Capability template for product management
Assumption map
For a product owner whose plan rests on beliefs nobody has tested: each belief is written down, set against the evidence you hold, ranked by what it would do to the plan, and given a cheap test to choose from. You decide what to test; no evidence is invented.
Listed under
Add it to a workflow in the Fully Powered Planner, beside the teams and roles that will run it. Book a fit call
The outcome
The product owner sees, on one page, what the plan is standing on. Every plan item is listed with the beliefs behind it, and every belief carries one of four marks: held, thin, contradicted or none. The mark follows a fixed rule from the evidence you supply, so a friendly interview cannot stand in for something people did, and bad news cannot be left out.
The beliefs that are not held come out in rank order. They are ranked first by what each would do to the plan, then by how little backs it. For the top few the page offers a cheap test suited to the kind of belief, and leaves the learning criteria empty for you to fill in before anything is run.
Nothing is decided for you. The page ends with a choice line for each belief shown: test now, later, or accept the risk. It adds no evidence of its own and reports no result it was not given.
Who starts the work, and when
The product owner starts it when a concept becomes a plan with a date: before the build is approved, before a launch is committed, or when a plan item changes and the beliefs behind it may have changed too. Run it again when a test comes back, and the marks move.
The product owner lists the plan items, writes down the beliefs they depend on, and supplies what evidence exists. The AI assistant lays the map out and ranks it. The product owner confirms every belief, chooses what to test and sets what result would count.
What they supply
- A short brief. The concept, who owns the decision, the decision and its date, the review date and any context worth knowing.
- The plan, one item per row: what the plan does or commits to.
- The beliefs, one per row: what has to be true, the kind of belief, which plan items depend on it, what happens to the plan if it is false, and a test idea if you already have one.
- The evidence, one item per row: what was seen, what kind it was, when, which beliefs it bears on, and whether it supports or contradicts them.
What comes out
One page in a fixed order. The plan with the beliefs under each item and their marks; every belief with the evidence for it, against it and neither way, and what is missing; evidence not tied to any belief; the unheld beliefs in rank order; a cheap test for the first few, either yours as typed or a suggestion from a fixed list; the questions only the product owner can answer; and an empty choice line for each belief shown.
Where something is missing, such as a plan item with no belief or evidence with no date, the page says so and asks. It never fills the gap itself.
An invented example
Illustrative example with invented data. The company, its customers, its people and every date and figure below are invented. "Company A" is an invented software company whose invoicing app is used by small trades businesses, and its product owner is planning a paid add-on that lets a customer attach job photos to an invoice. People appear by role only. First come the four inputs the product owner supplies (the brief, the plan, the beliefs and the evidence), then the map.
The inputs
1. The brief (written by the product owner)
Concept: Job photo reports, a paid add-on to Company A's invoicing app
Product owner: Product owner
Decision this serves: Whether to start building in December 2026 for a March 2027 launch
Decision date: Friday 20 November 2026
Review date: Tuesday 10 November 2026
Context: Company A has 1,400 paying accounts. 9 customers were interviewed in September 2026.
Settings: defaults
2. The plan (typed by the product owner)
id,plan_item
P01,"Add photos to any invoice from the phone app, up to 10 photos per invoice."
P02,"Sell it as a $9 a month add-on to existing accounts from March 2027."
P03,"Announce it to existing customers by email and in the app, with no paid advertising."
P04,"One engineer and one designer build it from December 2026 to February 2027."
P05,"Store the photos with the cloud storage supplier Company A already uses."
P06,"Answer customer questions through the existing help desk with no new hires."
3. The beliefs (typed by the product owner)
id,belief,type,plan_items,if_false,test_idea
B01,"Tradespeople want proof of finished work to send with the invoice.",customer,P01;P02,stops,
B02,"Enough existing customers will pay $9 a month for it.",behaviour,P02,stops,"Ask 10 active accounts for a written pre-order at $9 a month"
B03,"Customers will open the announcement and turn the add-on on.",behaviour,P03,changes,
B04,"One engineer and one designer can finish it by the end of February 2027.",delivery,P02;P04,changes,
B05,"Photos upload reliably from a phone on a job site with a weak signal.",delivery,P01,changes,
B06,"Photos at up to 10 per invoice stay inside the current monthly storage budget.",delivery,P05,small,
B07,"The phone app stores allow us to bill the add-on through our own checkout.",environment,P02,changes,
4. The evidence (typed by the product owner)
id,evidence,kind,date,beliefs,direction
E01,"31 of the last 120 support tickets asked how to show a customer a photo of the finished job.",measured,2026-10,B01,supports
E02,"4 of the 9 customers interviewed said they would pay for it.",said,2026-09,B02,supports
E03,"Three competing apps charge between $6 and $15 a month for photo features.",measured,2026-08,B02,unclear
E04,"The 2025 reminders add-on was turned on by 11% of accounts in its first 3 months.",behaviour,2025-09,B03,supports
E05,"A test upload of a 4 MB photo over a slowed connection failed 3 times out of 5 on a spare phone.",measured,2026-10,B05,contradicts
E06,"Engineering's estimate for one engineer and one designer is 14 weeks.",estimate,2026-10,B04,contradicts
E07,"The October 2026 storage bill is 38% of the monthly storage budget.",measured,2026-10,B06,supports
E08,"6 of the 9 customers interviewed said they send job photos by text message today.",said,2026-09,B01,supports
The map
ASSUMPTION MAP: Job photo reports, a paid add-on to Company A's invoicing app
Concept: Job photo reports, a paid add-on to Company A's invoicing app.
