Why referral traffic from AI search engines is not all equal
Most e-commerce brands have started noticing a trickle of traffic from ChatGPT, Perplexity, and Gemini showing up in their analytics. The instinct is to lump it all together under "AI referrals" and move on. That's a mistake. These three platforms work very differently, attract very different users, and send very different kinds of visitors to your site.
The question that actually matters is not which AI engine gets the most queries. It's which one sends people who are ready to buy.
Let's break down each platform honestly, based on what we know about their users, their citation behaviour, and the referral patterns brands are actually seeing.
ChatGPT: high volume, but buyer intent varies wildly
ChatGPT is the biggest name in AI search by a significant margin. As of mid-2025, OpenAI reported over 200 million weekly active users. That's a huge potential audience. But size isn't the whole story.
The majority of ChatGPT queries are still conversational and exploratory. People use it to brainstorm, draft content, learn about topics, and solve problems, not primarily to shop. When someone asks ChatGPT "what are the best standing desks under £500?", they're often at the very start of a research journey. They're not necessarily about to open their wallet.
That said, ChatGPT's Shopping features (rolled out in early 2025) have changed things somewhat. When users search with clear product intent, ChatGPT now surfaces product cards with images, prices, and links. Traffic from those product-focused interactions is considerably warmer than general conversational referrals.
What ChatGPT referral traffic actually looks like
Brands that have tracked ChatGPT referrals consistently report a split between two very different visitor types. The first group arrives with a specific question, reads one page, and leaves. Bounce rates for this segment can be above 70%. The second group, typically driven by a product or comparison query, explores multiple pages and has conversion rates that rival organic Google traffic.
The implication: if your schema markup and site structure clearly signal product availability, pricing, and reviews, you have a real chance of catching that second group. Without it, you're mostly getting curious browsers.
Perplexity: smaller audience, but the intent is exceptional
Perplexity is often overlooked because its user base is much smaller than ChatGPT's. That's the wrong way to think about it. Perplexity positions itself explicitly as a research and answer engine. Its users are not messing about - they are people who want a fast, sourced, accurate answer so they can make a decision.
This self-selection matters enormously. A person who downloads and uses Perplexity is almost by definition more deliberate than the average web user. They've actively chosen a tool that prioritises cited sources and structured answers over casual chat. When that person asks "which project management tool is best for a five-person agency?", they are not daydreaming. They are about to make a choice.
Citation behaviour and how Perplexity chooses sources
Perplexity is also notable for how transparently it cites sources. Every answer includes numbered citations that link directly to the pages it drew from. This means that if Perplexity cites your product page or your blog post, users can and do click through. The click-through rates from Perplexity citations are, anecdotally, the highest of the three platforms we're discussing.
The challenge is getting cited in the first place. Perplexity favours pages that are clearly structured, factually dense, and authoritative. Schema markup plays a direct role here: well-structured product data, review aggregates, and FAQPage schema all help Perplexity parse and trust your content. This is exactly the kind of work that FlinnSchema approaches differently from traditional SEO agencies.
For e-commerce brands especially, Perplexity referrals tend to be smaller in raw volume but punching well above their weight in conversion rate. Some brands are reporting Perplexity-referred visitors converting at two to three times the rate of standard organic traffic.
Gemini: integrated into Google, but still maturing
Gemini sits in an unusual position. It's Google's AI, which means it has unparalleled distribution: built into Android, Gmail, Google Workspace, and increasingly woven into standard Google Search results through AI Overviews. The potential reach is enormous.
However, Gemini's referral traffic to third-party sites is currently the hardest to measure and, based on early data, the weakest converter of the three. There are a few reasons for this.
First, a large portion of Gemini interactions happen within Google's own ecosystem. When someone uses Gemini inside Google Search, they often get their answer without ever leaving. The AI Overview satisfies the query. Traffic that might have gone to your site stays on Google's pages.
Second, Gemini's user intent skews heavily towards information retrieval rather than purchasing. People use Gemini to understand things. That's valuable for brand awareness but less immediately useful for driving sales.
