You are on a call with a prospect who found you through a mutual connection. It is going fine. Then they say the thing: “Before we go further, I want to run this past a couple of people.” You know what happens next, because you have done it yourself. They open a browser. They search your name. They open your LinkedIn, your company site, whatever a search engine puts in front of them, and maybe they ask an AI assistant what it knows about you. Ninety seconds later they have formed an opinion you had no part in shaping.
That ninety seconds is the whole game. Everything you do to build authority as a founder is an attempt to control what a stranger finds in that window and what a machine says when somebody asks about you.
Most founders respond to this by making noise. More posting, more podcasts, more conference badges. Some of it helps. Most of it evaporates, because noise is not the thing being checked. What gets checked is proof, and proof has a structure.
Visibility and authority are different products
Visibility means people have heard of you. Authority means people believe you when you make a claim in your domain. These are not the same asset and they do not convert into each other on their own.
There are founders with fifty thousand followers who cannot get a serious journalist to call them back, and founders with two thousand followers who get quoted in trade press every quarter. The difference is not audience size. It is whether the person on the other side can verify something specific about you inside a minute.
Visibility is a volume metric. Authority is a verification metric. Volume can be bought, faked, farmed, and inflated. Verification cannot, which is why it holds value. Every hour you spend on visibility that does not produce a verifiable artifact is an hour spent on something with a two-day shelf life.
This distinction has become sharper as AI assistants have moved into the research step. When a model answers a question about who is credible in a category, it does not count your followers. It assembles an answer from public text that corroborates itself across independent sources. Volume is invisible to that process. Verification is the entire input.
There is a diagnostic that settles the question fast. Imagine your best prospect asks a mutual contact about you, and that contact has never worked with you. What can they say? If the answer is “I see them post a lot,” you have visibility. If the answer is “they wrote the piece on procurement cycles that everyone in our category passed around,” you have authority. The second answer requires an artifact that exists outside your own promotion of it. That is the difference in one sentence, and it explains why so much founder marketing produces warmth without producing conviction.
So the working definition I use with founders is simple. Authority is the set of things a stranger can confirm about you without asking you. Everything else is marketing.
The Proof Stack is what your buyer checks
The Proof Stack is six assets, in order, each one making the next one possible. I named it because founders keep trying to build authority as a founder from the top of the stack down, before the bottom exists, and it fails the same way every time.
Layer one is a position. One defensible claim about how your corner of the world works, narrow enough to be wrong.
Layer two is a body of work. The artifacts that demonstrate the position rather than assert it: writing, product decisions, talks, teardowns, data you gathered yourself.
Layer three is third-party corroboration. Coverage, citations, quotes, and mentions on properties you do not own.
Layer four is named peers. Who vouches for you by name, who you appear alongside, whose audience has seen you.
Layer five is the retrievable record. Your position, work, and corroboration rendered in text a search engine and an AI system can find, parse, and repeat without ambiguity.
Layer six is the referral loop. Customers and peers who repeat your claim in rooms you will never enter.
Notice what is missing from that list. Follower counts. Logos on a website. A podcast you host. Those are outputs of the stack or accessories to it. They are not load-bearing.
The stack is sequential for a reason. A journalist will not cover you without a body of work. A peer will not vouch for a position you have not stated. An AI system cannot retrieve a record you never wrote down. Skip a layer and the layers above it wobble.
A position and a body of work come first
Founders hate layer one because a position, by definition, excludes people. That is the point. A claim that offends nobody informs nobody.
A position is not a value statement. “We believe in customer obsession” is not a position, it is wallpaper. A position sounds like an argument you could lose: most sales teams are compensating for a pricing problem they refuse to name. Enterprise procurement cycles are long because vendors write proposals for the wrong reader. Your CAC problem is a retention problem wearing a costume. Each of those has an opposing view held by intelligent people, which is what makes it worth stating.
Then layer two makes the position real. This is where most founders discover whether they had an insight or a slogan.
Look at what Jason Fried and David Heinemeier Hansson built at 37signals. The position was clear for two decades: software companies are overbuilt, overfunded, and overworked, and you can run a profitable business without any of that. They did not assert it in a keynote and move on. Heinemeier Hansson released Ruby on Rails in 2004 and it became infrastructure for a generation of web applications. They wrote Getting Real in 2006, Rework in 2010, Remote in 2013. They ran a company blog for years where they argued the position in public, over and over, against people who disagreed. The books, the framework, and the company itself were all the same argument in different formats.
That is a body of work. Not content. Artifacts that would have to be dismantled to disprove the claim.
Paul Graham did the same thing with a different medium. He has been publishing essays at his own domain since the early 2000s, arguing a consistent set of positions about startups, and those essays became the reason Y Combinator could recruit founders before it had a track record. The writing preceded the credential. It often does.
