AI Visibility · AI Companion Apps
When a buyer asks an AI engine which one to use, a handful of names come back. We measure whether yours is among them, then build the third-party presence that puts it there.
AI Companion Apps specifically
The method below is the same whichever industry you are in. These are the parts that are not — and they are usually what decides whether the programme works.
AI tool aggregators are cited constantly by every engine for app recommendations, they are free, and most apps in this category have either no listing or a listing written in ninety seconds at launch. It is the cheapest correct work in the entire programme.
Perplexity in particular leans heavily on community discussion, and for this category the relevant threads are small, specific and highly influential. Presence there has to be genuine — the communities detect and punish marketing, and a burned account is worse than no presence.
"Best uncensored chat app" carries far more volume than any product name, and each capability question resolves through a different source mix. An app can be named consistently for image generation and invisible for roleplay, which is why we measure per capability.
This category is eighteen months old and the engines have almost no established answer to inherit. Being the named default now is achievable in a way it will not be once the market consolidates.
The method
Every major assistant declines explicit content. None of them decline the commercial question underneath it — which processor handles high-risk billing, which age-verification vendor to shortlist, which brand makes the better product. Those answers are being generated today, for your category, with someone named in them. The refusal everyone notices has distracted the whole industry from the recommendation nobody is measuring.
Brands are around six times more likely to be cited through third-party sources than through their own site. In restricted categories the gap is wider still, because the engines lean harder on sources they consider independent when the subject is sensitive. Roundups, comparison pages, community discussion, review platforms and directory profiles decide this. A site redesign moves none of it.
ChatGPT resolves heavily through Bing and licensed publisher content. Gemini and AI Overviews weight what Google has indexed and what it trusts. Perplexity leans unusually hard on community discussion, which is why Reddit threads decide more of this category than anyone expects. Copilot draws on Bing plus editorial. A programme optimising one surface will report healthy numbers while three others quietly fail.
Engines have to be able to tell that your company is one company — the legal name, the trading name, the product name and the domain all resolving to a single node rather than four unrelated stubs. In this industry that is routinely broken, because operators use different names across directories, media and their own footer for perfectly sensible commercial reasons. It is unglamorous, and it is frequently the single change that moves a score.
Each month we re-run your prompt set across all four engines and report the score, the per-engine breakdown, your share of voice against named competitors, and the citation source behind every mention. The number moves and you can see what moved it. Where an engine declines a prompt entirely we report that separately, because a refused prompt is a different problem from one a competitor won.
Category specifics
The method does not change between industries. The questions buyers actually ask, the sources the engines trust, and the rules you have to advertise inside all do.
Prompts we run for this category
Rules this category advertises inside
Citation layer
Each engine resolves a local recommendation through a different set of sources. These are the ones that decide whether a AI companion app gets named — and most of them are not your website.
Scope
Specifics rather than adjectives — and an equally specific list of what it does not cover, because unclear scope is the most common cause of a bad first quarter.
Included
Not included
What to expect, when
6 questions
They decline explicit content and they answer the commercial questions around it. Which payment processor, which age-verification vendor, which brand, which app — all of those return named recommendations today. The baseline measures which of your prompts return a list at all before it measures whether you are on it, and we report both numbers separately.
Entity and schema corrections can register within two to four weeks. Roundup placement and community presence compound over three to six months. We report monthly either way, so progress is visible before it is complete.
No, and treat anyone who does with suspicion. No platform publishes how it selects sources. We work from measured output and correlations documented across the field, we report honestly, and we re-baseline the source map quarterly because these systems change without notice.
Traditional SEO gets a page ranked. This gets a company named. The inputs barely overlap — most of the work sits in third-party sources, community and editorial rather than on your site.
We report it and explain what changed. These systems shift underneath everyone, and community content in particular gets weighted differently quarter to quarter. A report that only ever contains good news is not a report.
No. We do not produce, edit, generate, upload or distribute explicit material, and that exclusion is written into every contract. We work on positioning, acquisition, measurement and the third-party presence that drives it. This is a firm line, not a negotiating position.
Before any retainer, we run fifty prompts across four engines and show you exactly where you stand against the operators being named instead.
Request your audit