Practical guide
AI go-to-market brief verification checklist
A polished go-to-market brief can still fail at the decision point. Verification asks whether its segment, problem, promise, channel, offer, and learning plan are traceable to current evidence and feasible for the team that must deliver them. Use this checklist after drafting and before a launch, pilot, or campaign decision.
A polished go-to-market brief can still fail at the decision point. Verification asks whether its segment, problem, promise, channel, offer, and learning plan are traceable to current evidence and feasible for the team that must deliver them. Use this checklist after drafting and before a launch, pilot, or campaign decision.
Set the verification boundary
Complete these fields before reviewing the brief:
If the brief has no specific decision or product boundary, return it to the author. A checklist cannot verify a moving target.
- [ ] Decision to approve: pilot, launch, repair, hold, or stop
- [ ] Named decision owner and decision date
- [ ] Product version, geography, segment, offer, and launch window
- [ ] Frozen source manifest with owner, date, and evidence type
- [ ] Delivery, sales, support, policy, and claim reviewers
- [ ] Alternatives, including the current process and no launch
Run the PROVE checks
PROVE tests five independent parts of the brief. A failure in one part should not be hidden by strong evidence in another.
### P: Problem and first segment
The U.S. Small Business Administration suggests examining demand, market size, economic indicators, location, saturation, and pricing, then using competitive analysis to understand alternatives and possible advantage. It also identifies surveys, questionnaires, focus groups, and in-depth interviews as direct-research methods. These are research categories, not evidence that a particular segment will purchase. SBA: Market research and competitive analysis
### R: Rival and status-quo comparison
Ask a reviewer to reconstruct the alternative map from the manifest without seeing the conclusion. Material differences should be explained, not averaged away.
### O: Offer and operational readiness
A market argument does not establish delivery readiness. Record an explicit hold when the product promise depends on an unavailable capability or an unstaffed service.
### V: Value promise and claim support
The Federal Trade Commission’s small-business guidance explains that advertising must be truthful and non-deceptive, that advertisers need evidence for claims before publication, and that both express and implied meanings matter. It also notes that fine print cannot repair a misleading main message. Apply the evidence discipline to the complete asset; obtain market-specific legal review where required. FTC: Advertising FAQs
### E: Experiment and exit gates
Do not approve a test whose only outcome is more activity. A useful test distinguishes at least two plausible decisions.
- [ ] The first segment has observable inclusion and exclusion criteria.
- [ ] The buyer, user, reviewer, and decision owner are distinguished where they differ.
- [ ] Problem evidence comes from dated sources that represent the stated segment.
- [ ] Reported pain, observed workflow, inference, and hypothesis are labelled separately.
- [ ] Sample limitations and missing perspectives are visible.
- [ ] The brief includes direct products, adjacent tools, internal workarounds, and doing nothing.
- [ ] Comparisons use equivalent scope, date, geography, and buyer context.
- [ ] Competitor statements link to current first-party material where possible.
- [ ] An absent public feature is labelled “not found in reviewed material,” not “does not exist.”
- [ ] The proposed advantage follows from evidence rather than adjectives.
- [ ] The offer states price or pilot terms, scope, exclusions, and expiry.
- [ ] Product, onboarding, support, and review capacity match the proposed volume.
- [ ] Required security, legal, accessibility, or procurement reviews have owners.
- [ ] The sales path and handoffs are workable for the named buying group.
- [ ] Unit assumptions, service effort, and dependencies have been recalculated.
- [ ] Every material promise appears in a claim ledger.
- [ ] The ledger records exact wording, implied meaning, support, reviewer, and expiry.
- [ ] Product facts are checked against the shipped product and approved evidence.
- [ ] Quantified outcomes use a defined denominator, comparison, period, and source.
- [ ] Disclosures are near the claim and do not contradict the main impression.
- [ ] The largest uncertainty has a test that can change the decision.
- [ ] The population, channel, message, period, and evidence capture are defined.
- [ ] Scale, repair, hold, and stop conditions were set before results.
- [ ] Reach, response, qualified interest, purchase, and delivery outcomes remain distinct.
- [ ] The next evidence gate has an owner and review date.
Sources: [1] [2]
Copyable verification record
| Check ID | Brief claim or decision | Source and date | Verification performed | Result | Exception | Owner | Disposition | | --- | --- | --- | --- | --- | --- | --- | --- | | GTM-01 | exact statement | source anchor | comparison, recalculation, or review | pass / fail / missing | unresolved issue | named person | approve / repair / hold / stop |
Attach the source manifest, alternative map, delivery-capacity check, claim ledger, calculation sheet, reviewer notes, and approved experiment card. A slide deck alone is not a reproducible evidence packet.
Worked example: verification changes the pilot
This synthetic brief proposes a broad launch of a desktop AI-fluency assessment for “all AI-forward employers.” It cites eight interviews, two product pilots, competitor pages, and a delivery estimate. The planned message says the product shows whether a workforce is productive with AI, and the channel plan assumes a paid campaign can generate forty pilots in one quarter.
Problem and segment: the reviewer traces the interview notes and finds that six interviewees are learning or transformation leaders running structured capability programmes. None represents the full employer market. PROVE fails the broad segment and replaces it with explicit inclusion criteria for a first learning-programme cohort.
Rival comparison: the brief treats training surveys as the only alternative. The reviewer adds manager observation, course completion, adoption analytics, and internally built simulations. This does not invalidate the proposed assessment, but it changes the comparison and the buyer questions.
Operational readiness: the delivery estimate supports four concurrent pilots, not forty. Security review and scenario adaptation also require lead time. The offer is narrowed to a six-week, four-customer pilot with synthetic or approved non-sensitive materials.
Value promise: the two product pilots contain task-bounded assessment evidence but no workforce productivity comparison. The stronger statement fails the claim ledger. The replacement says the product creates a consistent opportunity to observe realistic AI-enabled work and provides evidence for human interpretation.
Experiment: paid reach is no longer the first test. The revised gate asks whether four qualified programme owners can use the evidence packet to make a defined learning decision without excessive review burden. Scale requires traceable evidence, acceptable delivery effort, and a documented follow-on decision. The verified disposition is pilot after repair, not broad launch.
Independent review and hold conditions
Use a reviewer who did not write the brief. Have them restate the decision, locate each material source, recalculate the economics, identify the weakest assumption, and explain what evidence would reverse the recommendation. Expand the review when any source is invented, stale, incomparable, or materially misquoted.
Hold the brief when the decision is undefined, the segment cannot be identified operationally, material research is stale or unrepresentative, the offer exceeds delivery capacity, a claim lacks support, the alternative set is incomplete, or the learning test has no decision threshold.
Two Prune can use go-to-market work as a realistic assessment task and evaluate how a participant frames a decision, navigates evidence, uses AI, checks claims, exercises judgment, and communicates a brief. It does not determine demand or make employment decisions.
Sources
U.S. Small Business Administration: Market research and competitive analysis
- Federal Trade Commission: Advertising FAQs, a guide for small business
Sources: [1] [2]
Sources
- 1.SBA: Market research and competitive analysis · www.sba.gov
- 2.FTC: Advertising FAQs · Federal Trade Commission