AI SEARCH RANKINGS ARE MOSTLY NOISE. HERE'S WHAT B2B FOUNDERS SHOULD BUILD INSTEAD.

A new paper surfaced this week with a finding that should change how most founders think about AI visibility. Research covered by Search Engine Journal on 11 July shows that AI visibility rankings fluctuate so significantly between runs that a single measurement is largely meaningless. The authors introduce a stopping rule: a threshold for how many readings you need before your data is statistically reliable. The conclusion is that most brands are reading noise and treating it as a strategy signal.

This matters now. AI search is not coming. It is here.

Your prospects are asking AI systems who to hire before they visit your website. "Which outbound agencies work with recruitment firms?" "Who do B2B founders use for LinkedIn pipeline?" These queries happen in ChatGPT, Perplexity, and Gemini before a referral conversation starts. If you appear in the answer, you are in the consideration set. If you do not, you are not.

The instinctive response is to check your AI visibility. Type your firm name, type your target question, see what comes back. The problem is that checking it once — which is what most founders do — is nearly useless. The research makes this explicit: rankings shift between sessions, between days, between different phrasings of the same question. A single run is a coin toss dressed up as data.

That means the right question is not "how do I get a good result today?" It is "how do I build the kind of presence that AI systems find reliably, across many runs, across many phrasings, across many entry points?"

Those are not the same question. The answer to the first is a copywriting exercise. The answer to the second is an infrastructure problem.

What we see in our own pipeline

We have run well over a thousand conversations through our own outbound pipeline, every message human-approved before it sent. When we look at what separates the conversations that progress from those that stall, the pattern is clear: no single data point tells you much. Accumulated signals are what move things forward.

Our system makes hundreds of stage decisions a month, advancing, flagging, or closing conversations based on patterns across timing, message content, and prior engagement history. A prospect who has replied twice, asked a clarifying question, and revisited the conversation carries a completely different signal profile from a prospect at the same funnel stage who replied once to a generic opener. The thread matters. The accumulation is the signal.

AI search citation works the same way. A brand that appears in three independent directories, has a published case study with a named outcome, has been mentioned in a relevant forum thread, and has a client review that names a specific result will be cited more consistently than a brand with a well-optimised homepage and nothing else. The systems are looking for the same thing our pipeline AI looks for: corroborating signals across multiple surfaces.

One optimised source does not build trust. Presence across multiple independent sources does.

Five moves that actually build durable AI visibility

Get onto more surfaces before you optimise any one of them. Directories, partner pages, industry lists, published articles, forum mentions: these are what AI systems draw from. If you exist only on your own site, you exist in one place. The first priority is named appearances in independent third-party sources, not better copy on your homepage.

Own the answer to the questions your buyers actually ask. What does a recruitment founder type into ChatGPT before hiring an outbound partner? Write the direct, specific answer to that question and publish it somewhere public and citable. Do three questions well before you try to cover twenty. Specificity is indexed. Breadth without depth is not.

Run visibility checks properly. A single query in one session tells you nothing reliable. Check the same question five to seven times across different sessions and days. Take the median. If the result varies wildly, you do not have stable visibility yet. That is useful information in itself. Build from that point.

Prioritise trust signals over copy. Named client outcomes, third-party reviews that mention specific results, partnerships with credible organisations: these are what AI systems weight heavily because they signal real-world validation. A testimonial from a client that names a specific outcome is worth more than a well-structured services page.

Measure what the citation actually produces. AI visibility is only valuable if it drives inbound contact. Track whether appearing in AI answers translates into enquiries or booked calls. If citation rate rises but conversation rate stays flat, the visibility is not doing the job you need it to do.

The window is open

The brands building durable, accumulated presence now will hold ground when AI search becomes the default top of funnel. The brands waiting until it is obvious will be playing catch-up in a market where the positions have already been taken.

SearchOS is built around exactly this logic: consistent presence that accumulates over time, across the surfaces that AI systems actually index and trust. Book a call and we will walk you through how it works for your category.