Does AI Search Traffic Convert Better Than Google?

An owner forwarded me a slide from an agency pitch deck last week: "AI search traffic converts 4.4x better than Google," bold text, no source, no fine print. She wanted to know if she should stop worrying about her Google ranking and pour everything into AI search instead. The honest answer took longer than a text message allows, so here it is properly.

The stat everyone's forwarding

The 4.4x number is real, and it comes from Semrush. In a June 2025 study, Semrush looked at conversion behavior across more than 500 high-value topics — mostly B2B, mostly the kind of subject a marketing or SEO agency would already be tracking closely — and found that the average visitor arriving from an AI search tool converted at 4.4 times the rate of the average visitor arriving from traditional organic search. It got picked up widely, and more than a year later it's still the number showing up in pitch decks and LinkedIn posts, usually stripped of the "500 B2B topics" part that explains what it actually measured.

That's not a knock on Semrush's methodology. It's a real finding, for the segment it studied. The problem is what happens to a real, narrow finding once it gets repeated a few hundred times without its scope attached — it quietly becomes "AI traffic converts better," full stop, applied to every kind of business by people who never read the original study.

What a peer-reviewed study found instead

Around the same time, a different, more rigorous piece of research was underway. Maximilian Kaiser and Christian Schulze, researchers affiliated with the University of Hamburg and the Frankfurt School of Finance & Management, spent twelve months — August 2024 through July 2025 — analyzing first-party Google Analytics data from 973 e-commerce websites that collectively generated $20 billion in revenue. They compared more than 50,000 transactions referred by ChatGPT against 164 million transactions from traditional channels: organic search, paid search, paid social, and affiliate.

Their finding, published in the peer-reviewed journal Marketing Science, ran the opposite direction of the Semrush number: for e-commerce specifically, ChatGPT-referred traffic converted worse than organic search, and revenue per session trailed both organic and paid search by a meaningful margin — organic outperformed ChatGPT by roughly 13%. ChatGPT traffic did beat paid social, so it wasn't the worst channel in the study. It just wasn't the standout the 4.4x figure implies.

Why these don't actually contradict each other

Two credible sources, two opposite conclusions, and neither one is lying. They measured different things.

Semrush's study covered B2B topics — the kind of purchase decision that involves research, comparison, and a longer sales cycle, where someone arriving via an AI tool has often already done more homework than someone doing a casual Google search. Kaiser and Schulze studied e-commerce — impulse-adjacent purchases, where a ChatGPT-referred visitor may be earlier in the decision, still comparing options the AI just summarized for them, rather than ready to buy. Different buyer, different moment, different result. Neither study is wrong about what it measured.

This is a known pattern, not a one-off. A separate analysis comparing how different research firms frame overlapping AI-search data found that Ahrefs and Semrush, looking at similar underlying traffic patterns, told two entirely different stories — one company emphasized a drop in click-through rate, the other emphasized a jump in conversion value — because each was measuring a different slice and each had a different story to tell about it. Nobody was fabricating numbers. Everybody was choosing which true thing to lead with.

The lesson isn't "don't trust studies." It's that a single multiplier, lifted out of its original scope and dropped into a pitch deck, tells you almost nothing about what happens on your specific website, in your specific industry, for your specific customer. It's also a reminder that conversion rate and visibility are two separate scoreboards — getting an AI tool to mention you at all is its own discipline, distinct from what happens after someone clicks through, and neither of the studies above says anything about the first one.

What to check on your own site instead

Neither 4.4x nor "13% worse" is the number that matters for your business. Your number is.

Here's the practical version: Google Analytics 4 added a native "AI Assistant" channel in May 2026 that automatically groups sessions with a recognized AI referrer — ChatGPT, Gemini, Perplexity, Claude — separate from your regular organic search line. We've covered how to actually find that traffic and what it can and can't tell you in more detail elsewhere, but the short version: open that channel, look at its conversion rate over the last few months, and compare it directly to your organic search conversion rate for the same window. That's the only comparison that's actually relevant to your business — not a B2B benchmark, not an e-commerce benchmark, yours.

If the sample is thin — a handful of sessions a month, which is normal for most small local businesses right now — don't force a conclusion out of a number that small. Watch it grow for a quarter before you draw anything from it either way.

What won't help

Adopting whichever stat flatters the plan you already wanted to run. If a pitch deck cites 4.4x to justify a budget increase, or a skeptic cites the e-commerce study to justify doing nothing, both are reaching for a number that fits a decision instead of informing one.

Assuming e-commerce findings apply to a service business, or B2B findings apply to a local retailer. The whole point of the disagreement between these two studies is that industry and purchase type change the outcome. A plumbing company, a boutique, and a B2B software vendor aren't the same buyer journey wearing different logos.

Treating a bad number for AI-referred conversion as proof AI visibility doesn't matter. Conversion rate and existing at all are two different questions. We've made the case elsewhere that getting cited isn't the same as getting paid — but the fix for that gap is measuring your own outcomes, not deciding the whole channel is worthless because one study measured a different industry than yours.

Waiting for a "final answer" before checking your own analytics. There won't be one. Studies will keep disagreeing as AI search keeps changing shape. Your own three months of data is available today.

FAQ

Does AI search traffic really convert 4.4 times better than Google?

For the specific segment Semrush studied — over 500 high-value B2B topics, in a June 2025 analysis — yes, that's what the data showed. As a claim about all AI search traffic across every industry, no; a separate peer-reviewed study of 973 e-commerce sites found the opposite for that industry.

Which study should I actually believe?

Neither one, applied blindly to your business. Believe the one that matches your situation — B2B versus e-commerce versus local service — and even then, treat it as a starting expectation to test against your own numbers, not a guarantee.

Why do two credible studies disagree this much?

They measured different industries, different time periods, and different definitions of "AI search traffic." Semrush studied B2B topic performance; Kaiser and Schulze audited first-party e-commerce transaction data. Different buyer behavior in each segment produced different results, and neither study's methodology was flawed — they just weren't measuring the same thing.

Does this mean I shouldn't bother with AI search visibility?

No. Conversion rate is a separate question from whether you show up in AI answers at all. A business that never gets mentioned by these tools has a bigger problem than a conversion-rate debate. The fix either way is measuring what's actually happening on your own site.

How do I check my own AI-search conversion rate?

In Google Analytics 4, look for the "AI Assistant" channel (added May 2026), which groups sessions with a recognized AI referrer separately from regular organic search. Compare its conversion rate to your organic search conversion rate over the same time window.

What if I don't have enough AI-referred traffic yet to compare?

That's normal for most small businesses right now. Let it accumulate for a full quarter before drawing a conclusion — a handful of sessions isn't a reliable sample in either direction.

Is a low AI-search conversion rate the same problem as low AI-search visibility?

No, and mixing them up leads to the wrong fix. Low visibility means AI tools rarely mention you at all — that's a citation problem. Low conversion means you're being mentioned but the visitor doesn't act — that's a page-experience or offer problem. They call for different work.

---

Sorting out which industry-wide stat actually applies to a specific business — instead of adopting whichever one showed up in an inbox first — is exactly the kind of ongoing work getting found on Google and AI is built around. The data changes every few months. Your own numbers are the one thing worth checking directly.