When Gemini Traffic Spikes in GA4: What Ecommerce Directors Should Check First

Written By Digital Team

If Gemini suddenly appears as a line in your GA4 traffic report, the first question is not “how do we scale this?” It is “where did this traffic sit last month?”

In early October 2026, