Build a strategy review decision record
Compare observed evidence with prior assumptions before authorised people choose the next commitment.
- Preparation
- Until the evidence owners, authorised decision owners, independent counsel, and relevant jurisdiction advisers complete their reviews
- Steps
- 3
- Difficulty
- Advanced
- Task
- Record a pivot-or-persevere decision
Requires: Confirm the company body and people authorised to own this decision. This Practice provides an internal evidence workflow. It does not provide legal, financial, tax, investment, or valuation advice.
Uses: ChatGPT evidence workspace
Review boundary: Prepare internal facts from an approved extract with neutral party keys and source links. Keep originals in the approved repository. Require qualified review before changing rights, making an offer, completing a transaction or publishing externally.
Back to the Task: Record a pivot-or-persevere decisionBefore you start
Prepare
- Confirm the company body and people authorised to own this decision.
- This Practice provides an internal evidence workflow. It does not provide legal, financial, tax, investment, or valuation advice.
- Obtain named pre-upload approval for the workspace, plan, account settings, retention settings, and each approved document category.
- Treat privileged, personal, customer, employee, security, and third-party information as prohibited. Keep it out of this workspace and use a separately approved system.
- Verify plan-dependent data controls and retention before use. Test access with a non-owner account before sharing.
- Keep unsupported facts, missing documents, conflicts, and disagreement visible.
- Treat Tool output as an internal draft that a person checks against every cited source.
- Obtain independent counsel and qualified advisers’ review of every external communication and transaction decision in each applicable jurisdiction before use.
Steps
Work through the method
Use the listed inputs and tools. Check the evidence when you need to verify a step.
Compare expected and observed evidence
Restate the dated strategy, model, assumptions, review trigger, and expected evidence. Add actual observations without rewriting the original hypothesis, and label explanations as hypotheses.
Step details and sources
Why it matters: The review needs the original expectation and current observation in separate fields.
- Input
- The dated strategy, original assumptions, agreed evidence, current observations, and constraints.
- Output
- A versioned comparison of expected evidence, observations, and explanation hypotheses.
Tools in this stepChatGPT evidence workspaceEvidence for this step
Strategyzer presents the Business Model Canvas as nine connected business-model blocks and says a completed canvas starts discussion rather than deciding the strategy.
Strategyzer: “Why use the Business Model Canvas?” and “Download the Business Model Canvas”The framework structures a dated model. It does not rank initiatives, validate assumptions, or prove viability.Strategyzer instructs users to distinguish present and future business-model states and to identify assumptions.
Strategyzer: “Define present vs future states” and “Acknowledge assumptions”The checklist structures review. It does not validate an assumption or authorise a direction change.OpenAI says file availability and limits depend on the plan and account settings, retention varies by use and plan, and deletion can take up to 30 days subject to stated exceptions.
OpenAI: “Availability”, “How do I delete files I upload?”, “How are files vs chats retained?”, and “Will OpenAI use files uploaded to train its models?”; version checked 26 September 2026The FAQ does not establish an organisation’s approval, access controls, or retention policy. Deletion is not described as immediate and has de-identification, security, and legal exceptions.
Compare four strategy options
Describe persevere, bounded pivot, pause, and closure as separate options. For each, record the smallest coherent commitment, remaining constraint, reversible step, stakeholder duty, and evidence that would reopen it.
Step details and sources
Why it matters: Explicit options prevent a copied founder threshold from becoming an automated viability judgement.
- Input
- The evidence comparison, current constraints, and stakeholder obligations.
- Output
- A four-option consequence record with no score, ranking, or Tool recommendation.
Tools in this stepChatGPT evidence workspaceEvidence for this step
Gumroad’s founder describes stalled growth, a venture-financed plan that no longer fitted, a defined attempt to change trajectory, and a later move to a smaller model.
Sahil Lavingia: “Failing in style”, lines 39–58; “From skeleton crew to lifestyle business”, lines 90–94 in the checked research captureGumroad’s thresholds, sequence, and later model are company-specific retrospective facts, not decision rules or viability benchmarks.
Review consequences and decide
Finance, Product, People, and independent legal, insolvency, tax, and board advisers review their domains. The authorised body records persevere, pivot, pause, or close, owners, consequences, and review date.
Step details and sources
Why it matters: A strategy choice can trigger duties and consequences outside the Strategy work area.
- Input
- The four-option record, domain reviews, company authority rules, and unresolved matters.
- Output
- A governed strategy decision with named owners and no automated viability judgement.
Tools in this stepChatGPT evidence workspaceEvidence for this step
Strategyzer instructs users to distinguish present and future business-model states and to identify assumptions.
Strategyzer: “Define present vs future states” and “Acknowledge assumptions”The checklist structures review. It does not validate an assumption or authorise a direction change.
Success checks
Check the result
- The record preserves original assumptions and separates observations from later explanations.
- Persevere, bounded pivot, pause, and closure each show commitments, constraints, duties, and reopening evidence.
- Domain owners reviewed consequences before the authorised body recorded one option and its review date.
Failure modes
What could go wrong
- The record rewrites an original assumption or presents a later explanation as an observation. Restore the dated hypothesis and label the explanation separately.
- A copied founder threshold, score, ranking, or Tool recommendation decides whether the company should continue. Remove it and return the choice to the authorised body.
- A path advances while cash, workforce, contract, investor, insolvency, tax, or board consequences remain unreviewed. Stop and escalate each consequence to its qualified owner.
Tools
Choose tools
ChatGPT evidence workspace
A restricted workspace for organising supplied company records into internal tables that authorised people and independent advisers check.
Check fit, limits and pricingSources
Check the sources
Check what each source supports and where the advice has limits.
- Sahil Lavingia: Reflecting on My Failure to Build a Billion-Dollar CompanyFirst-hand founder retrospective · Publication date unavailable
- Strategyzer: The Business Model CanvasOfficial framework guidance · Publication date unavailable
- Strategyzer: How to create a crystal-clear Business Model CanvasOfficial framework guidance · Publication date unavailable
- OpenAI: File Uploads FAQOfficial product help · Publication date unavailable