Ask ChatGPT for the best project-management software, or Perplexity for the top CRM for a small team, and you won’t get ten blue links to sift through. You’ll get a short list of named companies. Your brand is on that list — or, for that buyer in that moment, it doesn’t exist. Not “ranked number seven.” Invisible.
That shift is now measurable. In a 2026 analysis of 117 AI answers across ChatGPT, Perplexity, and Gemini, specific companies were named in roughly 98% of commercial answers. For “which tool should I buy” questions, the page of links is effectively gone, and the shortlist is the whole game.
The encouraging part is who makes that list. In the same study, the top three domains earned only about 18–21% of the citations in any given category — a long tail, not a lockout. And only about 31% of the pages the engines cited contained any original data.
Two-thirds of “cited” content is opinion dressed up as authority. That gap is the opening, and here’s how to take it.
1. Find the questions that trigger a shortlist
Write down the ten questions a buyer actually types right before spending money — “best [category] for [use case]” — and run each through ChatGPT and Perplexity yourself. The brands that come back are your real competitive set, which is usually not the one on your slide deck.
You can’t earn a citation for a question you never checked.
2. Match the format AI reaches for
About 77% of cited pages were long, structured “best-of” and comparison posts — the shape that answers “what are my options?” directly. Thin, 500-word blog posts don’t get quoted; comprehensive pages with clear headers, real comparisons, and specific detail do. Build the page an engine can lift a clean, self-contained paragraph from.
3. Put one piece of original data in every article
This is the emptiest lane in the whole game. Only about 31% of cited pages carried a real number the author generated — a benchmark, a small survey, a before-and-after result. Add one, and you hand the model something quotable and vault into the minority that actually has evidence. It’s the single most repeatable move for getting cited.
4. Be present where buyers actually talk
Reddit was cited in roughly 87–90% of the Perplexity answers in the study. The AI often isn’t reading your carefully built blog; it’s reading the conversation. Show up in the communities where your category is discussed — disclosed, genuinely helpful, never spam. The forum thread is now a distribution channel.
5. Optimize for each engine — they barely agree
The three engines shared a single common recommendation only 56% of the time, and their full shortlists overlapped about 7%. Winning on ChatGPT tells you almost nothing about how you show up on Perplexity or Gemini. Treat them as three separate channels with three different source diets, not one “AI” to be gamed once.
6. Let the models in
None of this works if the engines can’t fetch your pages.
Confirm your robots.txt isn’t quietly blocking AI crawlers, and that your key pages are eligible for Google’s AI features. It’s a ten-minute check that gates everything above it. The pattern underneath all six is the same: getting cited isn’t about publishing more — it’s about publishing the kind of content an AI is comfortable quoting to a stranger who’s about to spend money. Comprehensive, specific, backed by real data, and present where the conversation happens. Everything cheap and generic is already invisible.
Which reframes the entire content budget. A $5 article that never gets cited isn’t cheap — it’s a donation.
In an AI-answer world, the number that matters is no longer cost-per-article; it’s cost-per-citation.
Start by auditing what the engines already say about your category, then close the gap one citable page at a time.