Product owner: Product owner. Decision this serves: Whether to start building in December 2026 for a March 2027 launch. Decision date: Friday 20 November 2026.
Context: Company A has 1,400 paying accounts. 9 customers were interviewed in September 2026.
Review date: Tuesday 10 November 2026. Prepared from the plan, the beliefs list and the evidence list; default settings. No evidence has been added.
THE PLAN AND WHAT IT RESTS ON
P01 Add photos to any invoice from the phone app, up to 10 photos per invoice.
Rests on: B01 HELD; B05 CONTRADICTED.
P02 Sell it as a $9 a month add-on to existing accounts from March 2027.
Rests on: B01 HELD; B02 THIN; B04 THIN; B07 NONE.
P03 Announce it to existing customers by email and in the app, with no paid advertising.
Rests on: B03 THIN.
P04 One engineer and one designer build it from December 2026 to February 2027.
Rests on: B04 THIN.
P05 Store the photos with the cloud storage supplier Company A already uses.
Rests on: B06 HELD.
P06 Answer customer questions through the existing help desk with no new hires.
Rests on: No belief written.
Plan items: 6. Resting only on held beliefs: 1. Exposed: 4. With no belief written: 1.
BELIEFS, EVIDENCE HELD AND MISSING
B01 customer. HELD. Tradespeople want proof of finished work to send with the invoice.
Plan items: P01, P02. If false: Stops the plan.
For:
E01 measured, 2026-10: 31 of the last 120 support tickets asked how to show a customer a photo of the finished job.
E08 said, 2026-09: 6 of the 9 customers interviewed said they send job photos by text message today.
Against: none.
Neither way: none.
Missing: Nothing flagged.
B02 behaviour. THIN. Enough existing customers will pay $9 a month for it.
Plan items: P02. If false: Stops the plan.
For:
E02 said, 2026-09: 4 of the 9 customers interviewed said they would pay for it.
Against: none.
Neither way:
E03 measured, 2026-08: Three competing apps charge between $6 and $15 a month for photo features.
Missing: No supporting behaviour or measured item within 12 months.
B03 behaviour. THIN. Customers will open the announcement and turn the add-on on.
Plan items: P03. If false: Changes the plan.
For:
E04 behaviour, 2025-09: The 2025 reminders add-on was turned on by 11% of accounts in its first 3 months.
Against: none.
Neither way: none.
Missing: No supporting behaviour or measured item within 12 months. Latest supporting item: E04, 2025-09, 14 months before the review date.
B04 delivery. THIN. One engineer and one designer can finish it by the end of February 2027.
Plan items: P02, P04. If false: Changes the plan.
For: none.
Against:
E06 estimate, 2026-10: Engineering's estimate for one engineer and one designer is 14 weeks.
Neither way: none.
Missing: No supporting behaviour or measured item within 12 months. Points against: E06 (estimate).
B05 delivery. CONTRADICTED. Photos upload reliably from a phone on a job site with a weak signal.
Plan items: P01. If false: Changes the plan.
For: none.
Against:
E05 measured, 2026-10: A test upload of a 4 MB photo over a slowed connection failed 3 times out of 5 on a spare phone.
Neither way: none.
Missing: Contradicted by E05 (measured).
B06 delivery. HELD. Photos at up to 10 per invoice stay inside the current monthly storage budget.
Plan items: P05. If false: Small change.
For:
E07 measured, 2026-10: The October 2026 storage bill is 38% of the monthly storage budget.
Against: none.
Neither way: none.
Missing: Nothing flagged.
B07 environment. NONE. The phone app stores allow us to bill the add-on through our own checkout.
Plan items: P02. If false: Changes the plan.
For: none.
Against: none.
Neither way: none.
Missing: No evidence supplied.
Beliefs: 7. Held 2, thin 3, contradicted 1, none 1.
EVIDENCE NOT TIED TO A BELIEF
None.
RANKED BELIEFS NOT YET HELD
Order: what happens if false, then the mark, then reach, then the id.
1. B02 behaviour. THIN. Stops the plan. Reaches 1 plan item. Enough existing customers will pay $9 a month for it.
2. B05 delivery. CONTRADICTED. Changes the plan. Reaches 1 plan item. Photos upload reliably from a phone on a job site with a weak signal.
