, which takes the total to $500,000 against the $360,000 the board approved, with completion by mid-2027. Pausing means stopping development and holding the work done so far. I haven’t worked out whether a pause would keep a reduced team going; I don’t know that yet.
   2. It was the contractor’s revised estimate in September 2026. That pushed the total past the approved budget, so I have to go back to the board.

4. Turn 4. Method agent, following the guide

   That is a clear statement. Here is Question 1 as I will carry it forward:

   • Presented decision: recommend to the board on 22 October whether to put an estimated $190,000 more into the client portal to finish it by mid-2027, or pause it.
   • Trigger: the contractor’s revised estimate in September 2026, which took the projected total to $500,000 against the $360,000 the board approved.
   • Still open: what a pause would actually involve, including whether a reduced team stays on. You said you don’t know that yet, so I’ll record it as open rather than fill it in.

   Question 2: The decision underneath

   Suppose the finish-or-pause choice disappeared tomorrow. What larger question would still be sitting there?

   To make that concrete: what was the client portal meant to change for the business, and for whom? I mean the outcome it was approved to deliver, not the features. Is that outcome still the one that matters to you and the board today?

5. Turn 5. Operator agent, answering from the fact sheet

   The approved business case was to cut inbound client calls. The portal was meant to let our clients get their certificates of currency and lodge claims online, so they wouldn’t need to ring us for those things. When the board approved it, inbound calls averaged 2,400 a month, and certificate requests were about 55% of them.

   As far as I’m concerned, cutting those calls is still the outcome that matters. I don’t know whether the board would see it differently today; I haven’t asked them.

6. Turn 6. Method agent, following the guide

   Thanks. Here is Question 2 as I will carry it forward:

   • Outcome the portal was approved to deliver: fewer inbound client calls, by letting clients get certificates of currency and lodge claims online.
   • Baseline at approval: inbound calls averaged 2,400 a month, and certificate requests were about 55% of them.
   • Decision underneath: what is the best way to cut inbound client calls, and in particular certificate requests? The portal is one means to that outcome. Finishing or pausing it is a choice about that means, not about the outcome.
   • Still open: whether the board still sees cutting calls as the outcome that matters. You said you haven’t asked them, so I’ll record that as untested rather than assume it.

   Question 3: The missing option

   So far the choice has been framed as two paths: put in $190,000 more to finish, or pause. Which genuine third path have you not been treating as real?

   It has to be credible enough to take seriously later, and it can include keeping the current position. The guide’s own examples are “do nothing for six months” and “A, but half the size”. Since the outcome is fewer calls rather than a finished portal, a third path might also be one that goes at the calls by a route other than the full build. I won’t evaluate whatever you name. I only want it on the table.

7. Turn 7. Operator agent, answering from the fact sheet

   Honestly, I haven’t been treating any third path as real. I’ve only looked at it as two options: fund the extra $190,000 and finish, or pause. I don’t have a third option worked out, so I can’t name one for you with any confidence.

8. Turn 8. Method agent, following the guide

   That’s a useful answer in itself. It confirms the choice reached you as two options and nobody has built a third yet. I won’t make one up for you. It has to come from you to be worth carrying forward.

   I’ll ask one follow-up on this question, then move on whatever the answer is. Take the shapes the guide uses and test them against your own programme:

   1. A smaller version of finishing. The portal has two jobs: certificates of currency and claims lodgement. Certificates were about 55% of inbound calls at approval. Is there a part of the build you could fund and finish on its own, for less than $190,000?
   2. Holding the current position for a set period. Is there a version of “pause” with an end date and a condition for restarting, rather than an open-ended stop?
   3. Reaching the outcome another way. Is there any route to fewer certificate calls that doesn’t depend on the portal being finished?

   If one of these is real for you, name it in your own words. If none of them is, say so and I’ll record that no third path was identified. That is a legitimate result too. I won’t judge whichever you pick, here or later in this framing.

9. Turn 9. Operator agent, answering from the fact sheet

   Taking them in turn:

   1. On a smaller version: the certificate module is about 80% built, and the claims lodgement module hasn’t started. I don’t know how the $190,000 splits between the two, so I can’t tell you what finishing certificates alone would cost.
   2. On a time-limited hold: I haven’t considered a pause with an end date or a restart condition. I don’t know what that would look like.
   3. On another route: in March 2026 our main underwriter launched its own online service where clients can download certificates of currency directly. I hadn’t thought of that as an option for us, but it is a route to certificates that doesn’t depend on our portal.

   If I had to name a third path in my own words, it would be finishing only the certificate module and not building claims. I can’t say yet whether that’s credible, because I don’t know the cost.

10. Turn 10. Method agent, following the guide

    Thanks. That was the one follow-up for this question, so here is Question 3 as I will carry it forward:

    • Missing third option, in your words: finish only the certificate module and do not build claims lodgement.
    • What you know about it: the certificate module is about 80% built, and the claims module has not started.
    • What you don’t know: how the $190,000 splits between the two modules, so the cost of finishing certificates alone is open. You said you can’t yet say whether this path is credible, and I’ll record it that way.
    • Also on the table: since March 2026, your main underwriter has run its own online service where clients can download certificates of currency directly. You hadn’t treated that as an option. I’ll record it as a fact about the situation, not as a path you have chosen.
    • Not considered yet: a pause with an end date or a restart condition. I’ll record it as not explored.

