The Buyer Verification Playbook

Companion to Report 005, “Can Buyers Verify Your Market-Leading Claim?” The report establishes what public evidence can prove. This playbook turns that boundary into a decision sequence. It is built for one claim, one named buying group, one decision owner, and one decision date.

The scene this playbook is built for

Your team wants to say “the leading AI platform for regulated teams.”

Marketing can point to the homepage. Sales has three customer stories. Product has a security packet. An AI answer describes the company as a workflow tool. A review site places it in a neighboring category. One competitor claims the same leadership position. Procurement will ask for the evidence behind “regulated.”

The claim may be right. The evidence around it is carrying five different versions.

That is the moment for buyer verification. The job is to decide whether to carry, revise, verify, pause, or stop the claim before the company spends money teaching it to the market.

The five-part brief

Write these five lines before opening a search result.

1. The claim: one sentence, quoted exactly as it will appear.
2. The buying group: the commercial champion plus the roles that can slow or stop the decision.
3. The decision owner: one person with authority to accept an adverse finding.
4. The decision date: the day the claim or the money behind it becomes hard to change.
5. The consequence: what gets funded, printed, briefed, sold, or promised if the claim survives.

“We need to understand trust” is too broad. “Decide by October 12 whether ‘the leading AI platform for regulated teams’ can anchor the enterprise campaign” can be researched.

Step 1. Split the claim into parts

Most market claims hide several claims inside one sentence. Separate them before gathering proof.

Each part needs a different standard. A customer count may come from an internal record. A causal claim needs a design that can separate the company's effect from other causes. A leadership claim needs a declared market universe and comparison rule. A testimonial can support experience. It cannot establish prevalence.

Step 2. Declare the market universe

A comparison set answers “how do we compare with these companies?” A category claim answers “how do we compare with the market?” Those are different research questions.

Before using a superlative, write:

• the category definition;
• the geography;
• the buyer segment;
• the time period;
• the inclusion and exclusion rule;
• the source used to identify eligible companies;
• the share of the declared universe actually covered.

Three hand-picked rivals can reveal contradictions and missing proof. They cannot establish market leadership unless those three exhaust the declared universe.

Step 3. Map the inspectable surfaces

Build a dated capture of the places a buyer can inspect without access to the founder's private reasoning.

Record the publisher, location, publication date, capture date, access conditions, source class, exact observation, and relation to the claim. A screenshot without that record becomes an anecdote the moment the page changes.

Step 4. Ask the buying group different questions

The commercial champion and procurement do not verify the same thing.

Treat the table as a starting hypothesis. Replace it with category evidence, client records, or direct research when available.

Step 5. Build the claim-to-evidence ledger

Use one row for every material claim part.

The ledger prevents one strong proof point from laundering the weak parts around it. It exposes duplicated evidence too. Ten articles that repeat the same vendor claim count as one source chain.

Step 6. Use five decision postures

Carry

The claim is stated at the level the evidence supports. The relevant surfaces agree. Counterevidence does not change the decision.

Revise

The core claim has support, but the comparison, population, time period, causal language, or certainty outruns the evidence. Narrow the sentence until it becomes defensible.

Verify

A private or unobserved question could change the decision. Name the method required: eligible interviews, a survey, technical testing, legal review, customer records, or another specialist. “More research” is not a method.

Pause

The evidence is incomplete and the cost of being wrong is material. Freeze the claim or the spend until the named gap is resolved.

Stop

The claim conflicts with strong evidence, depends on an undeclared comparison rule, or cannot be substantiated at the level the company plans to publish.

A valid process can end with no external work. If the claim holds and the right next action is internal, close the review.

Step 7. Write the Decision Record

The report is the durable evidence layer. The Decision Record is the completion test.

Write one page with:

• the original decision and date;
• the selected posture;
• the accepted evidence and its limits;
• the alternatives considered;
• the strongest dissent;
• the decision owner;
• the first action and due date;
• the person who must sign off with the owner, if any;
• the events that reopen the decision.

Useful reopen triggers include a new category entrant, a changed certification, a material customer loss, a major model-answer shift, an analyst reclassification, contradictory primary research, or a change in the planned market.

The research firewall

Keep these practices:

• visible scoring or decision logic;
• adaptive evidence routing;
• source and date traceability;
• calibrated confidence;
• clear routes to primary research or another specialist;
• a legitimate no-sale result;
• separation between the finding and any later implementation scope.

Reject these practices:

• pain agitation that pressures the founder into a diagnosis;
• identity language that makes disagreement feel irresponsible;
• yes-ladders that preordain the result;
• a calculated-looking score with no disclosed rule;
• questions designed to turn every path into a Petrichor sale;
• sales follow-up hidden inside research participation consent.

The instrument may qualify fit. It cannot manufacture an external buyer verdict from the founder's answers.

When public evidence is enough

Stay with public and permissioned evidence when the live decision turns on findability, consistency, freshness, contradiction, source quality, declared comparisons, or an observable approval artifact.

Add primary research when the decision turns on motive, prevalence, private objections, influence among roles, or how a message is interpreted by eligible buyers.

Route to technical, legal, security, financial, or regulatory specialists when the claim requires their sign-off. Route to broader company diagnosis when several claim failures reveal a system problem. Route to a workshop when the evidence is sufficient and the unresolved issue is leadership choice.

The 60-minute founder version

If you need a first pass today:

1. Write one exact claim and one decision date.
2. Split the claim into its factual, comparative, causal, predictive, and evaluative parts.
3. Name the buying group and one question per role.
4. Capture the claim across owned pages, search, one AI system, customer proof, and three named alternatives.
5. Mark every observation as support, contradiction, absence, or unknown.
6. Write the strongest case against your preferred answer.
7. Choose carry, revise, verify, pause, or stop.
8. Assign one owner, one first action, and one reopen trigger.

This pass can expose an obvious contradiction. It cannot stand in for interviews, a market-wide comparison, technical assurance, or a causal study.

The one-line operating principle

Bring one claim and one decision date. Test the evidence a named buyer can inspect. Mark everything else as a question.

Petrichor Projects | Category Gravity | petrichorgrowth.com/research