When it became clear that ChatGPT, Gemini, and Perplexity answer questions people used to ask Google, we did not start a new company. We retooled. 20 years working with algorithms, external confirmations, and entity alignment, plus our audit engine for the new models and a network of 100,000+ real people for research.
The same query a user once typed into Google now more often lands in ChatGPT, Gemini, or Perplexity. The answer structure has changed: instead of ten links — a connected summary; instead of word-level relevance — an assessment of how credible the brand looks to the model. The basic task is the same: make sure the brand is recognizable, described correctly, and included in shortlists.
Category architecture, entity alignment, external confirmations, technical markup, content strategies — all of this moves into the new era with adjustments. The adjustments are big, but incremental. Teams that have worked in search for a long time have a working model that the new mechanisms fit into faster than starting from scratch.
Treating AI visibility as a task like "we need one more writer", "we need more links", "let's rewrite the homepage". Sometimes one of those pieces really does pull. But on their own, they rarely move the summary the model builds around the brand — that summary is assembled from a dozen layers at once, and tweaking one layer does not change the overall portrait.
Buying a dashboard to measure AI visibility and expecting it to fix everything by itself. A good tool sees signals — we build one ourselves, and we know how it works from the inside. But a dashboard does not coordinate vendors, write copy, fix markup, or run research. It shows where things are bad. People hold the layers together.
Hiring one "full-stack" firm and hoping it will do everything. Technical fixes, PR, entity work, execution across multiple markets, research — no team honestly does all of this equally well. A firm that says it does usually does each direction at a mediocre level. The working model is one operator holding the overall logic while specialists join for specific steps.
What works is different: diagnosis comes first, then a plan of directions is formed, and one person is responsible for making them add up to one picture.
The most important decision in visibility work is which direction to launch first. If the order is wrong, even good tactics will not move the result: owned surfaces are not ready to receive traffic yet, external links lean on a split entity, new content is written in a category that is phrased differently in that language. Diagnosis almost always pulls out at least one of these patterns.
Our audit engine runs an audit in 20 minutes. Then a live specialist spends a couple of hours: clarifying, checking edge cases, setting priorities. $20 is realistic because most of the analysis is automated. After that, the client decides whether they need a specialist conversation for $200 and a work plan — or whether the report is enough.
Most often, it turns out that they do. The machine report shows patterns. The specialist conversation answers "what do we do with this in our specific situation".
A network of 100,000+ real people, each with their own device and their own history of internet use. Through this network, you can run surveys, assemble focus groups, launch public discussions with neutral wording, and test a product through specific use cases. The resource is rare, and potentially dangerous too — so we write the rules for using it ourselves, in advance and explicitly.
We connect real people from the network only to transparent work: surveys, focus groups, perception checks. When we initiate a public discussion, we publish a neutral question and do not interfere in that conversation again. Fake reviews, steering discussions, presenting participants as independent experts, political or government orders — all of that is outside what we do. The reason is operational, not moral: working against these rules over time destroys both the network and the reputation.
The full position is on the "Approach and principles" page.
It works best with companies for which AI visibility has already become a real channel. Equipment manufacturers with long sales cycles, research labs, technology companies, and brands working across multiple language markets. High-trust service businesses — medical technology, legal practices, educational institutions. Regulated industries also fit — we only enter after an individual review of the jurisdiction and registration form.
It works worse where the client wants speed at the cost of documentation. Visibility for AI systems builds slowly — several months, sometimes a quarter. Without a log of what was done, six months later it is unclear what exactly worked — and teams that initially brushed off the operating rules usually ask in month two to reconstruct from emails what was done in month one.
Who this fits and who it does not fit in detail — in the "Who it fits" block on the homepage and in the prohibited zones on the "Approach and principles" page.
UserSignals is the trading name of ECONDATA TECHSCRIBE LTD, a private limited company incorporated in the Republic of Cyprus under registration number HE455945. The company is registered for VAT in Cyprus under number 60046398P. Director: Vitalis Makris. Full legal identification, registered office address, and supervisory data protection authority are on the Imprint page.
If there is a suspicion that the brand is getting lost in AI results — $20 and a couple of hours will show exactly where. Get an express audit — $20 →