Guide for agencies

Why does my AI and directory traffic show as 'Direct'?

The short answer

AI and directory traffic shows as 'Direct' because the AI apps and in-app browsers that send it strip the referrer header, so around 70% of AI-driven visits reach GA4 with no source it can read. GA4's May 2026 AI Assistant channel recovers some of that, but it is a floor, not a ceiling: it recognises a handful of chat apps, misses Perplexity (which lands in Referral) and Google AI Overviews (which lands in Organic), and still cannot see the referrer-less majority. The 'Direct' bucket is therefore part real dark AI traffic and part everything else, and that traffic converts several times better than what you can see. The honest response is not to guess which listing or engine created it, but to size the unmeasurable slice with self-reported source data and time-correlation, and to measure the channel by prompt-level citation rather than by clicks GA4 can attribute.

The size of the blind spot

How much of the AI channel is invisible by default.

These figures explain why the 'Direct' line in GA4 overstates typed-URL traffic and hides the highest-intent visits you have — and why a referred-click count is the wrong denominator for an AI or directory listing.

70.6%Share of AI-driven visits that arrive with no referrer header and land in GA4 as 'Direct', across a 446,405-visit sample — traffic a listing or engine may have created but cannot be traced to itLoamly, AI Traffic Attribution Crisis

4.1xHow much better this dark AI traffic converts than non-AI direct traffic (10.21% vs 2.46% transactional rate) — the visits you most want to measure are the ones hardest to seeLoamly, AI Traffic Attribution Crisis

May 2026When GA4 added a native AI Assistant channel (ChatGPT, Gemini, Copilot, Grok, Deepseek), broad by ~7 Jun — a floor: Perplexity still lands in Referral and Google AI Overviews / AI Mode in Organic SearchDigitalApplied, GA4 AI Assistant Channel

-42.6%Fall in AI referral traffic from its July 2025 peak even as visits to the AI platforms themselves rose ~28.6% year over year — the click channel is small and volatile, so do not read it as the whole storyThe Digital Bloom, Gen AI Traffic Share Feb 2026

47%Leads who pick the first option in a 'how did you hear about us' menu and are misattributed — self-reported source recovers dark traffic only if the field is designed to avoid first-option biasRuler Analytics, self-reported attribution

85%Share of AI-answer citations that come from third-party sources, not a brand's own domain — the reason the AI channel is better judged by prompt-level citation than by referred clicksOmnibound, AI Search Statistics

The diagnosis

Three reasons the traffic hides — and one ceiling you can't remove.

When an agency claims a directory listing or starts earning AI citations and then opens GA4 to see the payoff, the disappointment is almost always the same: the referral line barely moves while 'Direct' swells. That is not a tracking mistake you made. 'Direct' in GA4 is a default bucket for any session GA4 cannot attribute, and AI traffic falls into it structurally, for reasons that sit above your analytics setup. Understanding the three causes tells you which part of the gap you can close and which part you can only estimate — and stops you from either ignoring a real channel or crediting a listing for traffic it never sent.

1. The referrer header is stripped before the visit reaches you

Most AI answers are read inside an app or an in-app browser, and those environments frequently do not pass a referrer header at all. Whether it is stripped, truncated or never sent depends on the tool, the platform (app versus web), the operating system and the user's privacy settings — but the net effect is consistent: roughly 70% of AI-driven visits arrive with no source for GA4 to read, so they are filed under 'Direct' alongside genuine typed-URL and bookmark traffic. This is the single largest cause, and it is not something a UTM tag on your own site can fix, because the referrer is lost at the source, before the click ever lands on a link you control.

2. GA4's native AI channel is a floor, not a ceiling

In May 2026 GA4 added a native AI Assistant channel that recognises a set of chat assistants — ChatGPT, Gemini, Copilot, Grok and Deepseek — automatically, with broad availability by around 7 June. It is a genuine improvement, but its coverage is narrow in two ways that matter for anyone measuring AEO. Perplexity, one of the highest-intent AI sources, is not in the channel and still lands in Referral; and Google AI Overviews and AI Mode are routed to Organic Search, so Google's own AI surfaces never appear as AI traffic at all — you can only see them in Search Console. The channel captures visits that already carry a referrer; it does nothing for the referrer-less majority, which is why practitioners describe it as a floor.

3. 'Direct' is a mixed bucket, so you can't just relabel it

Because 'Direct' holds real typed-URL visits, bookmarks, some app traffic and the dark AI slice together, you cannot honestly reclassify all of it as AI. A server-side tag or a custom channel group can recover the fraction that carries any usable signal, and should — but a large residue will always be genuinely unattributable. The mistake to avoid is deciding, on no evidence, that the swelling 'Direct' line is 'obviously' your new Clutch listing or your ChatGPT citations. It may be, in part; it may also be a seasonal bump in returning visitors. Sizing that uncertainty, rather than resolving it by assumption, is the whole task.

