AI engines recommend people based on verifiable evidence, not self-description. A proof portfolio is the organized inventory of that evidence across three shelves: what you own, what others say about you, and what machines can parse. Audit it once, fill the gaps in priority order, and every future mention lands on a foundation that compounds instead of scattering.
Machines cannot take your word for anything. Before an engine names you, it looks for evidence: things other people and other systems say about you that it can read, cross-check and attribute. That collection is your proof portfolio, and most experts have never once inventoried theirs.
Why machines need evidence, not claims
Humans buy stories. Machines weigh corroboration. When an engine assembles an answer to "who should I trust with this," it is choosing between names based on what it can retrieve and verify right now, plus what it already associates with each name. Your homepage says you are an expert. So does every competitor's homepage. Self-description is table stakes, which means it breaks no ties. What breaks ties is evidence that exists outside your control: a byline on a publication you do not own, a review you could not have written, a conference agenda that lists your talk, a credential confirmable on the issuing body's own site.
This is not a new idea dressed up for AI. Google has been explicit for years that its systems look for signals of experience, expertise, authoritativeness and trustworthiness, and its own documentation on helpful content describes exactly the kind of first-hand, verifiable expertise it wants to surface. AI answer engines inherited that instinct and sharpened it, because a chatbot that recommends the wrong professional embarrasses its maker in a way a page of ten blue links never did. The engine is cautious by design. Your job is to make trusting you the cautious choice.
The payoff for getting this right is not vanity. Visitors referred by AI assistants convert dramatically better than search traffic: Conductor's 2026 benchmarks put ChatGPT referral conversion at 14.2 to 15.9 percent against roughly 1.76 percent for Google organic, as compiled in SEO Sherpa's AI search statistics roundup. A person who arrives because a machine vouched for you is most of the way to a decision. Proof is what earns that vouching.
What is a proof portfolio?
A proof portfolio is the complete, organized set of credibility assets attached to your name: every artifact a machine or a diligent stranger could find that supports the claim that you are who you say you are and good at what you say you do. Think of it the way a fund manager thinks of a portfolio. It has holdings of different types, some appreciate and some decay, it needs periodic rebalancing, and its value comes from the mix rather than any single position.
Every asset in the portfolio sits on one of three shelves:
- Owned proof: evidence you control. Your site, your bio, your published body of work under your own name.
- Earned proof: evidence others control. Press quotes, reviews, podcast appearances, event listings, awards. This is the heavy shelf, the one machines weight hardest.
- Structured proof: evidence formatted for machines. Person schema, consistent identity facts across profiles, markup that ties every scattered mention back to one entity.
Owned proof makes you legible. Earned proof makes you believable. Structured proof makes the first two attributable. A portfolio missing any shelf leaks value from the other two.
The credibility asset inventory: the full checklist
Set aside two hours and walk this list. For each item, record three things: does it exist, is the name and bio on it consistent with everywhere else, and could a stranger find it from your name alone.
Shelf one: owned proof
- An about page that states your full name, role, specialty, track record and location in plain sentences, not slogans.
- A services or work page that says precisely what you do and for whom.
- A bylined body of work: articles, essays or guides published under your exact name, on your site or elsewhere, deep enough to demonstrate a position rather than summarize common knowledge.
- One canonical headshot used everywhere, so image models and humans connect your appearances.
- One canonical short bio, 40 to 80 words, reused verbatim across platforms.
- A contact path that works, because proof of reachability is itself a trust signal.
- Case narratives: what you did, for what kind of client, with what outcome, told without invented numbers.
Shelf two: earned proof
- Press mentions and journalist quotes, however modest the outlet, as long as it is real and independent.
- Podcast episodes where you are the named guest, with show notes that spell your name correctly and link your site.
- Conference and event listings that show you spoke, on pages you do not control.
- Client reviews on third-party platforms relevant to your field, since machines discount testimonials that live only on your own site.
- Guest bylines on respected industry publications.
- Awards, rankings and "top people in X" lists compiled by someone other than you.
- Credentials that resolve: degrees, licenses and certifications a machine can confirm on the issuer's site.
- A published book with an ISBN, if you have one, because it creates durable catalog records.