    I’m not weighing any of these against each other.

    Question 4: The load-bearing assumption

    Which single belief would change the decision if it proved wrong? I’m asking for the one that carries the decision, not a list of risks.

    One way to find it is to finish this sentence in your own words: “Putting more money into the portal makes sense provided that …”

    Then tell me what has actually been seen, measured or checked in support of that belief, and when. For example, the business case rests on the 2,400 calls a month and the 55% share for certificates at approval. Has anyone measured either figure since then? If not, say so. A belief nobody has checked is still worth naming.

11. Turn 11. Operator agent, answering from the fact sheet

    In my own words: “Putting more money into the portal makes sense provided that the work we’ve already done would be wasted if we paused.” That’s what I’ve been working from. I’ve told colleagues we’ve spent too much to stop now.

    What’s been checked behind it: nothing. Nobody has assessed what of the work done so far could be reused if we paused.

    On the business case figures: inbound calls have averaged about 1,500 a month since April 2026, against 2,400 at approval. I don’t know whether anyone has re-measured the share that are certificate requests since approval; I haven’t seen a figure for it. I also don’t know how many clients would actually use the portal.

12. Turn 12. Method agent, following the guide

    Thanks. Here is what I have for Question 4 so far:

    • Belief you have been working from, in your words: “Putting more money into the portal makes sense provided that the work we’ve already done would be wasted if we paused.”
    • What has been checked behind it: nothing. Nobody has assessed what of the work done so far could be reused if you paused.
    • Business case figures: inbound calls have averaged about 1,500 a month since April 2026, against 2,400 at approval. Nobody you know of has re-measured the certificate share since approval. How many clients would use the portal is not known.

    I want to push on this once before moving on, because the belief as stated is about money already spent. The $310,000 is gone whichever way the board decides. What the board is actually being asked to approve is the next $190,000. So the belief that carries the decision may be a different one.

    Follow-up: try finishing the sentence about the new money rather than the old: “Putting a further $190,000 into the portal makes sense provided that …”

    As you answer, two facts you have given me may be relevant. I’m putting them side by side, not drawing a conclusion from them:

    1. Calls fell from about 2,400 to about 1,500 a month, and the lower level dates from April 2026.
    2. Your main underwriter launched its own certificate download service in March 2026.

    Nothing you’ve told me establishes whether the second has anything to do with the first. If you know why calls fell, say so. If you don’t, say that, and I’ll record the cause of the drop as unknown.

    If you would still finish the sentence the way you did first time, that’s fine too. I’ll record your answer as it stands and we’ll move to Question 5.

13. Turn 13. Operator agent, answering from the fact sheet

    I hadn’t framed it around the new money before. If I try: “Putting a further $190,000 into the portal makes sense provided that it cuts inbound client calls the way the business case said it would.” But I don’t know how many clients would actually use the portal, and I don’t know whether the $190,000 estimate will hold.

    On why calls fell: I don’t know. Nobody has looked into the cause, as far as I know, so please record it as unknown.

14. Turn 14. Method agent, following the guide

    Thanks. That was the one follow-up for this question, so here is Question 4 as I will carry it forward:

    • Load-bearing assumption, in your words: “Putting a further $190,000 into the portal makes sense provided that it cuts inbound client calls the way the business case said it would.”
    • What has been checked behind it: you don’t know how many clients would actually use the portal, and you don’t know whether the $190,000 estimate will hold. Nobody you know of has re-measured the certificate share of calls since approval.
    • Related facts, side by side: inbound calls have averaged about 1,500 a month since April 2026, against 2,400 at approval. Your main underwriter launched its own certificate download service in March 2026. The cause of the drop in calls is unknown, and I’m recording it that way.
    • Earlier belief, also recorded: you had been working from “the work we’ve already done would be wasted if we paused”. Nobody has assessed what of that work could be reused. I’m keeping it on the record as a belief that hasn’t been checked, not discarding it.

    Question 5: Untested constraints

    Several things have been treated as fixed so far. I’ll list them as you’ve given them. For each one, tell me whether it is verified or assumed: who set it, when it was last tested, and how you would know if it moved.

    1. The board date of 22 October. Is that the only point at which the board can decide this, or could the decision go to a later meeting or be made out of session?
    2. The $190,000 estimate and the mid-2027 completion date. Both come from the contractor’s revised estimate in September 2026. Has anyone checked either one independently, and is the contractor committed to them on any terms?
    3. The $360,000 the board approved. Is that a hard ceiling, or is there a delegation or tolerance that applies to overruns?
    4. What “pause” means. You said it means stopping development and holding the work done so far. Is there anything in the contractor’s agreement about suspending work, such as notice, holding costs or loss of the team, or has nobody looked?

