The Homepage Is No Longer the Front Door: Leah Nurik on AI Visibility, GEO, and the Future of Brand Discovery

Full analysis

Your draft

The old free traffic pipe from Google to your website is closing, and nobody has yet agreed on how you get discovered instead. Leah Nurik, CEO of Brandi AI, sat down with Signal & Noise hosts Rio Longacre and Brett House to name the new discipline (GEO, for Generative Engine Optimization) and sell you the fix. Some of what she says is real. Some is her own book. Sorting the two is the job.

What's actually being decided for a business reader: do you reallocate marketing money and attention this quarter toward getting cited inside AI answers, or do you wait for the numbers to firm up? That reallocation is easy to undo. You can shift PR and content spend back if it doesn't pay. Nothing here forces a deadline, so there's no reason to panic-hire or panic-buy a GEO platform. But the traffic decline is real enough that ignoring it entirely is its own bet.

The Skeptic. Almost every eye-popping number here comes from the person selling the cure. The "6x conversion" claim is Brandi AI's own demo funnel, one company, self-selected, with a direct financial interest in you believing it. The "40% of Google traffic gone, now only 21% of web traffic" stat Longacre ran with is sourced to something called "People Link," which nobody recognizes, and it contradicts every credible measurement firm. Nurik's "90% of citations are earned media" has no source either. What survives the discount? The referral traffic decline itself: multiple client reports, ~30% average, is directionally believable because it shows up across many independent sites. The rest is a pitch deck with footnotes missing.

The Researcher. Strip the marketing and one claim is genuinely useful: AI answer engines source authority from a different part of the web than Google's blue links. Nurik says 30% of AI Overview citations don't appear on page one of traditional search. Unverified, but directionally sound, because these systems weight independent, corroborating sources (Reddit, G2, trade press) rather than whoever won the SEO game. That changes what you optimize for. The other real signal Longacre and House flagged: Google search revenue grew ~17% in Q2 2026 while click-throughs to publishers fell. Those two facts coexisting is the important pattern, and it's the opposite of what most people assume.

The Compute Pragmatist. House flagged the paradox and half-answered it. Here's the full mechanism, because it's counterintuitive. Google's revenue holds while referral traffic to publishers collapses because those outbound clicks to publishers never paid Google a cent. That traffic was a cost Google absorbed, not a revenue line. When AI Overviews answer the query on-page, the user stays, and the ads move onto the answer surface itself. Google keeps the auction, drops the delivery cost, and raises minimum spend thresholds on top. Publishers eat the entire loss. This is why Google can cannibalize its own link model and still print 17% growth.

The Builder. What would I actually do Tuesday morning with this? Not much that's new, and that's the good news. GEO is not a tool you buy or a specialist you hire. It's your existing PR, content, and SEO people getting credible third parties to say true things about you, because that's what the models cite. The one concrete move: go run your own brand and your top three competitors through ChatGPT, Perplexity, and Google AI Mode this week. Read what they say about you and where they source it. That's your baseline. Do it monthly. The gaming tactics (fake reviews, astroturfing Reddit) will get you penalized the way Google's Panda update wiped out content farms, so don't. Nurik made that same point, and she's right to.

The Enterprise Buyer. The unsolved problem is measurement. Nobody can yet show you, with audited numbers, that money spent on AI visibility drove revenue. The conversion claims are single-vendor anecdotes. Before signing a GEO platform contract, ask the vendor: what's your source for citation-share data, can I see it validated against a third party, and can you attribute a closed deal to an AI-referred visitor? Most can't, yet. The category is real; the accountability layer isn't built. That gap is exactly where an early buyer overpays.

Where the council splits

Two real disagreements. First, the Researcher and the Skeptic part ways on whether GEO is a discipline or a rebrand. The Researcher sees a genuine shift in where authority comes from. The Skeptic sees Nurik putting a new acronym on "do good PR and get indexed." Both are right, which is the point: the underlying change is real, the packaging is opportunistic.

Second, the Builder and the Enterprise Buyer split on urgency. The Builder says the work is free and cross-functional, so just start. The Enterprise Buyer says the moment you spend real money, you're buying on numbers nobody can audit, and the conversion multipliers are the vendor's own funnel. The resolution: the free version (check your citations, brief your PR team) starts now; the paid version waits for third-party measurement.

What it hinges on

The whole thing turns on one fact and one unknown. The fact: referral traffic from search is falling, ~30% and real, because AI answers satisfy queries without a click. That's not seriously in dispute. The unknown: whether AI-referred visitors actually convert better, or whether that's a story a GEO vendor needs to be true. The council leans toward "traffic decline real, conversion premium unproven." Verify it yourself: tag AI-referred sessions in your analytics, watch their conversion against organic search for a quarter, and don't take Nurik's 6x on faith. Prediction: By Alphabet's Q3 2026 earnings, expected in late October 2026, Google's search advertising revenue will again grow year-over-year (mid-single digits or better) even as third-party measurement firms report continued declines in outbound referral traffic to publishers.

Confidence: High — the mechanism decoupling search revenue from referral traffic is structural, not temporary.

Why: The pattern was already visible in Q2 2026, with search revenue up roughly 17% year-over-year while click-through rates to publishers kept falling. The reason the two move in opposite directions is that clicks sent to publishers were always a cost to Google, never a revenue line. Keeping the query on-page, serving ads against the AI-generated answer, and raising minimum spend thresholds protects the money while destroying the traffic. For this to fail, a rival such as ChatGPT or Perplexity would have to pull search-query volume away from Google at a scale that dents ad revenue by the end of Q3 2026, and 2 billion AI Overview users plus Google's embedded position in the Android and Chrome toolbar make that hand-off unlikely in this window.

Revisit by 2026-10-30: We're right if Alphabet reports year-over-year search ad revenue growth in its Q3 2026 earnings while Similarweb, SparkToro, or a comparable firm shows continued referral-traffic decline to publishers. We're wrong if search ad revenue is flat-to-down year-over-year, or if referral traffic to publishers recovers materially.