A growing share of visitors now arrive after asking an AI assistant a question rather than typing a search query. ChatGPT alone accounts for the large majority of trackable LLM referral traffic industry-wide, with Claude, Gemini, and Perplexity each carving out smaller but fast-growing slices. The problem is that most teams’ analytics still can’t see this traffic for what it is. Between a third and most of these sessions arrive with no referrer header at all and land in “Direct” — the same bucket as someone who typed your URL from memory — which means the channel is often bigger than reports show, and almost nobody is measuring it accurately yet.
Getting this wrong has real consequences. If AI-assistant traffic is invisible, it looks like organic growth has stalled even when a meaningful and rising number of visits are coming from being cited in an AI answer. Worse, teams that don’t separate this channel can’t tell whether it converts differently — early data suggests AI referral visitors often land straight on high-intent pages (product, pricing) rather than browsing from a homepage, which changes what “good” looks like for that traffic.
Data Points to Track
- AI-assistant channel classification: sessions with a referrer matching known AI platforms (chatgpt.com, perplexity.ai, claude.ai, gemini.google.com, copilot.microsoft.com, meta.ai) grouped as a distinct channel, not folded into “Referral” or “Direct”
- Referrer-less landing pattern: direct-traffic sessions that land straight on a deep page (pricing, a specific feature, a docs article) rather than the homepage — a strong proxy for AI-assisted visits missing a referrer
- Landing page distribution for AI traffic: which pages actually receive AI-referred visits, since early data shows heavy concentration on pricing, product, and comparison pages rather than even distribution
- Conversion and engagement delta: bounce rate, time on page, and conversion rate for confirmed AI-assistant sessions compared with organic search sessions on the same landing pages
- Citation-to-click gap: where measurable, how often your brand is cited in an AI answer versus how often that citation actually produces a visit — a large gap signals content that’s being summarised rather than driving traffic
Setup Steps
- Add a custom channel grouping rule matching session source against a regex of known AI assistant domains, so this traffic stops being silently absorbed into existing channels.
- Tag confirmed AI-referral sessions in your analytics tool as a first-class dimension, not a filtered view, so every downstream report (conversion, retention, funnel) can be sliced by it.
- Build a “referrer-less but deep-landing” segment to approximate the AI sessions that lose their referrer header, and track its size over time even though it can’t be attributed with full confidence.
- Cross-reference landing pages against your GEO/AI-visibility efforts — if you’re optimising content to be cited by AI assistants, confirm the pages you’re targeting are the ones actually receiving the resulting traffic.
- Review the channel monthly rather than daily — AI referral volume is still small relative to organic search for most sites, and daily numbers are noisy enough to produce false trend signals.
Actionable Insights
A rising AI-assistant channel with landing pages concentrated on pricing and product pages is a signal that AI-driven visitors already have purchase intent by the time they arrive — worth treating more like a high-intent referral than a top-of-funnel discovery channel, and worth checking those specific pages convert well for a visitor who’s arrived with only a two-sentence AI summary as context. A large and growing “referrer-less but deep-landing” segment that outpaces your confirmed AI channel is a sign the true number is materially higher than what you can directly attribute — a reason to treat any AI-referral figure as a floor, not a ceiling, when reporting it upward.
Related Resources
Need help tracking this in your app?
Our team sets up analytics pipelines for mobile and web teams every day. Talk to us and get your first events flowing in under an hour.
Talk to an expert