    If you don’t know the answer to one of these, say so and I’ll mark it assumed. That is a useful result here.

15. Turn 15. Operator agent, answering from the fact sheet

    Taking them in turn:

    1. The board date: the recommendation is due to the board on 22 October 2026. I don’t know whether it could go to a later meeting or be decided out of session; I haven’t asked.
    2. The $190,000 and mid-2027: both come from the contractor’s revised estimate in September 2026. I don’t know of anyone checking either one independently, and I don’t know whether the contractor is committed to them on any terms. I don’t know whether the $190,000 will hold. Mark both as assumed.
    3. The $360,000: that’s what the board approved in March 2025. I don’t know of any delegation or tolerance for overruns. On the ceiling, I’ve been working on the basis that the board won’t approve more than $500,000 in total, but that’s my own reading of the chair. Nobody has tested it, so mark it assumed.
    4. What pause means: I’ve been assuming a pause means losing the contractor’s team. That hasn’t been checked; the contractor hasn’t been asked about a pause. I haven’t looked at what the agreement says about suspending work, notice or holding costs, so I don’t know.

16. Turn 16. Method agent, following the guide

    Thanks. Here is Question 5 as I will carry it forward. Where you said you don’t know, or nobody has asked, I’ve marked it assumed.

    1. Board date of 22 October 2026. The date the recommendation is due is as you stated it. Whether 22 October is the only point at which the board can decide, or whether it could go to a later meeting or out of session, has not been asked. ASSUMED fixed.
    2. The $190,000 and mid-2027 completion. Both come from the contractor’s revised estimate in September 2026. Nobody you know of has checked either independently, and you don’t know whether the contractor is committed to them on any terms. ASSUMED, both of them.
    3. The $360,000 approved budget. The board approved $360,000 in March 2025. You know of no delegation or tolerance for overruns, and nobody has checked for one. The approval itself is your record of the board decision. Treating it as a hard ceiling with no tolerance is ASSUMED.
    4. A $500,000 total limit. This is your own reading of the chair, and nobody has tested it. ASSUMED.
    5. Pause means losing the contractor’s team. The contractor hasn’t been asked about a pause, and nobody has read what the agreement says about suspending work, notice or holding costs. ASSUMED.

    None of the constraints you’ve treated as fixed has been verified. I’m recording that as it stands, not as a judgement on the decision.

    Question 6: The evidence map

    This is the last question. Below is a draft map of the material claims from what you’ve told me, each tagged with one of four labels: GIVEN (you supplied it or it’s on record), DERIVED (calculated from given figures), INFERRED (reasoned from what is known) or UNKNOWN (material but not established). Approximate figures stay approximate inside the claim.

    GIVEN

    • $310,000 spent to date.
    • The board approved $360,000 in March 2025.
    • The contractor’s revised estimate in September 2026 is a further $190,000, with completion by mid-2027.
    • The business case was to cut inbound client calls by letting clients get certificates of currency and lodge claims online.
    • At approval, inbound calls averaged 2,400 a month, and certificate requests were about 55% of them.
    • Inbound calls have averaged about 1,500 a month since April 2026.
    • The certificate module is about 80% built. The claims lodgement module has not started.
    • Your main underwriter launched its own online certificate download service in March 2026.
    • The recommendation is due to the board on 22 October 2026.

    DERIVED

    • Finishing would take the total to $500,000, which is $140,000 (about 39%) over the $360,000 approved.
    • $50,000 of the approved $360,000 has not yet been spent.
    • At approval, certificate requests were roughly 1,320 calls a month (55% of 2,400).
    • Inbound calls are down about 900 a month, or roughly 37%, against the approval baseline.

    INFERRED

    • If the certificate share were still about 55%, certificate calls would now run at roughly 825 a month. That share has not been re-measured, so this holds only on that condition.

    UNKNOWN

    • Why inbound calls fell, including whether the underwriter’s service had anything to do with it.
    • The current share of calls that are certificate requests.
    • How many clients would use the portal.
    • Whether the $190,000 and mid-2027 will hold.
    • How the $190,000 splits between the certificate and claims modules.
    • What of the work done so far could be reused if you paused.
    • What the contractor’s agreement says about suspending work, and what a pause would cost or involve.
    • Whether the board could decide later than 22 October, whether any overrun tolerance exists, and whether the board still sees cutting calls as the outcome that matters.

    Please check it. Is any claim tagged wrongly, for example something I’ve marked GIVEN that you have only heard second-hand, or something marked UNKNOWN that you can in fact confirm? Is there any material claim missing? Once you’ve corrected it, I’ll put the full framing together and stop there.

17. Turn 17. Operator agent, answering from the fact sheet

    I’ve checked it. As far as I know the tags are right. I can’t confirm anything you’ve marked UNKNOWN; those really are open for me.

    Two facts are missing that I’d add under GIVEN:

    • Work started in April 2025.
    • The original go-live date was June 2026.

    Otherwise it matches what I’ve told you.

18. Turn 18. Method agent, following the guide

    Thanks. I’ve added both facts under GIVEN and worked out what follows from them under DERIVED. Here is the full framing.

