The method, in sequence. Fixed on purpose.

What you supply, and what we deliberately do not ask

Before anything starts: brand guidelines if they exist, and an honest answer if they do not. Production files for anything printed or fabricated. Access to analytics and search console, plus a list of everyone who has published in your name. Contact for whoever answers the phone. Recent invoices, quotes and standard correspondence. We ask what the business does, and who it serves.

We do not ask what you think your brand problems are. That question is the standard opener, and it is also why most audits confirm what the client already suspected. The answer becomes the hypothesis, and the rest of the work turns into a search for supporting evidence. The diagnosis waits until there is something to diagnose from.

Anything requested and not supplied goes into the unexamined section automatically. A client who declines to supply something has made a decision, and the report reflects it rather than arguing about it.

The sequence

The two cases

Every claim gets both, built to the same evidence standard. The case against is constructed deliberately, and it has to be one you would find uncomfortable, because an argument designed to lose proves nothing. Disconfirming material is searched for first, since supporting material will always be found. The case for establishes not only that a claim holds but why, which is what lets you defend the position later when we are not in the room.

A claim survives only when the attack on it fails. Where both cases are strong, the claim is inconsistently expressed, and that is itself the finding. The attack is recorded either way, including on claims that survived. That shows the work and hands you the language.

"We argue against our own recommendations too. Before anything is proposed, the strongest case for doing nothing gets stated."

Some identities are working, and the honest answer is to leave them alone, which is a conclusion a firm selling brand work is structurally disinclined to reach.

What a finding is

Never an impression, because every entry names what was examined and where, so a reader can go and look. Three columns that never merge: what the brand claims, what the brand shows, and what the world says. Most of the value sits in the gaps between them, and collapsing them into one narrative hides exactly what the audit exists to find.

Observation, interpretation and recommendation stay separate. Six color variants appear across live surfaces is the first. The palette expanded without governance is the second. Consolidate to three and encode them is the third. Every finding carries an evidence grade: a direct artifact, a third-party source, an inference from pattern, or something you told us. A reader then sees instantly which would survive a challenge.

What the report contains

What could not be examined is stated rather than omitted: sources unavailable, surfaces you could not supply, periods beyond reach. An audit implying completeness it does not have has the same defect as a claim with no source behind it. What must not be touched is named explicitly, because an audit recommending change without protecting what works invites you to modernise away the thing earning your living.

True claims you are not making get their own section, meaning things that would survive scrutiny and appear nowhere you publish. Every recommendation carries an effort estimate: something your own team can do, something needing a supplier, or something amounting to a project, while anything in the third band states what it displaces. Disagreements are recorded as disagreements, in your words, along with whether new evidence was supplied and whether the finding stands, is revised or is withdrawn. The report has an expiry date, agreed at the start.

Findings outside the brief

A thorough sweep surfaces things that are not brand problems: exposed access that should have been revoked, a former supplier still acting in your name, or something concerning an individual rather than the business. We report it and do not investigate it, since pulling that thread is a different engagement nobody commissioned. Anything involving a named person is handled with more care than anything involving the company, and some findings get raised immediately rather than saved for a deliverable. Live access that should not exist is the clearest example.

The second audit

A single audit is a snapshot. The value compounds in the comparison. The same instruments are re-run, not a fresh look, because a different method produces a different picture and tells you nothing about change. The census counts are the headline, because three numbers that moved or did not are the most legible evidence of progress anyone will show you.

Acted, ignored and reversed get reported separately. The third category is the most instructive and the one nobody reports. A recommendation that did not move needs a reason rather than a repetition, because frequently the reason is that it was wrong, too expensive, or displaced something more important. Anything from the prior audit that turned out to be wrong is withdrawn explicitly and named.

Where the audit stops, and where it connects

This examines what a business claims, what it shows, what the world says about it, and whether those agree. Where a finding touches something beyond what a brand review can properly reach, we say what we found, say it is outside scope, and stop. The audit connects into every other discipline of brand integrity: competitive position, claim verification, digital footprint, AI brand perception, visual continuity, brand voice, internal alignment and duty of care.