I recently finished a major rebrand of my own business.
The website changed. The positioning changed. The language changed. I had gotten very clear on the category I wanted to occupy: Strategic Advisor — AI, Digital Authority & Real Estate Innovation.
Then I did something that turned out to be more useful than another round of copywriting.
I asked an outside AI system to research me cold.
I did not tell it how I wanted to be positioned. I asked it to look at the public evidence and tell me who it thought I was.
Its answer was basically: real estate coach.
My first instinct could have been: the AI got it wrong.
But once I traced the answer back to the sources it was reading, I realized the more useful conclusion was the opposite.
The problem wasn’t the AI. It was my evidence.
Your brand is the pattern the internet can infer
The AI was synthesizing the evidence I had distributed over years: old pages, old bios, old footer language, historic positioning, current pages, videos, articles, and third-party references.
I had changed my intended identity faster than I had changed the public evidence around it.
That led to a rule I think is going to matter a lot more in the next few years:
Fix the evidence, not the AI.
Your brand is no longer only the sentence you put at the top of your website. It is the conclusion a person — or a system — can reasonably reach after looking at the evidence you have made available.
If you say you are a luxury agent, relocation specialist, listing expert, neighborhood authority, strategic advisor, innovative brokerage, or technology leader, your headline cannot carry the whole burden. Your reviews, case studies, videos, articles, profiles, proof, and third-party references all teach the market what to believe.
What my audit actually found
The useful part wasn’t that an AI called me a real estate coach. The useful part was tracing that conclusion back to the evidence.
The audit found two versions of me living online at the same time.
The current version said Strategic Advisor — AI, Digital Authority & Real Estate Innovation.
The legacy version still taught the internet “Coach Patrick Ferry” through old pages, article bios, footer language, historic positioning, and other retrievable surfaces.
That changed the job completely.
I did not need another prettier homepage. I needed to make the evidence more coherent.
Some evidence needed to be corrected because it was outdated. Some needed to stay because it accurately described my history. The goal was not to erase more than two decades of work. The goal was to stop letting the past masquerade as the present.
My working rule became:
- Do not erase your history.
- Do not argue with the machine.
- Make the evidence truthful, current, and easy to interpret.
That is evidence architecture.
The framework: Experience → Evidence → Distribution → Recommendation
Experience is the real work you have done and the expertise you have earned.
Evidence is the proof that makes that expertise visible: articles, case studies, pages, videos, reviews, frameworks, interviews, credentials, results, and examples.
Distribution is where that evidence lives and how consistently it is expressed across the web.
Recommendation is the downstream result: a person, a search engine, YouTube, or an AI system has enough credible evidence to understand when you are relevant.
The common mistake is trying to jump directly from experience to recommendation.
“I am great at this. Why doesn’t the market know?”
Because expertise that exists only in your head is not evidence. Evidence that exists in one hidden corner of the internet is poorly distributed. And contradictory distribution creates ambiguous recommendations.
The AI Brand Alignment Test
Use this after a rebrand, positioning shift, website rebuild, new offer direction, or whenever you suspect the internet still describes an older version of you.
Important: do not begin the cold test by telling the AI how you want to be positioned. If you seed the answer, you are no longer testing retrieval; you are testing compliance.
Step 1 — Write the intended identity first
Before you look at the web, answer these in plain language:
- My current professional category is: ________
- The three to five subjects I most want to be known for are: ________
- The people or organizations I most want to serve are: ________
- The work I most want the market to understand I do is: ________
- The old category, offer, or identity I do not want mistaken for my current primary identity is: ________
- My strongest proof that supports the new identity is: ________
Step 2 — Run the cold external identity test
Use a web-enabled AI system with little or no prior context about you. If possible, run the test in more than one system.
Research my public digital footprint using only publicly available web sources. Do not rely on anything I tell you about how I want to be positioned. Based on the evidence you find, tell me: 1) Who do you think I am? 2) What professional category would you put me in? 3) What am I best known for? 4) Who do I appear to serve? 5) What products, services, or work do I appear to sell? 6) What three to five public sources most strongly shaped your conclusion? 7) Which evidence appears current and which appears historical? 8) Where is my public identity contradictory, ambiguous, or outdated? Be specific and cite the evidence.