Your version does not need to be a bestselling book. It needs to be a stack of public artifacts that make your claim harder to dismiss each time you add one.
Third-party corroboration is the layer founders skip
Here is the asymmetry nobody warns you about. Everything you say about yourself is discounted. Everything a credible outsider says about you is not.
You can publish a hundred posts explaining your expertise. One paragraph in a trade publication that describes you the same way does more work, because the publication had no obligation to say it. That is what corroboration means: a claim that survived contact with someone who could have refused to make it.
This layer is where founders stall, and the reasons are predictable. Pitching feels like begging. Rejection from editors stings. The payoff is slow and hard to attribute. So they go back to posting, which is under their control and produces a dopamine response the same day.
The mechanics of getting corroborated are not mysterious. You need a position sharp enough to be quotable, a body of work that makes you a defensible choice for an editor, and a reason for the story to exist this week rather than any other week. Journalists are not looking for impressive people. They are looking for people who make a specific story easier to write. Those are different searches.
Start below the top. A quote in a trade publication your buyers read beats a mention in a national outlet they do not. Industry podcasts, association newsletters, expert-source platforms, guest columns in category publications: these are lower-friction, faster to land, and they compound. Five specific citations in your category outperform one general-audience feature nobody in your market saw.
Layer four follows from layer three faster than founders expect. Once you have been quoted somewhere credible, other people in your category start treating you as a known quantity, and the invitations change character. Podcast hosts book you because a booking is a bet and prior coverage lowers the risk. Panel organizers want a name that has been vetted by somebody else. Peers link to you because linking to a covered source costs them nothing in credibility. None of this is a favor. It is risk transfer, and third-party corroboration is what moves the risk off the person deciding whether to associate with you.
There is a second reason this layer matters more than it did five years ago. AI answer engines weigh corroboration across independent sources when deciding what to say about a company or a person. Your own site is one source. Six unrelated publications describing you the same way is a pattern. Patterns are what get repeated back to whoever asked.
The retrievable record decides what machines say about you
Layers one through four can all exist and still fail if they are not written down in a form that can be found.
The retrievable record is the version of you that exists in text a machine can parse: your site’s about page, your bio as it appears across the properties that covered you, your LinkedIn profile, the structured markup that tells a crawler this entity is a person, this is their role, this is the organization, these are the sources.
Consistency is the variable that matters most, and it is the one founders neglect. If your name appears three different ways, if your title on one site does not match another, if the company description shifts every time somebody rewrites it, you have handed a retrieval system a set of conflicting facts. Conflicting facts produce hedged answers or no answer at all.
Fix this in an afternoon. Write one canonical paragraph about yourself: name, role, organization, the position you are known for, two verifiable credentials, one link. Use it everywhere without editing it for flavor. Put the same facts in structured markup on your site. Make sure the pages that describe you are indexable, which sounds obvious and gets violated by founders whose bio lives inside a JavaScript modal that no crawler will ever open.
The goal is not to trick anything. The goal is to remove ambiguity so that when a system assembles an answer about your category, you are a clean entity it can name rather than a fuzzy set of half-matching strings.
Order beats effort
The people who build authority as a founder faster than their peers are not working harder. They are working in sequence.
The failure pattern is always the same shape. A founder decides they need press, hires a publicist, and discovers there is nothing to pitch because layers one and two are empty. Or they build an enormous following on one platform, then find that none of it transfers when they need a journalist, a partner, or an AI system to take them seriously. Effort applied out of order produces motion without accumulation.
Six months of correct sequencing beats two years of scattered activity. Spend the first month on the position, and be willing to throw out three versions before one survives an argument. Spend months two through four on the body of work, publishing on a schedule you can hold, in public, under your own name. Start pitching in month four, when you have artifacts to point at. Fix the retrievable record in parallel, because it is a weekend of work with a permanent payoff. The referral loop takes care of itself once the other five layers exist, which is the only part of this you cannot force.
One warning about delegation, since this is where the money tends to go. You can hire help for layers two through five. You cannot hire out layer one. A writer can turn your thinking into publishable prose. A publicist can turn your artifacts into coverage. Neither can invent a position you do not hold, and every attempt to outsource that step produces the same result: content that is competent, generic, and attributable to nobody. If you are going to spend money here, spend it on people who extract and sharpen what you already believe, and be suspicious of anyone who offers to supply the belief.
You will know it is working when the pattern reverses. Instead of you explaining who you are, people arrive already knowing. The prospect on the call does their ninety seconds of research and comes back warmer instead of cooler. A journalist emails you rather than the other way around. Somebody quotes your position back to you in a meeting and does not know it came from you.
So the question worth sitting with is not how to get more visible. It is this: if a stranger spent ninety seconds researching you tomorrow, which layer of the stack would they fail to find, and what are you going to do about it this month?