    1. Presented decision and trigger

    • Decision: recommend to the board on 22 October 2026 whether to put an estimated $190,000 more into the client portal to finish it by mid-2027, or pause it.
    • Trigger: the contractor’s revised estimate in September 2026, which took the projected total to $500,000 against the $360,000 the board approved.
    • Still open: what a pause would actually involve, including whether a reduced team stays on.

    2. Decision underneath

    What is the best way to cut inbound client calls, and certificate requests in particular? The portal is one means to that outcome. Finishing or pausing it is a choice about that means, not about the outcome. Whether the board still sees cutting calls as the outcome that matters has not been asked.

    3. Missing third option

    In your words: finish only the certificate module and do not build claims lodgement. The certificate module is about 80% built and the claims module has not started. How the $190,000 splits between the two is unknown, so you can’t yet say whether this path is credible.

    Two related items, recorded but not chosen:

    • Since March 2026 your main underwriter has run its own online service where clients can download certificates of currency directly. You hadn’t treated it as an option. It is recorded as a fact about the situation.
    • A pause with an end date or a restart condition has not been explored.

    4. Load-bearing assumption

    In your words: “Putting a further $190,000 into the portal makes sense provided that it cuts inbound client calls the way the business case said it would.”

    Nothing has been checked behind it. Nobody knows how many clients would use the portal, whether the $190,000 will hold, or what share of calls are certificate requests today.

    Your earlier belief is also on the record: “the work we’ve already done would be wasted if we paused.” Nobody has assessed what of that work could be reused. The $310,000 is spent whichever way the board decides. The board is being asked to approve the next $190,000.

    5. Untested constraints

    Constraint	Status	Basis
    22 October 2026 is the only point at which the board can decide	ASSUMED	The due date is as you stated it. Whether a later meeting or an out-of-session decision is possible has not been asked.
    The further $190,000	ASSUMED	Contractor’s September 2026 estimate. Not checked independently. Contractor’s commitment to it on any terms is unknown.
    Completion by mid-2027	ASSUMED	Same source and same position as the $190,000.
    $360,000 is a hard ceiling with no tolerance for overruns	ASSUMED	The March 2025 approval is your record of the board decision. Whether any delegation or tolerance applies has not been checked.
    The board won’t approve more than $500,000 in total	ASSUMED	Your own reading of the chair. Not tested.
    A pause means losing the contractor’s team	ASSUMED	The contractor hasn’t been asked. The agreement hasn’t been read on suspension, notice or holding costs.

    None of the constraints treated as fixed has been verified.

    6. Evidence map

    GIVEN

    • $310,000 spent to date.
    • The board approved $360,000 in March 2025.
    • Work started in April 2025.
    • The original go-live date was June 2026.
    • The contractor’s revised estimate in September 2026 is a further $190,000, with completion by mid-2027.
    • The business case was to cut inbound client calls by letting clients get certificates of currency and lodge claims online.
    • At approval, inbound calls averaged 2,400 a month, and certificate requests were about 55% of them.
    • Inbound calls have averaged about 1,500 a month since April 2026.
    • The certificate module is about 80% built. The claims lodgement module has not started.
    • Your main underwriter launched its own online certificate download service in March 2026.
    • The recommendation is due to the board on 22 October 2026.

    DERIVED

    • Finishing would take the total to $500,000, which is $140,000 (about 39%) over the $360,000 approved.
    • $50,000 of the approved $360,000 has not yet been spent.
    • The original plan ran about 14 months, from April 2025 to June 2026. Completion by mid-2027 is roughly a year after the original go-live date, which has already passed.
    • At approval, certificate requests were roughly 1,320 calls a month (55% of 2,400).
    • Inbound calls are down about 900 a month, or roughly 37%, against the approval baseline.

    INFERRED

    • If the certificate share were still about 55%, certificate calls would now run at roughly 825 a month. That share has not been re-measured, so this holds only on that condition.

    UNKNOWN

    • Why inbound calls fell, including whether the underwriter’s service had anything to do with it.
    • The current share of calls that are certificate requests.
    • How many clients would use the portal.
    • Whether the $190,000 and mid-2027 will hold.
    • How the $190,000 splits between the certificate and claims modules.
    • What of the work done so far could be reused if you paused.
    • What the contractor’s agreement says about suspending work, and what a pause would cost or involve.
    • Whether the board could decide later than 22 October, whether any overrun tolerance exists, and whether the board still sees cutting calls as the outcome that matters.

    7. Reframed decision and the single evidence gap to verify next

    Reframed decision: how should the business reduce the certificate and claims calls that remain, now that inbound calls have already fallen by about a third for reasons nobody has established? Finishing the portal, finishing the certificate module alone, and pausing are all means to that outcome. The board can only weigh them once it knows what call volume each one is meant to remove.

    Evidence gap to verify next: the current share of inbound calls that are certificate requests, and whether it has changed since March 2026. Your load-bearing assumption is that the portal cuts calls the way the business case said. That figure tells you how many certificate calls there are left for the portal to cut. It also shows whether certificate calls account for part of the unexplained drop since April 2026. The answer changes what the certificate-only path and the full build are each being asked to achieve.