The ceiling: some return is real but permanently dark

Here is the uncomfortable part. The dark AI slice converts about 4.1 times better than ordinary direct traffic, so the traffic you cannot see is disproportionately the traffic you most want to prove. That means a listing or an engine may be creating pipeline you will never trace to it — and the temptation to claim all of it is strong precisely because it converts so well. Resist it. The correct posture is to attribute what you can, estimate the rest with the two techniques below, and treat anything beyond the estimate as upside, never as the basis for a renewal or budget decision.

Read the bucket correctly

What the 'Direct' line hides, and what actually proves the channel.

The instinct is to make 'Direct' smaller or to explain it away. The productive move is to separate the part you can recover from the part you can only estimate, and to measure the AI channel where it actually shows — in citations, not clicks.

How to read AI and directory traffic that lands in GA4 as 'Direct'.
The question	The tempting wrong read	The honest read
What is in 'Direct'?	"It's mostly my new AI/listing traffic"	A mix of typed-URL, bookmarks and referrer-less AI — separable only in part
Did the listing send it?	Assumed from the timing	Provable only via a tagged link or a self-reported source
How big is the AI slice?	Guessed at 100%	Estimated from the ~70% dark benchmark plus your own recovered signal
Is the native channel enough?	"GA4 now tracks AI"	A floor: no Perplexity, no AI Overviews, no referrer-less visits
How to value the channel	Referred clicks in GA4	Prompt-level citation, mention and recommendation across engines
What to do with the residue	Credit the listing for all of it	Report it as unattributable upside, not as proof

The workflow

How to size and recover dark AI traffic, step by step.

You will not eliminate the 'Direct' blind spot, but you can shrink it, estimate what remains, and measure the channel where it is visible. Six steps, set up before you need the answer.

1. Recover the signal you can with a custom channel group

Keep a custom channel group alongside GA4's native AI Assistant channel, with rules that catch Perplexity (which otherwise lands in Referral) and the wider set of AI sources by referrer and landing-page pattern. This does not touch referrer-less traffic, but it rescues the fraction of AI visits that do carry a source and would otherwise scatter into Referral or Direct — and it is the only bridge to comparable data from before May 2026.

2. Tag every link you control with UTMs

The referrer is lost at the source, but any link you own — the outbound link on a directory profile, a newsletter, a signature — can carry UTM parameters so its clicks read as that source instead of dissolving into 'Direct'. Use one lowercase naming convention, tag directory and marketplace profiles the same disciplined way, and never tag internal links, which corrupts the model. This closes the part of the gap that is genuinely within your control.

3. Add a self-reported 'How did you hear about us?' field

A short source question on every inquiry form recovers what analytics cannot: the buyer who read your reviews in an app, then typed your URL. It is the highest-signal input you have on dark traffic — but design it to avoid the 47% first-option bias: rotate or randomise options, keep a free-text fallback, and never put your best guess at the top of the list. Persist the answer on the lead record next to the UTMs.

4. Estimate the dark slice instead of ignoring it

Use the ~70% referrer-less benchmark as a sizing tool, not a fact about your account: if your recovered AI signal is, say, 30% of the AI visits you can see, treat the true figure as materially higher and say so as a range. Anchoring the estimate to a published benchmark is more honest than either pretending 'Direct' is all typed-URL or claiming it is all AI. State the assumption every time you report the number.

5. Corroborate with time-correlation, carefully

Note when a listing went live or a citation appeared, then watch whether signed work and self-reported 'found you via AI' answers rise while other channels hold flat. Correlation is a hypothesis, not a receipt — a coincident campaign or season can move the same line — so use it to raise or lower confidence in the estimate, never to assert attribution you cannot tag.

6. Measure the channel by citation, not by clicks

Because roughly 85% of AI-answer citations are third-party and the click channel is small and volatile (down ~42.6% from its 2025 peak even as AI usage grew), the truest measure of AI visibility is prompt-level: are you cited, mentioned or recommended across engines, and with what share of voice? Track that in a prompt portfolio alongside GA4, and let the citation data — not the 'Direct' line — carry the verdict on whether the channel is working.

Signal vs noise

What to trust, and what to discount.

Two of these move you toward an honest number; two feel like proof and are not.

Trust: recovered + self-reported

A visit tagged by a custom channel group or a UTM, or a buyer who tells you 'I found you via ChatGPT', is real signal. Report it as the attributable floor of the AI channel.

Trust: prompt-level citation

Whether engines cite, mention or recommend you is measured independently of referred clicks, so it survives the 'Direct' blind spot. It is the metric the channel is actually about.

Discount: the whole 'Direct' line

'Direct' mixes typed-URL, bookmarks and dark AI. Reading all of it as your new listing or citations is the most common way agencies over-credit a channel.

Discount: 'it must be the AI'

A rising 'Direct' line after a listing launches is a hypothesis to test with tags and self-reported source — never attribution you can bank or bill against.

