Hidden AI competitors: the ones models cite in your place

Understand that AIs cite competitors you're not tracking, and how to reveal them.

Direct answer

When you ask an AI for the best brands in a sector, it often cites players you hadn't thought of — "hidden AI competitors." These are the brands that models surface spontaneously, well beyond the ones you track. In a real measurement, a brand that claimed to follow 4 competitors saw several dozen appear in the AIs' answers. Knowing them is decisive: in the eyes of a customer querying the AI, they are your real competitors — not the ones in your usual watch list.

The problem

You know who your competitors are. You follow them, you benchmark yourself against them, you know their strengths. Three, four, five names — your competitive landscape, the one you've had in mind for years.

Except that landscape is yours. The AI's is different. When a customer asks ChatGPT "what are the best brands in [your sector]," the AI answers with its list — and it may contain names you've never tracked, or even heard of. Those are the ones taking customers from you in the AI's mind, not necessarily the ones on your watch board.

The idea to grasp

You have to distinguish two types of competitors.

Declared competitors: the ones you name yourself. Your view of the market, based on your experience, your sector, your historical rivals.

Discovered competitors: the ones the AI surfaces spontaneously when queried about your field, without anyone having named them. It's the view of the market as the AI conveys it to your potential customers.

The gap between the two is often striking. In a real measurement in the tire industry (France configuration), a brand tracked 4 declared competitors (Continental, Pirelli, Bridgestone, Goodyear). The AIs, for their part, surfaced 76 discovered competitors in web search mode. Among the most cited were names the brand wasn't necessarily watching as a priority: Hankook, Falken, Uniroyal. Nineteen times more competitors than the declared list.

Why it matters: a customer doesn't consult your watch board. They ask the AI a question, and the AI serves them its list. If a player you're unaware of ranks high in it, then for that customer it's a real competitor — perhaps ahead of you. Not knowing the discovered competitors means ignoring half the playing field.

And the number itself depends on the conditions: from memory (stock), the AI surfaced 25 competitors; with the web (flow), 76. The competitive snapshot changes with the querying mode — all the more reason to measure rather than assume.

What you hear everywhere

"I know my competitors." You know your competitors. Not necessarily the ones the AI cites to your customers. The two lists can diverge sharply.

"The competitors are always the same." In your historical sector, maybe. In the AIs' answers, new names emerge — sometimes foreign, niche, or recent brands that your classic watch doesn't catch.

"Tracking that many competitors is unmanageable." You don't track them all the same way. The point first is to know who appears and how often — to tell the real AI rivals from marginal mentions.

My stance: facts only. Your list of competitors is an opinion based on your experience. The AI's list is measurable. Better to know the latter than to stick with the former.

My vision: reveal the competitive landscape as the AI sees it

From here on, the register changes: we describe the instrument.

Measuring hidden competitors means:

  • Recording every player cited spontaneously by the AIs on queries in your field, not just the ones you declare.
  • Quantifying each one: citation frequency, average rank, share of voice — to tell a serious rival from an isolated mention.
  • Comparing the modes (stock / flow): the underlying landscape (memory) differs from the current-events landscape (web).
  • Comparing providers: a competitor dominant with one can be absent with another.
  • Dating and sealing to track changes over time (a newcomer on the rise stands out over time).

Where LirenPrism stands

On its brand measurement axis, mAIr (LirenPrism) systematically records the discovered competitors: the brands the AIs cite in your field beyond the ones you track, each one quantified (frequency, rank, share of voice). It's one of the most striking contributions of a measurement — revealing rivals invisible to your classic watch.

mAIr shows you who the AI cites in your place and how many times. What it doesn't do: tell you how to overtake them — that's GEO, the agencies' job. mAIr provides the map; the strategy to conquer it belongs to others.

In brief

  • The AI cites discovered competitors your watch doesn't follow — its view of the market, not yours.
  • Real example (tires, France): 4 declared competitors vs. 76 discovered in flow (25 in stock).
  • In the eyes of a customer querying the AI, these hidden competitors are your real rivals.
  • mAIr reveals and quantifies them; it doesn't tell you how to overtake them (that's GEO).

Frequently asked questions

Why does the AI cite competitors I don't know?

Because its view of the market comes from what it learned and what it finds on the web — not from your watch. It can surface niche, foreign, or recent brands you're not tracking.

Are all these discovered competitors a threat?

No. Some appear often and high up (serious rivals), others are marginal mentions. That's why we quantify them (frequency, rank, share of voice) rather than settle for a list.

Does the number of competitors change with the query?

Yes. From memory (no web), the AI cites fewer; with web search, far more. The competitive landscape depends on the querying mode — hence the value of measuring both separately.