    That is the framing. I have not ranked the options, scored them or recommended a course of action. Framing prepares the decision; it does not make it.

Written before the run. The operator agent was told never to volunteer its last section.

1. Turn 1. Operator agent, answering from the fact sheet

   Use Frame Free in this PDF. Ask me one question at a time.

   Our gross margin has dropped from 31% to 24% in a year, and I’m deciding between a 9% price rise and cutting deliveries from three a week to two. Customers need 30 days’ notice, so I want to send letters by 1 November.

2. Turn 2. Method agent, following the guide

   We’ll use Frame Free. I’ll ask one question at a time, and I’ll stop before ranking options or recommending anything.

   Here’s what I have so far:

   • Decision: a 9% price rise or cutting deliveries from three a week to two.
   • Context: gross margin fell from 31% to 24% over a year. Customers need 30 days’ notice, so you want letters out by 1 November.

   Before we move on, one point needs sharpening. The margin fell over a full year, so the drop on its own doesn’t explain why this is live now and wasn’t last quarter. What made it urgent now? For example, a specific month’s result, a cost increase you were told about, a cash pressure, or the 1 November date itself. If 1 November is the reason, where does that date come from?

3. Turn 3. Operator agent, answering from the fact sheet

   It was the September management accounts. They showed a 24% gross margin for the year to 30 September 2026, and that’s what made me act. The 1 November date comes from the notice period. Our standard terms let us change prices on 30 days’ written notice, and I want the change to start on 1 December, so the letters have to go out by 1 November.

4. Turn 4. Method agent, following the guide

   Thanks. That gives the decision a clear trigger. The September management accounts showed a 24% gross margin for the year to 30 September 2026. The 1 November date follows from your 30-day notice term and a chosen start date of 1 December. I’ll note that 1 December is a date you picked, not one the terms impose, and we’ll come back to that when we look at constraints.

   Question 2 is about the decision underneath this one.

   Suppose neither option existed: no price rise and no change to delivery frequency. What larger question would you still face? Put another way, a price rise and fewer deliveries are both means. What outcome are they meant to change, and do you know yet what caused the margin to fall from 31% to 24%?

5. Turn 5. Operator agent, answering from the fact sheet

   The outcome I’m trying to change is the gross margin. It fell from 31% ($992,000) to 24% ($768,000), a drop of $224,000, on revenue of about $3.2 million that’s roughly flat on last year. Without either option, I’d still be facing how to get that margin back.

   On the cause, I know part of it. Gas costs rose $96,000 over the year, and wages rose about $52,500, mostly from award increases, on a wage bill of about $1.4 million. Together that’s about $148,500 of the $224,000. I’ve assumed the rest is general costs, but I haven’t broken it down.

6. Turn 6. Method agent, following the guide

   Thanks. So the decision underneath is how to recover about $224,000 of gross margin on revenue that has stayed roughly flat at $3.2 million. A price rise and fewer deliveries are two ways to do that.

   Your figures check out. 24% of $3.2 million is $768,000, and 31% is $992,000. Gas ($96,000) and wages (about $52,500) together come to about $148,500. That leaves about $75,500, roughly a third of the drop, with no identified cause. You’ve assumed it is general costs, but you haven’t broken it down yet. I’ll carry that into the evidence map as unknown rather than settle it now, because it affects which part of the margin each option would act on.

   Question 3 is about the missing option.

   At the moment you have two paths: a 9% price rise, or cutting deliveries from three a week to two. Which genuine third path haven’t you treated as real yet? Keeping things as they are counts, and so does a smaller or slower version of either option. It only needs to be credible enough to look at seriously later. We won’t evaluate it now.

7. Turn 7. Operator agent, answering from the fact sheet

   Honestly, I don’t know. I’ve only been weighing those two: a 9% rise across the board, or dropping standard delivery from three a week to two at current prices. I haven’t seriously looked at a third path, whether that’s leaving things as they are or a smaller or slower version of either one.

8. Turn 8. Method agent, following the guide

   That’s a useful answer. It tells us the choice came to you as two options, and neither of them has been tested against a third. I won’t make one up for you, but I’ll ask one follow-up before we move on.

   Both of your options apply to every customer from 1 December: a 9% rise across the board, or two deliveries a week instead of three at current prices. The forms a third path usually takes are these:

   • Keep things as they are for a set period, for example until you know what the remaining $75,500 of the drop is.
   • A smaller version of either option, such as a lower rise or a delivery cut for some customers only.
   • A slower version of either option, such as a later start date or a change made in stages.

   Which of these could you credibly put on the table, even if you don’t like it yet? One is enough, and we won’t judge it now. If none of them is credible for your business, say so. I’ll record the third path as not yet identified and we’ll move on to the assumption the decision rests on.

9. Turn 9. Operator agent, answering from the fact sheet

   I don’t know. I haven’t looked at any of those, so I can’t tell you which one would be credible for us. Record the third path as not yet identified.