Track ROI on a paid listing UTM + attribution starter

Definition

Dark AI traffic, defined.

Dark AI traffic

AI-driven visits that reach your site with no referrer header — stripped by an AI app or in-app browser — so analytics files them as 'Direct' and cannot attribute them to the engine, answer or listing that sent them.

Dark AI traffic is the large share of AI-referred visits (around 70% in published samples) that arrive without a readable source because the referrer was stripped, truncated or never sent by the app the answer was read in. Analytics tools default these sessions to 'Direct', mixing them with typed-URL and bookmark traffic, so the AI channel looks smaller than it is — even though this traffic tends to convert several times better than ordinary direct visits. It cannot be fully attributed by tags alone; the practical response is to recover the fraction that carries any signal, estimate the rest against a benchmark, and measure AI visibility at the prompt level rather than by referred clicks.

Disclosure

Our own relationship, stated plainly.

This portal is an independent reference site, not an agency, and it does not sell optimization services or take a fee for placement. The operator also runs the agency Blobic, which is listed in this directory under exactly the same public criteria as every other agency, with a disclosure badge. Blobic paid nothing for its place and is never ranked above others; no position in this directory is for sale, to Blobic or anyone else. We note it here because a guide about measurement should be clear about incentives: our directory link carries no fee whose traffic we would have any reason to over-count, and we would rather you measure the AI channel accurately — citations included, dark traffic estimated honestly — than credit any listing, ours or another's, for pipeline it cannot prove it sent.

FAQ

Common questions about AI traffic in 'Direct'.

Why is my AI traffic showing as 'Direct' in GA4?

Because the AI apps and in-app browsers that send the visit usually do not pass a referrer header, so GA4 has no source to read and files the session under 'Direct' by default. Roughly 70% of AI-driven visits arrive this way in published samples. It is not a mistake in your setup — the referrer is lost at the source, before the click reaches a link you control, so a UTM tag on your own site cannot recover it. GA4's May 2026 AI Assistant channel captures visits that still carry a referrer, but the referrer-less majority remains in 'Direct'.

Does GA4's new AI Assistant channel fix the problem?

Only partly. The native AI Assistant channel, added in May 2026, automatically recognises ChatGPT, Gemini, Copilot, Grok and Deepseek when a referrer is present, which is a real improvement. But it is a floor, not a ceiling: Perplexity still lands in Referral, Google AI Overviews and AI Mode are counted as Organic Search, and any visit that arrives without a referrer — most AI traffic — still falls into 'Direct'. Keep a custom channel group alongside it to catch Perplexity and the wider set of sources, and to bridge to data from before May 2026.

How do I measure dark AI traffic if I can't attribute it?

Size it rather than resolve it. Recover the fraction that carries any signal with a custom channel group and UTM tags on links you own; add a 'how did you hear about us' field, designed to avoid first-option bias, to capture what analytics cannot; then estimate the remaining dark slice against the ~70% referrer-less benchmark and report it as a range, stating the assumption. Corroborate with time-correlation carefully, and treat the unmeasurable residue as upside, never as proof that a specific listing or engine sent it.

Should I just assume all my 'Direct' traffic is now AI?

No. 'Direct' is a mixed bucket that holds genuine typed-URL visits, bookmarks and some app traffic alongside dark AI, so reclassifying all of it as AI over-credits the channel exactly as badly as ignoring it under-credits it. A rising 'Direct' line after a listing goes live is a hypothesis to test with tags and self-reported source, not attribution you can bank. Recover what you can, estimate the rest, and keep the parts you cannot prove clearly labelled as unattributable.

If the AI click channel is small and falling, is AEO even worth measuring?

Yes, but measure it in the right place. Referred AI clicks are a small and volatile channel — down about 42.6% from a 2025 peak even as AI usage grew — so judging AEO by GA4 clicks alone will systematically undervalue it, especially since the dark slice you cannot see converts several times better. Because around 85% of AI-answer citations are third-party, the truest measure is prompt-level: whether engines cite, mention or recommend you, and with what share of voice, tracked in a prompt portfolio alongside your analytics.

How does this change how I judge a paid directory listing?

It means a referred-click count is the wrong denominator. Because most listing-influenced AI traffic can arrive with the referrer stripped, a paid listing may create pipeline that never traces back to it — so judge it by attributable signed work inside a window, add self-reported source to recover the dark path, and estimate the unmeasurable slice as upside rather than crediting the listing for it. An independent, criteria-based listing is also better judged by whether it feeds a citation than by the clicks it can prove, since its value is being read as evidence by the engine, not counted in GA4.

Next step

Measure the channel where it shows: in citations.

The AI channel is judged more truthfully by whether engines cite and recommend you than by the clicks GA4 can attribute. See how the portal tracks visibility, how it judges agencies, or list your agency under public criteria — there is no fee here whose traffic to over-count. Companies looking for a provider can use the directory rather than being sold to here.

The AI Visibility Index How we choose agencies List your agency