// AI ATTRIBUTION / CONVERSION BY ENGINE
Why Claude and Perplexity Out-Convert ChatGPT
Two studies ranked ChatGPT, Claude, and Perplexity by value. They disagree with each other completely. I went and read both, and the disagreement turned out to be more useful than either study alone.
Most reporting on AI referral traffic treats it as one channel with one number: how much did AI send you, how did it convert. That framing was always going to break once three assistants with three different user bases started sending meaningfully different traffic. It broke faster than I expected, and it broke in a specific, instructive way.
One term first, because this piece is about exactly this trap and I am not going to fall into it in my own headline. "Out-convert" here means economic value: revenue per visitor. On that measure, Perplexity and Claude beat ChatGPT clearly. On the narrower measure of conversion rate, the share of visitors who buy, the picture flips for ChatGPT specifically, and I show that data in full below rather than burying it. Both things are true. They are not the same claim, and the gap between them is the actual point of this post.
By revenue, Perplexity and Claude win
A 2026 benchmark joined 200 Stripe-connected SMB sites to their AI referral logs and calculated revenue per visitor by engine, meaning total revenue divided by every visitor that engine sent, converting or not. The ranking: Perplexity at $1.42 per visitor, Claude at $1.18, ChatGPT at $0.87. Gemini and Google's AI Overviews trailed further behind at $0.41 and $0.29.
The volume split makes the result sharper. ChatGPT sent 71 percent of the sessions in the cohort but only 53 percent of the revenue. Perplexity sent 8 percent of sessions and 19 percent of revenue. Claude sent 6 percent of sessions and 14 percent of revenue. Both smaller engines are punching roughly two and a half times above their traffic share. If you are reporting AI traffic as one blended number, ChatGPT's volume is quietly dragging your average down, not up.
This is a vendor benchmark, not an audited study, and 200 SMB sites is a real but modest sample. Treat the ranking as directionally solid and the exact cents as a snapshot, not a permanent scoreboard.
By conversion rate, ChatGPT wins outright
Here is the complication nobody puts in the same article as the number above. Seer Interactive ran a case study on a single client's traffic, industry undisclosed, and found a real reversal on one engine specifically: ChatGPT-referred sessions converted at 15.9 percent, ahead of Perplexity's 10.5 percent, Claude's 5.0 percent, and Gemini's 3.0 percent. All four beat Google organic's 1.76 percent by a wide margin, which is the part every recap article keeps. The part they drop is that ChatGPT led the pack, not trailed it.
The sample is real and disclosed: 1,370 AI-referred conversions against nearly 14 million organic sessions over the study window. That is one company's funnel and one buying cycle, with no disclosed industry to explain why the gap between AI and organic is this dramatic here specifically. It is a legitimate, disclosed data point, not a number to build a universal rule on.
So which is it. Do Claude and Perplexity out-convert ChatGPT, or does ChatGPT out-convert them? Both, depending on which word in that sentence you mean.
Rate and revenue are not the same question
Conversion rate answers "what share of visitors bought." Revenue per visitor answers "how much did all of them spend, on average, whether they bought or not." An engine can lose on the first question and win on the second if the buyers it does send have a meaningfully higher average order value.
That is the simplest honest explanation for why ChatGPT's rank moves between the two studies while Perplexity and Claude's relative order does not: the studies may not disagree about ChatGPT's traffic at all. ChatGPT converts more often, at a lower deal size. Perplexity and Claude convert less often, at a higher one. Neither study proves that mechanism directly, since they are two different companies' traffic, not one company's traffic split three ways. But it is the only reading that makes both results true at once instead of forcing you to pick a winner and discard the other paper.
The practical version: if your business lives or dies on volume of transactions, ChatGPT referrals may be your best channel. If it lives on deal size, Perplexity and Claude referrals may be quietly outperforming a channel three times their volume. A single "AI traffic converts well" line in a board deck cannot tell you which business you are.
And the ranking is already moving under you
Whichever ranking you trust today has a shelf life. Goodie Research Lab ran the same B2B referral panel twice, eight months apart, and found Claude's share of B2B AI referrals grew from 1.4 percent to 18.5 percent, while ChatGPT's B2B share fell from roughly 89 percent to roughly 63 percent over a comparable window. That is not a rounding change. That is a fundamentally different competitive picture from the one that existed eight months earlier, in the specific segment, B2B, where deal sizes are largest and where the revenue-per-visitor gap above matters most.
Put the three findings together and the picture is not "Claude and Perplexity are better." It is that the value of each engine is business-specific, metric-specific, and moving fast enough that a study from eight months ago is closer to history than to guidance.
