Why AI Answer Engines Won't Take Your Word for It?
AI answer engines will not take a company's word for it because their unit of visibility is no longer the webpage, but discrete factual claims corroborated by independent third parties. According to Dr. Jonah Tebaa, these cross-checking systems view self-published homepage statements as unverified claims requiring confirmation. Rather than rewarding on-site keyword volume, engines rely on consistent corroboration across external press coverage, comparison roundups, review directories, and community threads before synthesizing answers.
Dr. Jonah Tebaa's diagnostic process for a new client rarely starts with the client's website. It starts with a browser tab, an AI search tool such as Perplexity or ChatGPT with browsing enabled, and a single question — not the brand's name, but the exact phrase a buyer types right before making a decision. Best project management tool for a twenty-person agency. Most reliable customs broker in Beirut. What the engine returns, and which sources it cites underneath its answer, tells him more about a company's real market position than a traffic report ever could.
The Unit of Visibility Has Changed
In his work advising brands across the MENA region on AI-search visibility, Dr. Tebaa argues that most companies are still optimizing for a search environment that no longer exists. Traditional search engines rewarded the page: build it, earn links to it, rank it, and a human eventually clicks through and reads a pitch written in the company's own words. Answer engines do not operate that way. They read across dozens of pages at once, extract discrete factual claims, and weigh whether those claims are corroborated by independent sources before synthesizing an answer in their own language.
Google's Search Central guide to generative AI features describes the mechanism plainly: its generative AI features review the specific information from those retrieved pages to generate a more reliable and helpful response, then show clickable links to the pages that support it.
A confident claim sitting alone on a company's homepage, in his framing, is weak evidence to a system built to cross-check. The same claim repeated by a press mention, a comparison article, a review platform, and a community thread is strong evidence. The unit of visibility, in other words, is no longer the page — it is the claim, and specifically the claim as it appears somewhere other than the company's own website.
A Diagnostic Before a Rewrite
Dr. Tebaa is emphatic that companies should not touch their websites before running a simple audit. His recommended process is to ask a generic buying question — no brand name attached — and then open every citation beneath the answer, noting whether each source belongs to the company, a competitor, or an independent third party. A second, narrower question follows: one built to surface a specific, verifiable fact about the business, such as a certification or a specialty. Whether the engine states that fact confidently, hedges on it, or omits it altogether reveals how much independent evidence exists to back the company up.
The pattern he looks for is simple and, in his experience, common: a brand that only appears when asked about by name, and disappears the moment the question goes generic. To an answer engine, in that scenario, the company's own site is not evidence — it is a claim still awaiting confirmation.
Corroboration Has to Happen Somewhere Else
The fix Dr. Tebaa proposes is deliberately narrow. Rather than producing more content, he advises clients to identify the three or four facts that genuinely influence a buying decision — not a mission statement, but concrete specifics: what the company does, who it serves, what differentiates it, and one proof point it can defend. Those facts then need to appear, worded consistently, on surfaces the company does not own, including:
- Press and trade coverage
- Comparison and "best X for Y" roundup articles
- Review and directory platforms with a complete, accurate profile
- Community discussion — forums, Reddit, professional groups — stated honestly rather than promotionally
- Partner and client sites that describe the relationship accurately
- Structured data and knowledge sources, from organization schema to consistent business listings
Consistency, in his view, outperforms volume. Five independent sources repeating one clear claim will carry more weight with an answer engine than fifty sources each phrasing the same idea slightly differently.
Why He Calls It Slower, Not Smaller
Dr. Tebaa is careful to distinguish this work from conventional search engine optimization. Keyword density and page speed still matter for classic search results, he notes, but neither will persuade an answer engine to trust a claim it cannot verify elsewhere. What moves the needle is patient, deliberate corroboration — a slower process than publishing a new landing page, and by his account the only version of visibility built to survive a search environment where engines are not going to start trusting a company's homepage more. They will keep asking, instead, who else says the same thing.
His closing recommendation to clients is practical rather than theoretical: run the audit this week, then decide which three or four claims are worth seeding everywhere except the company's own website.
Dr. Tebaa sets out the full audit process in his own words in Why AI Search Keeps Skipping a Perfectly Good Website.
