Back to blog
aeoai-citationsoriginal-researchstate-of-ai-citations
FogTrail Team·

ChatGPT's brand mention overlap with Gemini is just 58%, the lowest pairwise agreement between any two AI engines across three waves of data covering 300 engine-query pairs. ChatGPT has the lowest overlap with 3 of the other 4 engines, a volatility score of 39% (compared to Claude's 8%), and it dropped a well-established brand from its responses entirely between consecutive weeks.

The engine that carries the most AI search traffic is also the one that behaves least like the others. If you are tracking your brand's AI visibility by watching Perplexity, Gemini, or Claude, you are not seeing what ChatGPT is doing.

Context: 300 Data Points Across Three Waves

These findings come from FogTrail's three-wave citation study, which tracked 25 B2B SaaS brands across 20 queries and 5 AI search engines (ChatGPT, Perplexity, Gemini, Grok, and Claude). Each wave produced 100 engine-query pairs. Three weekly waves yielded 300 total data points, enough to separate durable behavioral patterns from one-off noise.

As of March 2026, the data shows that AI engine recommendations are not converging toward agreement. They are oscillating, with strong consensus stuck at 50%. And ChatGPT is oscillating the most.

The Data: ChatGPT Is Pulling Away From the Pack

Here is the Wave 3 pairwise brand mention overlap. Each cell shows the percentage of brands that both engines mention for the same queries.

PerplexityChatGPTGeminiGrokClaude
Perplexity-64%67%67%62%
ChatGPT64%-58%71%62%
Gemini67%58%-74%69%
Grok67%71%74%-75%
Claude62%62%69%75%-

ChatGPT-Gemini at 58% is the lowest pair in the entire three-wave dataset. Meanwhile, Grok-Claude hit 75%, the highest. Four engines are forming a loose cluster. ChatGPT is drifting away from it.

ChatGPT's Overlap Trend Across Three Waves

The three-wave trajectory makes the divergence clearer.

ChatGPT Overlap WithWave 1Wave 2Wave 3Direction
Gemini62%67%58%Diverging
Claude61%67%62%Diverging
Perplexity64%65%64%Flat
Grok58%70%71%Converging

ChatGPT is selectively aligning with Grok while pulling away from the Perplexity-Gemini-Claude cluster. The ChatGPT-Gemini relationship shows the sharpest decline: a 9-point drop from Wave 2 to Wave 3.

Volatility Score: ChatGPT Changes 5x More Than Claude

The volatility score measures the percentage of brand citations that changed between consecutive waves. It captures how much an engine's actual recommendations shift week to week.

EngineVolatility ScoreCitation Count (W1/W2/W3)
ChatGPT39%23, 12, 14
Grok27%2, 7, 7
Perplexity21%7, 5, 4
Gemini18%7, 6, 5
Claude8%6, 6, 6

ChatGPT's brand citation count swung from 23 to 12 to 14 across three identical runs. Claude produced exactly 6 brand citations in all three waves. That contrast, 39% volatility versus 8%, is the single clearest measure of how differently these engines behave.

Supporting Evidence: Three Moves Only ChatGPT Made

The overlap numbers tell a structural story. The individual brand movements make it tangible.

ActiveCampaign: Cited to Invisible in One Week

ActiveCampaign appeared on all 5 engines in Waves 1 and 2. ChatGPT cited it twice in Wave 2, including direct URL links. In Wave 3, ActiveCampaign vanished from every ChatGPT email marketing response. Zero mentions across all 4 email queries.

EngineActiveCampaign (W1)W2W3
Gemini444
Claude333
Perplexity332
Grok442
ChatGPT220

The other four engines continued mentioning ActiveCampaign normally. This was not a universal signal change. It was a ChatGPT-specific disappearance. A midmarket brand went from cited with links to completely invisible on the highest-traffic AI engine in a single week.

Netlify: A First-Ever #1 That Only ChatGPT Gave

Netlify had appeared in 28 engine-query pairs across two waves without a single #1 placement. In Wave 3, ChatGPT recommended Netlify ahead of Vercel for "best platform for deploying web apps." No other engine made this change.

MetricWave 1Wave 2Wave 3
Netlify at #10/14 (0%)0/14 (0%)1/16 (6%)
Vercel at #114/14 (100%)15/16 (94%)14/16 (88%)

Vercel still dominates at 88% position-1 rate. But the strongest structural pattern in the dataset, Vercel's lock on the top spot, was broken by ChatGPT and only ChatGPT.

Startup-at-#1 Rate: A 40% Drop

ChatGPT held a 25% startup-at-#1 rate across Waves 1 and 2, the highest of any engine. In Wave 3, it dropped to 15%.

EngineStartup at #1 (W1)W2W3
ChatGPT25%25%15%
Claude10%15%5%
Gemini5%10%10%
Grok5%10%10%
Perplexity0%5%5%

ChatGPT lost PostHog at #1 for analytics queries and Fly.io at #1 for hosting. No engine now exceeds 15% startup-at-#1 rate. The "ChatGPT is the most startup-friendly engine" narrative from earlier waves no longer holds.

What Makes ChatGPT Structurally Different

ChatGPT's divergence is not random noise. It has consistent behavioral quirks that no other engine shares.

Wikipedia reliance. ChatGPT sourced 10.4% of its citation URLs from Wikipedia in Wave 3. No other engine uses Wikipedia at all, across any wave. This unique source dependency means ChatGPT's knowledge graph draws from a fundamentally different information base than Perplexity, Gemini, Grok, or Claude.