Save the answer exactly as returned. Do not correct it yet.
Step 3 — Diagnose the gap
Compare your intended identity with the cold-test answer. Sort the evidence into five buckets:
- ALIGNED — the public evidence supports the intended identity.
- CONTRADICTORY — public evidence teaches a competing identity.
- MISSING — important parts of the intended identity have little or no visible evidence.
- AMBIGUOUS — the evidence could reasonably lead to multiple interpretations.
- HISTORICAL — accurate past information is being mistaken for the present.
For each item, identify the specific page, profile, bio, article, video, review pattern, or third-party source creating the conclusion.
Step 4 — Build the public evidence graph
Create a repair table with these columns:
- Source / page
- What it currently teaches about you
- Current / historical / contradictory / missing context
- How influential it appears to be
- Disposition: KEEP / UPDATE / REDIRECT / RETIRE / CREATE SUPPORTING EVIDENCE
Look beyond the homepage. Common evidence sources include About pages, author bios, old articles, old sales pages, footer language, navigation labels, YouTube descriptions, podcast bios, speaker pages, LinkedIn, Google Business Profile, press mentions, schema, directories, and high-ranking historical pages.
Step 5 — Prioritize the repair queue
Do not prioritize by visual annoyance. Prioritize by expected identity impact.
- High-authority current pages
- High-ranking legacy identity pages
- Shared/global elements such as headers and footers
- Author bios, metadata, and schema
- High-traffic or frequently cited articles
- Major third-party profiles
- Lower-value long-tail cleanup
For every priority item, ask: what is the smallest change that removes the contradiction while preserving accurate history, useful URLs, backlinks, and current proof?
Step 6 — Turn the decisions into a Brand OS
Do not leave the new identity scattered across a dozen chats.
Create one canonical Brand Foundation or Brand OS that records at minimum:
- Current category
- Positioning statement
- Primary audiences
- Topics to own
- Terminology to use and retire
- Current offers or engagement types
- How to describe historical work
- Important organizational affiliations
- Core proof
- The rule for resolving conflicts with older material
Then require future AI work to retrieve that source before doing material brand, positioning, messaging, offer, or identity work.
Step 7 — Propagate, do not merely rewrite the homepage
The homepage matters, but it is only one node in the evidence graph.
For every repair, verify: does this source now make the present identity clear? Does it preserve accurate history without presenting the past as the current category? Does the page connect to the right current destination? Do metadata or author information contradict the visible copy?
Do not claim completion until the source itself has been verified after the edit.
Step 8 — Re-run the cold test
After the meaningful evidence has been corrected and external systems have had time to recrawl it, run the same cold prompt again without showing the AI your intended identity.
The success condition is not perfect wording. The success condition is that an informed outside system can correctly infer your current category, primary subjects, audience, and work without being coached into the answer.
Step 9 — Treat disagreement as evidence
If different AI systems still describe you differently, do not immediately pick the answer you like best. Ask what source set each system used. The disagreement often reveals uneven distribution, stale indexing, conflicting third-party profiles, or a category that is not yet supported by enough evidence.
Can you actually repair the evidence?
This is where brand strategy becomes operational.
If the audit tells you the web is misunderstanding you, can you update the positioning, add a guide, publish a case study, repair an author bio, add proof, clarify an old page, or create the missing evidence without rebuilding your entire digital presence?
Your website and content systems should make your expertise easier to express and update, not harder.
Fast acceptance checklist
- A cold AI system identifies the correct current professional category.
- It names the subjects you actually want to be known for.
- It can distinguish current work from historical work.
- Your major owned pages teach the same current identity.
- Formal organizational titles are not being mistaken for your independent category.
- High-authority legacy pages are contextualized rather than blindly deleted.
- Author bios, footer language, navigation, metadata, and schema do not contradict the visible brand.
- Your strongest claims are supported by visible proof.
- Your current work has enough distributed evidence that an outside system can discover it without being told.
The principle
Your brand is not the sentence you prefer. It is the pattern a reasonable person or system can infer from the evidence you have made available.
The job of this audit is not to manipulate an AI answer. It is to make the public evidence more truthful, coherent, current, and useful.
