Your ChatGPT Fix Doesn't Help Perplexity: The 11% Problem

Most businesses treat AI visibility as a single channel. Fix your directories, add schema, clean up your NAP -- and assume the work carries across ChatGPT, Perplexity, and Gemini equally.

Our June 2026 investigation found a number that breaks that assumption: only approximately 11% of domains cited by ChatGPT also appear in Perplexity's citation pool for the same queries. Nine out of ten sources ChatGPT trusts are sources Perplexity ignores -- and vice versa.

That is not a rounding error. It means platform performance does not transfer.

Why the Citation Pools Diverge

The 11% overlap figure comes from practitioner analyses we reviewed in session 34 of our ongoing citation behavior research (2026-06-01). It is directionally consistent with what we documented in session 15 (2026-05-12) using the Yext 2026 AI Visibility study -- the most detailed source breakdown available, built from 17.2 million citations across ChatGPT, Perplexity, and Gemini.

The Yext data explains why the overlap is so low. Each platform is drawing from a structurally different source pool.

ChatGPT pulls approximately 49% of its citations from third-party directories: Yelp, BBB, G2, Capterra, industry registries. It also mirrors Bing's organic results closely -- practitioners consistently report approximately 87% overlap between ChatGPT's cited sources and Bing's top-10 results for the same queries. If you rank well in Google but not Bing, that gap may explain more of your ChatGPT weakness than any content or schema issue.

Perplexity runs a live web search for every query and prioritizes niche expert directories specific to the category. For healthcare, it pulls medical registries. For legal, legal directories. For SaaS, comparison platforms like G2 and category-specific roundups. The general-purpose directories that drive ChatGPT citations often do not appear in Perplexity's source pool at all.

Gemini is different again. It pulls approximately 52% of citations from brand-owned websites. Structured schema, E-E-A-T signals, and well-indexed service pages carry more weight here than on either ChatGPT or Perplexity. A business with a thin directory presence but a well-built site will consistently score higher on Gemini than on ChatGPT.

Three platforms, three source architectures, three separate optimization problems.

What a Real Audit Looks Like

The 11% overlap figure is practitioner data. We can anchor it against our own audit records.

In session 16 (2026-05-13), we ran the first systematic source URL analysis from a Sourcepull audit -- the May 11, 2026 audit, which captured 102 Perplexity source URLs across 12 queries. Perplexity is the only platform that populates source URLs in our audit records; ChatGPT, Claude, and Gemini records had empty source URL arrays.

Two findings from that data stand out.

First: zero Reddit appearances across all 102 URLs. At the time, our fix plan template still recommended Reddit as a Perplexity citation surface. The audit data showed that for this category (AEO audit tools / SaaS), Reddit was not in Perplexity's citation pool at all -- consistent with the 86% Perplexity Reddit drop we documented following the October 2025 lawsuit. The recommendation was wrong, and the data caught it.

Second: the actual sources Perplexity was reading for category queries were tryprofound.com, meltwater.com, g2.com, and aeoaudittool.com. These appeared across 3-4 of the 4 category queries in the audit. The business had a 0.0/10 category score on Perplexity not because it lacked directories in general, but because it was absent from the specific comparison articles Perplexity was reading for its exact category.

A generic "add directories" recommendation misses this entirely. The source data shows which specific domains Perplexity trusts for a given category. The fix plan should name those domains and explain how to get listed in them.

How Long Fixes Take on Each Platform

The structural divergence also affects timeline expectations. After a fix is implemented -- a new directory listing, a schema update, a NAP correction -- how long until each platform registers it?

Our session 34 research (2026-06-01) surfaced practitioner estimates for platform ingestion speed:

| Platform | Structural fix ingestion | |---|---| | Perplexity | 2-7 days | | ChatGPT | 7-21 days | | Claude | 14-45 days | | Google AI Overviews | 14-45 days |

These are practitioner estimates, not controlled measurements. But the relative ordering reflects known platform architecture. Perplexity is fastest because it runs live retrieval on every query -- a new listing that gets indexed can start appearing in Perplexity answers within a week. ChatGPT is slower because it blends training data with selective retrieval. Claude and AI Overviews depend more heavily on training data cycles and can take 4-8 weeks to register structural changes.

The practical implication: if a business's primary gap is Perplexity, early progress is possible within 3 weeks of implementing the right fixes. If the primary gap is Claude, expecting visible movement within a month is unrealistic. The re-audit timing should match the target platform.

What This Means for Fix Plans

Platform-generic fix plans are an averaging problem. They recommend actions that have some positive expected value across all platforms -- directories help a little, schema helps a little, content freshness helps a little -- but none of these is the specific high-leverage action for any given platform's gap.

A Perplexity-specific fix plan looks like: identify the comparison articles and niche directories Perplexity is actually reading for this category, then close the gaps in that specific source set. Speed of appearance: 2-7 days for structural fixes.

A ChatGPT-specific fix plan looks like: broad directory presence plus Bing organic ranking, since approximately 87% of ChatGPT citations mirror Bing's top 10. Wikidata entity anchoring for brand name disambiguation. Timeline: 7-21 days.

A Gemini-specific fix plan looks like: brand site structure, LocalBusiness schema with complete sameAs links, and GBP completeness, since Gemini pulls 52% of citations from brand-owned sources and Google properties. Timeline: 14-45 days.

If a full audit isn't possible yet, the free Sourcepull Signal Check gives a per-platform visibility score in about 60 seconds -- which tells you which platform gap is largest before you decide where to invest fix effort.

The Diagnosis Question

The 11% overlap finding reinforces a principle we keep running into in our audit work: a single-platform score tells you almost nothing about the others. A business with a strong ChatGPT score and a 0.0 Perplexity score is not unusual. It is the expected outcome when a business has strong directory presence but hasn't appeared in the niche comparison sources Perplexity reads.

The diagnosis question is not "what is my AI visibility score?" It is "which platforms have gaps, why does each gap exist, and what is the specific source set I need to close it?"

Those are three separate questions. The platforms are not interchangeable.