10. Turn 10. Method agent, following the guide

    Recorded. The third path is not yet identified, and the handoff will say so. That is a finding in its own right: the choice reached you as two options, and nothing has yet been tested against them.

    Question 4 is about the load-bearing assumption.

    Many things could go wrong with either option. I’m not asking for that list. I’m asking for the one belief that would change your decision if it proved wrong. Try finishing this sentence:

    “This decision makes sense provided that …”

    Then tell me what you have actually seen, measured or checked that supports it. For example, have you seen how customers reacted to a past price change, or worked out what moving from three deliveries a week to two would save in gas and wages? If the honest answer is that nothing has been checked yet, say so. That is useful to know at this stage.

11. Turn 11. Operator agent, answering from the fact sheet

    This decision makes sense provided that customers will accept a 9% rise, because costs are up everywhere. For the delivery option, it rests on cutting a delivery day saving running costs.

    Honestly, nothing has been checked on either. I haven’t asked any customers about the 9%. On a past price change, we put prices up 5% in 2024 and six customers left afterwards, but nobody followed up on why. On deliveries, I haven’t done any route or cost analysis, so I haven’t worked out what going from three to two a week would save in gas or wages.

12. Turn 12. Method agent, following the guide

    Thanks. Saying plainly that nothing has been checked is useful, and I’ll record it that way.

    You’ve given me two beliefs, one for each option:

    • Price rise: customers will accept a 9% rise because costs are up everywhere.
    • Delivery cut: dropping from three deliveries a week to two will save running costs.

    “Costs are up everywhere” is a reason customers might accept the rise. It isn’t evidence that they will. The only evidence you have points the other way and is incomplete: after the 5% rise in 2024, six customers left, and nobody found out why. I’ll record that as given, with the reason for leaving unknown.

    For a sense of scale: if every customer stayed, 9% on about $3.2 million would bring in about $288,000 a year. That’s my calculation from your figures, and it assumes nobody leaves. It’s more than the $224,000 drop, so the question becomes how much revenue you could lose before the rise no longer covers the gap. We won’t work that out now.

    My one follow-up on this question: the guide asks for the single belief the decision rests on, not one per option. If you could test only one belief before 1 November, which one would change what you do if it proved wrong? It could be one of the two above. It could also be something underneath both, for example that the cost increases you’ve identified are the main cause of the margin fall, when about $75,500 of it still has no identified cause. Pick the one that matters most to you, in your own words.

13. Turn 13. Operator agent, answering from the fact sheet

    If I could only test one, it would be whether customers will actually accept a 9% rise. If a lot of them would walk, the price rise stops making sense, and the last time we put prices up, 5% in 2024, six customers left and we never found out why. I take the point about the other $75,500, but I don’t know what’s in it yet, so I can’t say whether that’s the bigger issue.

14. Turn 14. Method agent, following the guide

    Recorded. The load-bearing assumption is that customers will accept a 9% rise. Nothing has been checked on it yet. The only evidence is the 5% rise in 2024, after which six customers left for reasons nobody found out. I’ll also note that you’ve put the $75,500 aside because you don’t know what’s in it yet, not because you’ve ruled it out.

    Question 5 is about untested constraints.

    Several things in this decision are being treated as fixed. For each one, tell me whether it is verified (someone has checked it in a document or with the person who set it) or assumed:

    1. Every customer is on your standard terms, and those terms allow a price change on 30 days’ written notice. Has anyone checked whether some customers have their own contracts, fixed prices or different notice periods?
    2. The 1 December start date, and so the 1 November deadline for letters. You chose 1 December. What makes it 1 December and not a later date?
    3. Delivery frequency can be cut from three a week to two on the same notice. Is three deliveries a week written into your terms or any customer contract, or is it a habit customers have come to expect?
    4. Anything else you’re treating as fixed: a budget, a bank or owner expectation, a capacity limit, or a commitment you’ve made to someone.

    For any you mark verified, say who checked it and when. “Our terms say so” counts as verified only if someone has read the clause recently.

15. Turn 15. Operator agent, answering from the fact sheet

    1. Partly verified. The 30 days’ written notice for price changes is in our customer terms, and that has been checked. I can’t tell you who checked it or when. Not every customer is on standard terms, though. Our 12 largest customers, about 38% of revenue, are on two-year contracts that only allow CPI increases on each contract anniversary.
    2. Assumed. I chose 1 December myself. Nothing I know of makes it 1 December rather than a later date.
    3. I don’t know. I haven’t checked whether three deliveries a week is written into our terms or any customer contract.
    4. The four vans are leased on fixed terms, and the drivers work fixed 38-hour weeks. I don’t know of any budget, bank or owner expectation that applies here.