- Peer citations: other practitioners referencing your work or naming you in their own content.
Shelf three: structured proof
- Person schema markup on your about page declaring your name, role, affiliations and profiles.
- A sameAs list connecting your site to your LinkedIn, X, GitHub, Crunchbase or other authoritative profiles.
- Identical core facts, name, title, specialty, location, across every profile you hold.
- Profiles actually completed, not half-filled skeletons from 2019.
- Review and rating markup where platforms support it.
- A Wikidata entry, only where genuinely merited, never faked.
How do machines weigh different credibility signals?
Not all proof is equal, and pretending otherwise wastes effort. Here is how the asset classes compare on the dimensions that matter.
| Asset class | What the machine reads | Trust weight | Decay speed |
|---|---|---|---|
| Self-published claims | Your bio, your site copy | Low alone, but required as the anchor | Slow, you control it |
| Bylined earned media | Independent pages naming you as author or source | High | Slow, archives persist |
| Third-party reviews | Ratings and text on platforms you cannot edit | High for service queries | Medium, recency counts |
| Event and directory listings | Agendas, member rolls, speaker rosters | Medium, strong in aggregate | Fast, pages get retired |
| Verifiable credentials | Issuer databases and registries | High in licensed fields | Slow |
| Structured markup | Schema connecting all of the above to one entity | A multiplier, not a source | Slow, but audit yearly |
Two readings of that table matter. First, earned proof dominates, which is why the loudest self-promoter loses to the quietly corroborated professional. Second, structured markup earns the strange label of multiplier: it creates no credibility by itself, but it makes sure the credibility you earn elsewhere actually accrues to you instead of dissolving into ambiguity about which "J. Morgan" gave that talk.
Run the gap analysis
With the inventory done, the gaps announce themselves, and they tend to follow three patterns. The first is all owned, no earned: a beautiful website, a rich blog, and not one independent source that confirms any of it. This is the most common profile among competent people who dislike self-promotion, and it caps how far any engine will go in recommending them. The fix is a deliberate earned-media push, and the fastest routes are guesting and reviews, which we cover in the podcast guesting playbook and the piece on reviews machines actually read.
The second pattern is earned but fragmented: real press, real talks, real reviews, scattered across three name variants, two old employers and an outdated bio. The evidence exists but does not compound, because the machine cannot confidently assemble it into one person. The fix is consolidation: standardize the name, the bio and the headshot, then add the structured shelf so every asset points home.
The third is no structured layer at all, which is nearly universal. If you have never added Person schema or a sameAs list, an afternoon of work upgrades everything you have already earned.
Score yourself honestly, then sequence the fixes: consolidation first, structure second, new earned proof third. That ordering matters because new proof landing on a fragmented identity is money left on the table. If you want the sequencing laid out week by week, the first 90 days plan does exactly that, and the trust ladder explains the order in which strangers, and machines, come to believe you.
Proof decays, so maintain the portfolio
A proof portfolio is not a monument. Semrush's AI Visibility Index found that 40 to 60 percent of the sources cited in AI answers rotate month over month, as reported in Similarweb's generative AI statistics. Engines keep re-sampling the evidence, which means stale proof quietly falls out of the answer set while fresh proof enters it. The professionals who stay named are the ones whose portfolios keep producing new corroboration.
Put a quarterly review on the calendar. Check that event pages listing you still resolve, and archive or replace the ones that vanished. Retire references to roles you left. Add the quarter's new assets to the inventory. Confirm your canonical bio still matches reality, because a portfolio that contradicts itself is worse than a thin one; conflicting facts force the machine to hedge, and hedging machines do not name people.
Where to start this week
Do the two-hour inventory first, before creating anything new. Most experts discover they are richer than they thought and messier than they feared: plenty of raw proof, badly filed. Consolidate, structure, then go earn the next asset. If you would rather have the audit done for you, with the gaps priced against the queries that actually pay in your field, that is precisely what our services engagements begin with.
Questions
What counts as a credibility asset? +
How many proof assets do I need before AI trusts me? +
Which credibility signal is the strongest? +
See what AI says about you today.
Start with a reading. We show you the words the engines return about your name, then map the fastest signal to move.
Get named →