3. B07 environment. NONE. Changes the plan. Reaches 1 plan item. The phone app stores allow us to bill the add-on through our own checkout.
4. B04 delivery. THIN. Changes the plan. Reaches 2 plan items. One engineer and one designer can finish it by the end of February 2027.
5. B03 behaviour. THIN. Changes the plan. Reaches 1 plan item. Customers will open the announcement and turn the add-on on.
Not yet rated: none.
SUGGESTED CHEAP TESTS FOR THE FIRST 3
1. B02 Enough existing customers will pay $9 a month for it.
Test: Ask 10 active accounts for a written pre-order at $9 a month
Learning criteria: owner to set.
2. B05 Photos upload reliably from a phone on a job site with a weak signal.
Test: Suggested: a time-boxed trial of the exact step in doubt, run by the team that would build or run it, with the result written down.
If that is not possible: ask that team to say in writing what they expect to go wrong and how they would find out.
Learning criteria: owner to set.
3. B07 The phone app stores allow us to bill the add-on through our own checkout.
Test: Suggested: read the published rules or terms for the exact point and note the passage.
If that is not possible: put the question in writing to the person who owns the rule, which a person sends.
Learning criteria: owner to set.
Beyond the first 3: B04, B03.
A behaviour or measured result moves a belief toward held.
What someone says does not.
OPEN QUESTIONS FOR THE PRODUCT OWNER
1. P06: no belief is written for this plan item. It holds that: Answer customer questions through the existing help desk with no new hires. Is that a belief the plan stands on, to be written into the beliefs list?
2. B03: the latest supporting item, E04, is dated 2025-09, 14 months before the review date; the limit is 12 months. Is there a more recent item to add to the evidence list?
3. Learning criteria: for B02, B05 and B07, what result would count as the belief holding, and what as failing, before any test starts?
OWNER'S CHOICE
B02 [ ] Test now [ ] Later [ ] Accept the risk
B05 [ ] Test now [ ] Later [ ] Accept the risk
B07 [ ] Test now [ ] Later [ ] Accept the risk
Prepared for the Product owner's decision. Nothing has been decided, tested, sent or changed. No evidence has been added.
The product owner reads the map, answers the three questions and ticks one box for each belief shown. The map only prepared it.
Where it fits
- Product management, when a roadmap item, launch or build rests on things no one has yet checked.
- Software and technology, where a feature plan often runs ahead of what customers have been seen to do.
- Hardware & physical technology, where a wrong belief about supply, build or fit is dear to correct once tooling starts.
- Consumer brands, when a new product or format is planned around how shoppers are assumed to behave.
- Startups and new ventures, where most of the plan is still belief.
- New business lines, before a team moves onto a line whose demand is assumed.
- Growth & scaling, when what worked at one size is planned for another.
It is not the commercial review of a whole business model: Business model assumptions review does that. It does not imagine the plan failing in a workshop: Premortem before a decision does that. This one starts from the plan's own items and asks what each is standing on.
What has to change for your company
Decide who owns the map and who may mark a belief as supported or contradicted. Keep or change how old evidence may be before it is rechecked (twelve months by default) and how many beliefs get a suggested test (three). Add your own kinds of evidence if some of your records count for more than the default four. Replace the suggested tests with the cheap tests your team can actually run, such as a pre-order page or a hand-run version of a step. Decide where the maps are kept, so the next one starts from the last and shows which beliefs moved. It runs in ChatGPT, Claude, Copilot or Gemini from the lists you type or paste; it needs no connection to any system.
What has been checked
Fully Powered AI wrote this capability template. It was editorially reviewed on 10 November 2026. It was tried once on its invented example in Claude on 10 November 2026; the record is published with the template. It has not yet been tested in a client environment or as a packaged version in ChatGPT, Claude, Copilot or Gemini.
Testing in your environment means running it on a plan whose outcome you already know, with the beliefs and evidence written as they stood then, plus hard cases written in advance: a plan item with no belief, evidence with no date, an estimate that points against a belief, a rating outside the scale. The product owner checks every mark and the ranking against the lists before relying on the page. Offer review takes the offer behind a belief further.
The full capability template
Everything above is the outline. The complete capability template is the working version: the roles, the inputs, the procedure, the exceptions and stop conditions, the output format, the reviewer checklist, the acceptance tests, the instructions for an AI assistant and what to customize for your company. It comes with a download and the link to its folder in Fully Powered’s public repository.
Free with an account. Sign in by email: no password and no card.
Next step
Put this capability to work for your team: on your plan, or built with you in a sprint.
Build this in a sprint
In a sprint, Fully Powered adapts this capability template to your rules, proves it on your own work and puts it in the hands of the roles that own it.
Two ways to begin: