Your buyer has four AI research assistants. None of them agree on your firm.
You've been to this party.
Packed room, industry event, the kind where everyone's working it and nobody's touching the canapés. Four people are standing in different corners. They don't know each other. They've never compared notes. But tonight, they're all forming an opinion about the same thing.
In the first corner: Alex. He's the most popular person in the room. He knows everyone, he's been to every conference, and he's already told three people tonight who they should call. Fast, confident, impossibly well-connected.
Across the room: Maya. She's holding a drink she hasn't touched and she's reading something on her phone. When you introduce yourself, her first question is, "What have you actually published on this in the last six months?" She has 47 tabs open. She's not impressed by reputation alone. She wants to see the work.
Near the door: Victoria. She's not looking at her phone. She's looking at you. She already ran the check before she walked in: credentials, rankings, third-party references. She barely nods when something passes muster. You notice when she does.
At the bar: James. Quiet. He's been here all evening, barely spoken, nursing the same drink. But whoever is making the decision checks in with him last. When James finally speaks, that's the one they choose.
Here's the problem: you can't win all four the same way.
What Alex talks about, Maya fact-checks. What impresses Victoria, Alex already assumed. James won't speak until the other three have. There's no single pitch that lands in every corner of this room.
The bigger problem: they're also not at a party.
Their names are ChatGPT, Perplexity, Gemini, and Claude. And the conversation about your firm is already happening, without you in the room.
Alex · ChatGPT
Meet Alex, the connector
Alex is ChatGPT. And 68% of B2B professionals use him weekly.
He runs on a Bing index, broad, current, conversational, and averages 7.92 citations per response. He's almost always the first call a buyer makes. Not because he's the most rigorous. Because he's the easiest. He gives a lay of the land. A mental map. A first shortlist.
"Oh, I know exactly who you should talk to."
If your firm isn't in Alex's answer, you don't enter the conversation. Everything else is downstream from this.
But here's the twist no one tells you: Alex is the hardest one to actually impress. Studies of brand visibility across all four platforms consistently show ChatGPT delivers the lowest owned citation share. That's because Alex works partly from memory. What he knows about you was built over time, from every third-party mention, every credible reference, every time your name appeared somewhere worth trusting. You can't walk up and introduce yourself at the party. You had to be talked about before you arrived.
Maya · Perplexity
Meet Maya, the fact-checker
Maya has 47 tabs open. She read the study, not the summary, the study. Ask her about a firm and she'll have primary sources before you've finished the question.
"I have 47 tabs open. Did you actually read the study, or just the summary?"
Maya is Perplexity. She is where the real vetting begins.
Maya's proprietary index, layered over Bing, produces an average of 21.87 citations per response, the highest of any AI platform. She doesn't trust old content: citation rates for sources older than six months drop from 82% to 37%. If your thought leadership is stale, Maya won't surface it. If you published that white paper 18 months ago and never followed up, she doesn't know you.
This is where firms that showed up in Alex's answer get eliminated. Maya is due-diligence mode. She expects recent, substantive, primary-source-backed content, quarterly at minimum.
Victoria · Gemini
Meet Victoria, the gatekeeper
Victoria doesn't ask where you went to school. She already knows. She checks credentials, rankings, endorsements, third-party mentions.
"Were you in the FT? What's your Chambers ranking? I'll need references."
Victoria is Gemini. She is the credentialed gatekeeper.
She runs on Google's own index, cross-referenced against the Knowledge Graph, the institutional memory of the internet. She averages 8.34 citations per response, and 81% of what she recommends comes from top-tier institutional sources only. Earned media, recognized outlets, established authority signals. She applies E-E-A-T criteria with no mercy.
A firm that produces excellent thought leadership but hasn't earned external mentions from recognized outlets doesn't pass her test. You can't self-certify with Victoria. The credential has to come from someone she respects.
James · Claude
Meet James, the closer
James speaks last. He says the least. He carries the most weight.
"I'll only recommend someone I'd stake my own reputation on."
James is Claude. He runs on the Brave Search index and averages 5.67 citations per response, the fewest of the four, by design. James isn't trying to be comprehensive. He's trying to be right. He synthesizes what Alex, Maya, and Victoria have surfaced, then gives a final assessment. He's the senior advisor who listens to everyone in the room before speaking.
If you weren't visible to the other three, James doesn't have enough material to make a recommendation. He can't vouch for a firm he doesn't know. The firms that appear in James's answers earned that position across the full breadth of the research process.
Why the room never compares notes
A study by Whitehat SEO and Qwairy analyzed 118,000 AI responses between January and March 2026 and found that only 11% of cited domains appear across multiple platforms. A separate analysis of 680 million citations reached the same conclusion independently: nearly nine out of ten sources Alex surfaces don't appear in Maya's results. And the reverse holds too.
But this isn't a zero-sum trade. A firm can appear on all four. Some do. What the data reveals is something more dangerous: appearing on one tells you almost nothing about whether you appear on the others.
Superlines' cross-platform analysis found citation-volume variance of up to 615x for the same firm between platforms. Not 6x. Not 60x. Six hundred and fifteen times. A firm dominant in one citation pool can be essentially invisible in another, not because they traded one for the other, but because these platforms don't share sources, indexes, or authority signals.
Alex, Maya, Victoria, and James are not comparing notes at this party. They never do.
"We show up in ChatGPT" is the beginning of measuring your AI visibility. Not the end. That check tells you Alex knows your name. It says nothing about Maya's vetting queue, Victoria's credential file, or James's final shortlist.
What this means for your firm
First: know which of the four your specific buyers use at each stage of their process. A general counsel at a Fortune 500 company may spend 15 minutes with Alex and two hours with Maya. A private-equity associate may run straight to Victoria. The buyer journey isn't uniform.
Second: build Alex over time. His answer depends on the weight of third-party mentions that existed before your buyer asked the question. You can't sprint to the party. You had to be worth talking about before it started.
Third: feed Maya. Fresh, substantive content, quarterly at minimum. Not recycled thinking. Primary-source-level analysis on the current issues your clients actually face.
Fourth: earn Victoria's respect. That means third-party authority: press coverage, rankings, peer citations, recognized-outlet mentions. You can't publish your way onto her list. Someone else has to vouch for you.
Fifth: trust that James follows from the other three. Make yourself visible, credible, and current across the full ecosystem, and James will have enough material to speak.
This isn't an SEO problem. It's a strategic presence problem, and it requires treating each AI platform as a distinct audience with distinct criteria.
The firms that figure this out first
They don't just appear in more AI responses. They appear in the ones that produce mandates.
The general counsel who started with Alex, ran due diligence with Maya, credentialed with Victoria, and got James's final sign-off, she's not calling firms she found on one platform. She's calling the firm that worked the whole room.
The cocktail reception doesn't end. It's running right now, in every browser where your next client is researching your category. The question is whether you're a name that comes up in every corner, or only one.
See where your firm stands across all four platforms.
Selectio maps your AI citation presence across ChatGPT, Perplexity, Gemini, and Claude, so you know exactly where you're visible, and where you're not.
Run your AI visibility checkSources: Whitehat SEO / Qwairy, "Perplexity vs ChatGPT vs Gemini: AI Citations," 2026 (118,000 AI responses); Averi, 680-million-citation analysis, 2026; Passionfruit, cross-platform citation research, 2026; Superlines, cross-platform citation analysis, 2026; LinkedIn B2B research, 2026.