ChatGPT Shows You a Map. Perplexity Shows You an Article. Same Query.

When a B2B buyer types "best PR agencies for SaaS" into ChatGPT, they get Google Maps links.

When they type the same query into Perplexity, they get editorial blog posts ranking agencies by specialization and results.

This is not a minor UX difference. It is a window into completely different trust architectures, and it showed up yesterday across six buyer-intent queries AT's brand visibility monitor ran through all four engines.

What the Data Actually Showed

AT's monitor runs 46 queries daily across Perplexity, ChatGPT, Gemini, and Claude. Yesterday's run included 15 buyer-intent queries: the kind real B2B buyers type when they are actively evaluating vendors, not doing research. Six of those queries revealed a structural divergence that has real operational implications.

For queries like "best PR agencies for SaaS," "best SEO agencies for AI search," "best IR agencies for VC-backed startups," and "best ghostwriting services for founders," ChatGPT returned Google Maps embed links as its primary citations. Perplexity returned editorial listicle content for the same queries.

Here is what that looks like in practice:

Query: "Best PR agencies for SaaS"

ChatGPT citations: Google Maps links for MADX Digital, 5WPR, Skale (all pulling from Maps entity data)

Perplexity citations: rankshift.ai/blog, shiftcomm.com/thinking, saashero.net (all editorial ranking content)

These sources have zero overlap. ChatGPT is not even seeing the editorial content Perplexity is drawing from. They are running entirely different retrieval pipelines on the exact same buyer query.

Query: "Best SEO agencies for AI search"

ChatGPT citations: Google Maps links for Growth Marshal, Graphite, First Page Sage

Perplexity citations: thriveagency.com/news, searchbloom.com/blog, sealglobalholdings.com

The pattern held across ghostwriting, investor relations, and earned media queries. ChatGPT consistently pulled from business directory and Maps sources. Perplexity consistently pulled from editorial review content. Gemini showed a third pattern: fewer citations overall, with more direct branded content from the agencies themselves appearing.

Three engines. Three completely different content type preferences. One buyer.

What Conventional Wisdom Gets Wrong

The dominant GEO playbook treats AI citation optimization as a single target. Write authoritative content, earn citations, get surfaced. The implicit assumption is that all engines pull from a similar pool of trusted sources and reward similar signals.

Yesterday's data breaks that assumption in a concrete way.

ChatGPT appears to be routing commercial buyer-intent queries through an entity-verification logic. If a business has a verified Google Business Profile with strong Maps data, it gets surfaced. From an anti-hallucination standpoint, this makes sense: Maps is verified, structured, and unlikely to contain fabricated information. But the consequence is that editorial content, detailed listicles, and thought leadership articles are largely invisible to ChatGPT for these query types. The engine is not consulting the editorial web. It is consulting an entity graph.

Perplexity is running the opposite logic. It treats editorial synthesis as the trust signal for commercial queries. The question it appears to be answering is: who wrote the most comprehensive, well-structured breakdown of these options? That editorial authority becomes the citation source. Your Maps presence is irrelevant to Perplexity for these queries.

What this means in practice is that a brand optimizing for Perplexity citation on "best PR agencies" will invest in editorial placement and listicle coverage. A brand optimizing for ChatGPT citation on the same query needs a verified Google Business Profile with complete categories, location data, and review velocity. These are not variations on the same strategy. They are different functions with different inputs, different owners, and different timelines.

The Machine Relations Implication

We have been building toward the idea that brands need to manage their relationships with AI engines as distinct institutional actors, each with its own preferences, trust hierarchies, and retrieval logic. This data is the clearest example yet of why that framing matters in practice.

If you are a SaaS PR agency and you are not appearing in AI answers to buyer-intent queries, knowing that gap exists is table stakes. The next question is which engine is missing you and why. ChatGPT absence and Perplexity absence have completely different root causes and completely different fixes.

A brand that treats AI visibility as a single optimization surface will spend budget on the wrong lever. They will build editorial content to fix a Maps gap, or build Maps presence to fix an editorial gap, and wonder why their citation rate is not moving.

The engines are not converging. The data from yesterday shows them pulling further apart in how they handle commercial intent. ChatGPT's entity-first logic and Perplexity's editorial-first logic are becoming more distinct as these products mature, not less.

Gemini's behavior adds another layer: it appeared to prefer direct brand sources over third-party editorial, which is a third model entirely. Claude was not included in the buyer-intent queries because it is not yet a primary discovery surface for B2B purchasing decisions, but its citation behavior on other query types shows yet another distinct preference pattern.

What to Watch

The practical question this raises for any brand selling to B2B buyers is: do you know where your citation gaps actually live by engine? Not in aggregate, but by engine and by query type.

The brands that answer that question with real data are going to hold very different competitive positions by the end of 2026 than the brands optimizing toward a generic "AI visibility" benchmark. The benchmark conceals the divergence. The divergence is where the actual leverage is.

That is what Machine Relations is built to surface.