What actually changes if you take this seriously
Stop reporting a single AI-traffic number. Split revenue and conversion rate by engine, the same way you already split them by campaign. GA4's native AI Assistant channel gives you the medium, but not a clean per-engine revenue view out of the box; you have to build that cut yourself, using source dimension against the domains each assistant actually refers from.
Weight your answer-engine optimization effort by revenue contribution, not by which engine sends the most sessions. A channel that is 6 percent of your traffic and 14 percent of your AI-attributed revenue deserves more attention than its session count implies, and a channel that dominates session count but trails on revenue per visitor is not automatically your top priority just because it is the loudest number in the dashboard.
Re-check the ranking on a calendar, not once. Given how fast Claude's B2B share moved in eight months, a per-engine strategy set once at the start of the year and never revisited is optimizing against a market that no longer exists. That same discipline, measuring rather than assuming, is what the referrer-recovery work in the dark AI traffic piece is really arguing for, and what the search-volume accounting problem in the search traffic piece is arguing for from the other direction. None of this works if the underlying number was never measured honestly in the first place.
Getting AI turned into money means tracking the metric that matches your actual business model, not the one that made the best headline, which is the same discipline underneath AI Commercialization: the complete guide.
Frequently asked questions
Do Claude and Perplexity really convert better than ChatGPT?
By revenue per visitor, yes: a 200-site Stripe-connected benchmark found Perplexity at $1.42 and Claude at $1.18 per visitor against ChatGPT's $0.87. By conversion rate, no: a Seer Interactive case study on a single client's traffic found ChatGPT converting at 15.9 percent against Perplexity's 10.5 percent and Claude's 5.0 percent. Both studies are real. They measure different things, and only ChatGPT's position moves: Perplexity stays ahead of Claude on both measures, ChatGPT is last by revenue and first by rate.
What is revenue per visitor and why does it differ from conversion rate?
Conversion rate is the share of visitors who buy. Revenue per visitor is total revenue divided by all visitors, which folds in how much each buyer actually spends. An engine can have a lower conversion rate and still produce more revenue per visitor if the customers who do convert have a higher average order value. That gap is exactly what the ChatGPT versus Perplexity and Claude data shows: ChatGPT wins on the rate, Perplexity and Claude win on the dollar.
How fast is Claude's referral traffic growing?
In B2B referrals specifically, fast. Goodie Research Lab's longitudinal panel found Claude's share of B2B AI referrals grew from 1.4 percent to 18.5 percent over eight months, while ChatGPT's B2B share fell from about 89 percent to about 63 percent over a comparable window. Any per-engine ranking measured before that shift is already out of date.
Should I trust a single conversion-rate study?
Treat it as one real data point, not a universal law. The Seer Interactive figures come from one client's traffic, industry undisclosed in the published case study, with 1,370 total AI-referred conversions against nearly 14 million organic sessions in the comparison period. That is a real, disclosed sample, not an invented number, but it is one company's funnel, one product category, and one buying cycle. A consumer ecommerce site would likely show a different ranking entirely.
How do I track AI referral value by engine instead of lumping it together?
Split revenue and conversion rate out per engine in your analytics rather than reporting a single blended AI-traffic number, using GA4's source and medium dimensions or a custom channel group per assistant. Track both conversion rate and revenue per visitor separately, since this piece shows they can rank engines in opposite orders. Then weight your answer-engine optimization effort toward whichever engine is actually paying for the traffic it sends, not the one sending the most of it.
Sources
Revenue-per-visitor benchmark by engine (Perplexity $1.42, Claude $1.18, ChatGPT $0.87, plus the session-versus-revenue share split): Attrifast, "AI Traffic Revenue Benchmark 2026" (May 26, 2026; 200 Stripe-connected SMB sites, a vendor benchmark, not an audited study). Conversion-rate-by-engine case study (ChatGPT 15.9 percent, Perplexity 10.5 percent, Claude 5.0 percent, Gemini 3.0 percent, Google organic 1.76 percent; 1,370 AI-referred conversions against roughly 14 million organic sessions): Seer Interactive, "Case Study: 6 Learnings, 1 Site, How Traffic from ChatGPT Converts" (one client, industry undisclosed, October 2024 to April 2025). Claude's B2B referral-share growth and ChatGPT's corresponding decline: Goodie Research Lab, AI search traffic report (Wave 2 published May 21, 2026, an anonymized business-site panel measured against an October 2025 baseline; a longitudinal vendor panel, not an audited study).
About the author
Jeff Brokaw is a sitting CMO and Certified Chief AI Officer who ships AI in production, not slideware. He has been building AI systems commercially since 2016. He built the engine behind $185M in new business for a defense manufacturer, and authored the go-to-market behind a $114M institutional raise that came together in under 30 days.