Where Gemini does send valuable traffic
There are specific scenarios where Gemini referrals are worth pursuing. Local and service-based queries work well: "best accountant in Manchester" or "where can I get bespoke curtains in Bristol" can result in Gemini surfacing specific business listings, particularly when those businesses have strong structured data. For e-commerce, queries that blend product research with local fulfilment (same-day delivery, in-store pickup, local availability) are areas where Gemini tends to pass traffic through.
Google's emphasis on structured data is no accident here. Gemini pulls heavily from the same signals that influence rich results in standard Google Search. If your Product, Offer, and Review schema is clean and complete, Gemini is more likely to surface your pages in contexts where it does pass traffic. You can read more about this in our post on how to use Offer schema to win AI shopping recommendations.
Head-to-head: conversion quality by platform
Let's be direct about what the patterns suggest, based on what brands and agencies working in this space are observing.
Perplexity sends the highest-quality traffic in terms of conversion intent. Users are deliberate, the citation model creates trustworthy click-throughs, and the platform actively rewards well-structured, authoritative content. Volume is low but quality is high.
ChatGPT sends the most volume overall, but it's mixed. Product-intent queries via ChatGPT Shopping can convert well. Conversational or research queries tend to result in low engagement. The gap between these two visitor types is wide enough that treating all ChatGPT traffic as equivalent is a significant analytical error.
Gemini currently contributes the least direct referral traffic for most e-commerce brands. Its strength is influence: appearing in Gemini's AI Overviews shapes brand perception even when it doesn't produce a direct click. This has value, but it's harder to measure and slower to monetise.
What this means for how you optimise
The practical takeaway is that optimising for all three AI engines at once is both possible and worthwhile, but your priorities should reflect your goals.
If you want near-term conversions, focus your energy on Perplexity and ChatGPT Shopping. Both reward structured product data, clean schema implementation, and pages that answer specific buyer questions clearly. Getting your Product, Offer, and Review schema in order is the single highest-leverage move you can make.
If you're playing a longer brand-building game, Gemini coverage matters because of its integration into the Google ecosystem. But don't expect Gemini to drive direct sales volume in 2026 the way Perplexity can.
It's also worth noting that the platforms are evolving fast. ChatGPT's paid advertising layer (launched in 2025) introduces sponsored results alongside organic citations. Whether organic AI visibility still matters now that ChatGPT runs ads is a real question brands should be asking.
The brands that will win AI search traffic across all three platforms are the ones building genuine authority: well-structured pages, real customer reviews, clear product information, and the kind of structured data that makes it easy for an AI to understand what you sell, at what price, and why someone should trust you.
If you're not sure where your site currently stands, an AI visibility audit is the fastest way to find out what's missing and what to prioritise first.
Frequently Asked Questions
Which AI engine sends the most traffic to e-commerce sites?
ChatGPT sends the highest raw volume of referral traffic because of its large user base. However, Perplexity typically sends visitors with higher purchase intent, and many brands find Perplexity referrals convert at a higher rate despite lower numbers. Gemini currently contributes less direct referral traffic for most e-commerce sites.
How do I track which AI engine is referring traffic to my site?
In Google Analytics 4, look for referral sources including chatgpt.com, perplexity.ai, and gemini.google.com. ChatGPT traffic may also appear as direct traffic in some configurations if the user clicks from the app rather than the web. Setting up UTM-tagged links where possible helps, though AI engines don't always preserve UTM parameters. Creating segments by source in GA4 and monitoring them weekly gives you the clearest picture over time.
Does schema markup actually influence whether AI engines cite my site?
Yes, significantly. Perplexity and Gemini both rely on structured data signals to understand and trust your content. ChatGPT's product features also pull from structured product data. Schema markup alone won't guarantee citations, but it removes friction and makes your pages far easier for AI models to parse accurately. Pages without schema are effectively invisible to these systems for anything beyond basic text extraction.
Should I treat AI referral traffic the same as organic Google traffic?
No. AI referral traffic has different intent patterns, different bounce behaviour, and different conversion paths compared to traditional organic search. Someone arriving from a Perplexity citation has often already read a summary answer and is clicking for confirmation or to purchase. Someone from a ChatGPT conversation may be early in their research. Analysing these segments separately in your analytics will reveal which platforms actually move the needle for your business.