16. Turn 16. Method agent, following the guide

    Thanks. That answer changes the shape of the decision more than anything so far, so here is how I’m recording it:

    1. The 30-day notice clause for price changes is in your customer terms. I’ll mark the clause as verified, but you can’t say who checked it or when, so I’ll note that gap. It applies only to customers on standard terms. Your 12 largest customers, about 38% of revenue, are on two-year contracts that allow only CPI increases on each contract anniversary. A 9% rise across the board from 1 December can’t reach them under those contracts, as you’ve described them.
    2. The 1 December start date is assumed. You chose it, and nothing you know of fixes it. That makes the 1 November deadline for letters assumed too.
    3. Whether three deliveries a week is written into your terms or any contract is unknown.
    4. The four van leases are on fixed terms and the drivers work fixed 38-hour weeks. I’ll record those as you gave them. I can’t tell from what you’ve said whether anyone has checked the lease terms or the employment arrangements recently.

    For scale only: if the 9% rise applied just to the roughly 62% of revenue on standard terms, and every one of those customers stayed, it would bring in about $178,560 a year (9% of about $1.98 million). That’s my calculation from your figures, not a forecast, and we won’t weigh it now.

    I’ll also note something without judging it. Your delivery option rests on the belief that cutting a delivery day saves running costs. The vans are on fixed leases and the drivers are on fixed hours, so it’s an open question which costs would actually fall. That goes into the evidence map as inferred, not settled.

    My one follow-up on this question: has anyone read the 12 two-year contracts recently, and do you know (a) when each one’s anniversary falls and (b) whether they say anything about delivery frequency? If nobody has, say so and I’ll record both as unknown. Then we’ll move to the last question, the evidence map.

17. Turn 17. Operator agent, answering from the fact sheet

    I don’t know whether anyone has read the 12 contracts recently. I also don’t know when each one’s anniversary falls, or whether they say anything about delivery frequency. Record all of that as unknown.

18. Turn 18. Method agent, following the guide

    Recorded. Nobody is known to have read the 12 contracts recently. Their anniversary dates are unknown, and so is whether they say anything about delivery frequency.

    Question 6 is the evidence map, the last question. Every material claim gets one of four tags. GIVEN means you supplied it or it has been checked. DERIVED means it is calculated from given figures. INFERRED means it is reasoned from what is known. UNKNOWN means it matters but hasn’t been established. Here is my draft from what you’ve told me.

    GIVEN

    • The September management accounts show a 24% gross margin for the year to 30 September 2026, down from 31% a year earlier.
    • Revenue is about $3.2 million, roughly flat on last year.
    • Gas costs rose $96,000 over the year.
    • Wages rose about $52,500, mostly from award increases, on a wage bill of about $1.4 million.
    • Your customer terms allow price changes on 30 days’ written notice. The clause has been checked, but you don’t know who checked it or when.
    • Your 12 largest customers, about 38% of revenue, are on two-year contracts that allow only CPI increases on each contract anniversary.
    • Prices rose 5% in 2024, and six customers left afterwards.
    • The four vans are leased on fixed terms, and the drivers work fixed 38-hour weeks.
    • You chose 1 December as the start date. Nothing you know of fixes it.

    DERIVED

    • Gross margin fell from about $992,000 to about $768,000, a drop of about $224,000.
    • Gas and wages together explain about $148,500 of that, which leaves about $75,500 with no identified cause.
    • The 1 November deadline for letters follows from the 1 December start date and the 30-day notice term.
    • If every customer stayed, a 9% rise on all revenue would bring in about $288,000 a year.
    • If the rise reached only the roughly 62% of revenue on standard terms (about $1.98 million) and every one of those customers stayed, it would bring in about $178,560 a year. That is about $45,000 less than the $224,000 drop. This is a calculation, not a forecast.

    INFERRED

    • Cutting from three deliveries a week to two will save running costs. With fixed van leases and fixed driver hours, it is open which costs would actually fall.

    UNKNOWN

    • Whether customers will accept a 9% rise. This is the load-bearing assumption, and nothing has been checked.
    • Why the six customers left after the 2024 rise.
    • What makes up the remaining $75,500. You’ve assumed general costs, but it hasn’t been broken down.
    • What going from three deliveries to two would save in gas or wages.
    • Whether three deliveries a week is written into your terms or any customer contract.
    • When each of the 12 contracts’ anniversaries falls, whether they mention delivery frequency, and whether anyone has read them recently.
    • Whether anyone has recently checked the van lease terms or the drivers’ employment arrangements.

    My one follow-up: is any tag here wrong, or is a material claim missing? For example, a figure I’ve marked GIVEN that is really your own estimate, or something you know about customers, costs or the contracts that you haven’t mentioned yet. If the map is right as it stands, say so and I’ll write up the full framing handoff.

19. Turn 19. Operator agent, answering from the fact sheet

    The tags look right to me. Two things are missing.

    First, we replaced more linen stock this year than usual. I haven’t pulled the amount out of the accounts, so I can’t tell you how much it was. That belongs under unknown.

    Second, I don’t know what competitors charge now. Please add that as unknown too. I also don’t know how many customers would leave at 9%, but I think your load-bearing item already covers that.

20. Turn 20. Method agent, following the guide

    Thanks. I’ve added both to the map as unknown: the extra linen stock replaced this year, and what competitors charge now. You’re right that how many customers would leave at 9% is already covered by the load-bearing item, so I haven’t listed it twice.

