An audit, not an opinion.

How it is built

Facts first, judgement second

The instrument is assembled in one direction only: your buyers, then the record, then the verdicts, then a number that code computes. Nothing further down this page is claimed without something behind it.

The room your ICP implies

• Owner

  Gen Z

  Relevance 0.9

• Finance

  Millennial

  Relevance 0.6

• Owner

  Millennial

  Relevance 0.9

• Ops

  Gen Z

  Relevance 0.5

The decision maker carries the heaviest weight. The committee around them is derived, not invented.

1. Built
2. Recorded
3. Judged
4. Scored

1. 01 Built

   It starts with your buyers, not a persona brief

   You supply one line about who you sell to. That ICP becomes the decision maker, and the roles who influence or veto around them are derived from it — profiled across the generations that actually hold those jobs. Each buyer carries a relevance weight, so the customer outranks the committee when the numbers are added up.

2. 02 Recorded

   The facts are fixed before anyone judges them

   A full URL map of the site makes absence provable: if no pricing page exists in the inventory, the site has no public pricing, we did not simply miss it. Pages are read in priority order and one pass records 26 objective signals, each with a verbatim quote and the page it came from. No verdicts yet.

3. 03 Judged

   Every buyer judges the same record

   Each persona reads the fixed signals and returns blocker, friction, neutral, or strength with a severity. Evidence is copied from the verified signal, so every quote in the dashboard came off your page. Two buyers never disagree about what the site says, only about whether it works for them.

4. 04 Scored

   The scorer is code, not a second opinion

   The model never returns a score. It returns verdicts and severities, and a deterministic function turns those into risk density, the blocker floor, and the 0–100 readiness number. Change nothing and the score does not drift.

The buying room

Every buyer carries what they were built from

A persona here is not a character with a backstory and a favourite coffee order. It is a role your ICP implies, a cohort that holds that role, a relevance weight, and the published research behind the habit being modelled. The card on each portrait is the same record the run stores.

Owner

Relevance 0.9

Gen Z

Grew up self-serve. Reads a sales call as friction, or as a sign something is hidden.

Behind this buyer

• Your ICP lineNamed as the buyer
• TrustRadius41% start at a review site
• Recorded signalsResearch and channel habit

Evidence base

What every check is built from

The signals we look for are not a matter of taste. Each one stands on a buyer survey or a controlled experiment, and the same literature sets the limits further down the page. None of these institutions sponsor, endorse, or review this product.

Signals are extracted once and reused, so cost grows with the number of buyers rather than buyers × pages. The four stages are the same on every run, which is what makes two audits comparable.

1. 01 Frame→
2. 02 Establish→
3. 03 Elicit→
4. 04 Score

Readiness is 100 minus the relevance-weighted average of each buyer’s risk density. Density is that buyer’s remaining risk as a share of the worst it could have been, which is what makes two runs comparable. Everything below happens in code after the model has returned its verdicts.

1. 01

   Layer 1. Four verdicts, one buyer at a time

   A blocker ends this buyer’s evaluation. Friction costs them effort or trust but they can still proceed. A strength actively helps them decide. Neutral does not move the needle and is omitted rather than padded in.

2. 02

   Layer 2. Severity is anchored, not vibes

   5 ends the evaluation on the spot. 4 means they probably will not proceed unless something rescues it. 3 is a real hit they carry into the decision. 2 is absorbable friction. 1 is an irritation. If everything comes back a 4 or 5, the judge has stopped discriminating.

3. 03

   Layer 3. Density, not a body count

   The judge returns a capped number of findings per buyer whatever the material is like, so raw totals say little. Density asks a different question: of everything this buyer noticed, how much of it was disqualifying?

4. 04

   Layer 4. A blocker puts a floor under the cell

   Strengths cancel up to half of a cell’s friction. They cannot un-block anyone. The worst blocker floors that buyer’s risk at its severity, because a buyer only leaves once and praise elsewhere does not undo a dealbreaker.

Fixed before the run

The model never returns the score

It returns a verdict, a severity, and the passage that caused it. The weights, the strength cap, and the blocker floor are ordinary code with tests around them. Run the same findings twice and you get the same number.

Same buyer, same twenty findings. One dealbreaker takes the cell from 5 to 100, because a buyer only leaves once and no amount of praise brings them back.

The readiness number is designed to move. Material that genuinely serves its buyers should be able to clear 90; material that loses the room should stay low. An unfinished run — buyers with no findings yet — is refused a score rather than shown as a flattering 100.

26

objective signals recorded on every website run

4

generation cohorts a role can be held by

0–100

absolute readiness, comparable between sites

These are real runs on the current instrument, not illustrations. Both numbers were recomputed from the findings each run stored. The instrument did not change between them.

An audit is a hypothesis about your buying room, backed by what your material actually says. It is not a substitute for talking to the people who would buy the thing. The method is reproducible; the room is still a model, and the literature is clear about where that model is weakest.

Built on BuyerLens methodology: published evidence first, a scorer written before the run, and results that are allowed to come back bad. Run it on your own material.