Brand-owned URL preference. Despite declining from 24.2% to 18.4% across three waves, ChatGPT still links to brand-owned websites at roughly 4x the rate of Perplexity, Gemini, or Claude. As of March 2026, it rewards product pages and documentation more than any other engine, a behavior that matters for how brands approach ChatGPT optimization specifically.

Source type distribution (March average):

Source TypeChatGPTPerplexityGeminiGrokClaude
Brand-owned20.5%5.9%5.5%5.5%4.1%
Third-party review21.5%2.0%2.9%15.1%0.6%
Wikipedia5.3%0%0%0%0%
Other (blogs, docs)46.3%89.4%90.1%73.2%91.9%

ChatGPT's source mix is visibly different from every other engine. It pulls more heavily from brand sites and third-party reviews, and it is the only engine that uses Wikipedia as a meaningful source. The other four engines rely on blogs and documentation for 73-92% of their URLs.

What This Means

The engine that matters most is the hardest to predict. ChatGPT commands the largest share of AI search traffic. It is also the engine whose recommendations you are least likely to guess by monitoring Perplexity, Gemini, or Claude. As of March 2026, this is the central problem for any brand doing AEO: the most important engine is the outlier.

Single-engine optimization is riskier than it looks. If you optimize for ChatGPT specifically, you are targeting an engine that changes its brand recommendations more aggressively than any other. ActiveCampaign learned this the hard way. A brand can be cited and linked one week and completely absent the next.

Two AI search ecosystems may be forming. ChatGPT is converging with Grok (58% to 71% overlap over three waves) while diverging from Perplexity, Gemini, and Claude. If this trend continues, brands may need to think about AI search as two distinct clusters rather than one unified channel. The pairwise overlap data across all five engines supports this splitting pattern.

ChatGPT's volatility creates openings. Netlify's first-ever #1 happened on ChatGPT. Beehiiv's newsletter breakthrough started on ChatGPT. The same instability that makes ChatGPT dangerous for incumbents makes it the most accessible engine for challengers. If you are trying to break into a category, ChatGPT is where the door is most likely to open.

What You Can Do About It

  • Track ChatGPT separately from your multi-engine dashboard. Its behavior diverges enough from the other four engines that aggregated metrics will mask ChatGPT-specific changes.
  • Monitor your brand's ChatGPT presence weekly, at minimum. ActiveCampaign's disappearance happened between weekly checks. Monthly monitoring would have missed it entirely. A 48-hour monitoring cadence catches these shifts before they compound.
  • Invest in brand-owned content for ChatGPT specifically. ChatGPT sources from brand websites at 4x the rate of other engines. Pricing pages, feature comparisons, and technical documentation directly influence ChatGPT citations more than any other engine.
  • Do not assume ChatGPT position changes are permanent. Netlify's #1 may revert. ChatGPT's startup-at-#1 rate may rebound. The engine's 39% volatility score means changes work in both directions.
  • Watch the ChatGPT-Grok convergence. These two engines are increasingly aligning (58% to 71% overlap over three waves). Strategies that work on ChatGPT may start paying off on Grok as well.

Methodology

We ran 20 queries across 5 AI search engines: ChatGPT, Perplexity, Gemini, Grok, and Claude. Each query was sent as a real-time API call, simulating how actual users interact with these platforms. We tracked 25 B2B SaaS brands across 5 categories (CRM, Project Management, Email Marketing, Analytics, Dev Tools) over three weekly waves (March 6, 10, and 15, 2026), producing 300 engine-query pairs. Pairwise overlap is calculated as the Jaccard index of brand mentions shared between two engines across all 20 queries. Volatility score is the percentage of brand citations that changed between consecutive waves.

Frequently Asked Questions

How much does ChatGPT agree with other AI search engines on brand recommendations?

As of March 2026, ChatGPT's brand mention overlap ranges from 58% (with Gemini) to 71% (with Grok). It has the lowest pairwise agreement with 3 of the other 4 engines, making it the most divergent AI search engine in a five-engine comparison across 300 data points.

Can a brand disappear from ChatGPT in a single week?

Yes. ActiveCampaign went from being cited with direct URL links on ChatGPT to zero mentions across all email marketing queries in one week. The other four engines continued mentioning it normally. Single-engine disappearances can happen without any corresponding change on other platforms.

Is ChatGPT the most volatile AI search engine?

Yes. ChatGPT has a volatility score of 39%, meaning nearly 4 in 10 of its brand citations change between consecutive weeks. Claude, the most stable engine, has a volatility score of 8%. ChatGPT's citation count swung from 23 to 12 to 14 across three identical query runs.

Why does ChatGPT behave differently from other AI engines?

ChatGPT has several unique structural behaviors: it sources 10.4% of its URLs from Wikipedia (no other engine does this), it links to brand-owned websites at roughly 4x the rate of Perplexity or Claude, and its citation counts swing dramatically between weeks. These sourcing differences produce different brand recommendations.

Should brands optimize for ChatGPT specifically?

Brands should monitor ChatGPT specifically because its behavior cannot be predicted from other engines. However, ChatGPT-exclusive optimization is risky because the engine's 39% volatility score means any position gained can shift quickly. A multi-engine strategy with dedicated ChatGPT tracking is the most practical approach.

Related Resources