    That completes the six questions. Here is the framing handoff.

    1. Presented decision and trigger

    Choose between a 9% price rise across the board and cutting standard delivery from three a week to two at current prices, with letters out by 1 November for a 1 December start. The trigger was the September management accounts, which showed a 24% gross margin for the year to 30 September 2026, down from 31%.

    2. Decision underneath

    How to recover about $224,000 of gross margin on revenue that has stayed roughly flat at about $3.2 million. The price rise and the delivery cut are two means to that outcome. About $148,500 of the fall has an identified cause (gas and wages). About $75,500 does not.

    3. Missing third option

    Not yet identified. The choice reached you as two options, and no third path has been tested against them. Keeping things as they are, a smaller version of either option, or a slower one have not been looked at. The handoff records that gap rather than filling it.

    4. Load-bearing assumption

    “This decision makes sense provided that customers will accept a 9% rise.” Nothing has been checked on it. The only evidence is the 5% rise in 2024, after which six customers left for reasons nobody found out. You set aside the unexplained $75,500 because you don’t know what’s in it yet, not because you ruled it out.

    5. Untested constraints

    • 30 days’ written notice for price changes under your customer terms: VERIFIED. The clause has been checked, but you don’t know who checked it or when.
    • All customers can be reached by a price change on that notice: ASSUMED, and contradicted by what you’ve told me. Your 12 largest customers, about 38% of revenue, are on two-year contracts that allow only CPI increases on each anniversary.
    • The terms of those 12 contracts as you’ve described them: ASSUMED. Nobody is known to have read them recently.
    • The 1 December start date, and so the 1 November deadline for letters: ASSUMED. You chose it, and nothing you know of fixes it.
    • Delivery frequency can be cut from three a week to two on the same notice: ASSUMED. Nobody has checked whether three a week is written into your terms or any contract.
    • Fixed van leases and fixed 38-hour driver weeks: ASSUMED. Stated as fact, but it isn’t known whether anyone has checked the lease terms or employment arrangements recently.
    • No budget, bank or owner expectation applies: ASSUMED. None that you know of, but none has been checked.

    6. Evidence map

    GIVEN

    • The September management accounts show a 24% gross margin for the year to 30 September 2026, down from 31% a year earlier.
    • Revenue is about $3.2 million, roughly flat on last year.
    • Gas costs rose $96,000 over the year.
    • Wages rose about $52,500, mostly from award increases, on a wage bill of about $1.4 million.
    • Customer terms allow price changes on 30 days’ written notice. Checked, but by whom and when is not known.
    • The 12 largest customers, about 38% of revenue, are on two-year contracts allowing only CPI increases on each anniversary.
    • Prices rose 5% in 2024, and six customers left afterwards.
    • The four vans are leased on fixed terms, and drivers work fixed 38-hour weeks.
    • You chose 1 December as the start date.

    DERIVED

    • Gross margin fell from about $992,000 to about $768,000, a drop of about $224,000.
    • Gas and wages explain about $148,500 of that, leaving about $75,500 with no identified cause.
    • The 1 November deadline follows from the 1 December start date and the 30-day notice term.
    • If every customer stayed, 9% on all revenue would bring in about $288,000 a year.
    • If the rise reached only the roughly 62% of revenue on standard terms (about $1.98 million) and every one of those customers stayed, it would bring in about $178,560 a year, about $45,000 short of the $224,000 drop. A calculation, not a forecast.

    INFERRED

    • Cutting from three deliveries a week to two will save running costs. With fixed van leases and fixed driver hours, it is open which costs would actually fall.

    UNKNOWN

    • Whether customers will accept a 9% rise (the load-bearing assumption), including how many would leave.
    • Why the six customers left after the 2024 rise.
    • What makes up the remaining $75,500. Assumed to be general costs, not broken down.
    • How much was spent replacing linen stock this year, above the usual level.
    • What competitors charge now.
    • What going from three deliveries to two would save in gas or wages.
    • Whether three deliveries a week is written into your terms or any customer contract.
    • When each of the 12 contracts’ anniversaries falls, whether they mention delivery frequency, and whether anyone has read them recently.
    • Whether anyone has recently checked the van lease terms or the drivers’ employment arrangements.

    7. Reframed decision and the single evidence gap to verify next

    Reframed decision: how to recover about $224,000 of gross margin when about a third of the fall has no identified cause, about 38% of revenue sits on contracts that a 9% rise from 1 December cannot reach as you’ve described them, the 1 December start is a date you chose, and no third path has yet been put alongside the two you inherited.

    Evidence gap to verify next: what the 12 two-year contracts actually say, meaning each anniversary date, the exact price-change clause, and anything on delivery frequency. Reading them settles how much of your revenue either option can reach, and when. That changes the scale of the question before customer acceptance is even tested.

    I’ll stop here. Nothing has been ranked, scored or recommended. Framing prepares the decision; it doesn’t make it.

Written before the run. The operator agent was told never to volunteer its last section.