Sign up with Google
Sign in with Zoho
A client came to us last year with a quiet panic. Their Google rankings were stable, their traffic reports looked fine, but new business had thinned to a trickle no one could explain. When we audited their brand visibility in AI, the reason surfaced fast. They had effectively disappeared from the place their buyers now started every decision.
This is their story, lightly generalized, because the pattern repeats across industries and the fix follows a predictable arc.
What AI Visibility Means, Briefly
AI visibility is how often, and how prominently, a brand appears inside answers generated by AI systems like ChatGPT, Perplexity, Gemini, and Google's AI Overviews. Unlike a search ranking, it is not a fixed position you climb. It is closer to a probability that a system names you when someone asks a relevant question.
That difference is why the client's dashboards showed nothing wrong. Rankings and sessions simply do not capture whether an assistant recommends you.
The Moment We Saw the Problem
We did something simple on the first call. We opened Perplexity and ChatGPT, typed the exact questions their customers ask before buying, and read the answers aloud.
Their name never appeared. Two competitors did, described confidently, positioned as the natural choices. The client had not lost a ranking battle. They had lost the recommendation, which happens earlier in the decision and matters more.
What struck them was how invisible the loss had been. No alert, no traffic cliff, just a slow erosion they only noticed in the sales pipeline weeks later.
Why They Had Disappeared
Digging in, the causes were unglamorous and common.
Their pages were written to impress humans, not to be read by machines. The plain facts about what they did sat three paragraphs down, wrapped in positioning language an assistant could not cleanly extract. Their information drifted across the web, with an old directory listing contradicting their current site. And almost no independent source mentioned them, so the systems had little outside signal to trust.
None of these were dramatic failures. Together, they added up to a brand that was technically online but illegible to the tools now shaping choices.
What the Evidence Told Us to Prioritize
We did not guess at fixes. The research gave us a clear order of priority.
Ahrefs, studying 75,000 brands in August 2025, found branded web mentions correlated with AI Overview visibility at 0.664, far above backlinks at 0.218, and found publishing volume barely mattered. Ahrefs notes this is correlation, not proven cause, but it pointed us away from churning out content and toward being referenced elsewhere. Muck Rack's review of over a million AI-cited links pointed the same way, with most citations coming from earned, non-paid sources rather than owned pages.
So the plan was not "write more." It was "be clearer, be consistent, and be talked about."
How We Rebuilt Their Visibility
The sequence mattered more than the volume of work, and we moved from cheapest to slowest.
First, we rewrote their core pages. The plain facts went to the top, stated directly enough that an assistant could quote them. This was fast and entirely in our control.
Second, we reconciled their information everywhere. We hunted down every listing and profile and made the details agree, correcting the contradictory ones before adding anything new.
Third, we worked on earned presence. Accurate directory listings, genuine reviews, and real coverage, built patiently over time. This was the slow part, and the one that eventually moved the needle most.
Fourth, we tracked each platform separately. This mattered more than expected. The client gained ground on one assistant well before another, and a blended score would have hidden that uneven progress. Because Perplexity behaves differently from ChatGPT in how it surfaces sources, we leaned on the specifics in this guide to getting ranked on perplexity to tune that platform's approach rather than treating all engines as one.
What the Recovery Looked Like
Progress did not arrive as a sudden spike. Their name reappeared first on lower-stakes questions, then gradually on the commercial ones that actually feed a pipeline.
The timeline was shorter than traditional SEO would have taken, which is consistent with what we see across this kind of work. NotionX, in its own engagements, reports early visibility gains for some clients within the first several weeks, while being clear that outcomes depend on market and competition. For this client, the lower-competition questions moved first, and the contested ones followed as the earned mentions accumulated.
The bigger shift was how they thought about it. They stopped treating AI answers as a curiosity sitting outside their real marketing and started tracking visibility inside them as a core number, reviewed monthly like any other metric.
If your own reports look healthy but your pipeline feels thin, do what we did on that first call before anything else. Ask your top buying questions inside two assistants and read the answers honestly. Whatever you find is your real starting point, and in our experience it is the moment the invisible problem finally becomes a fixable one.
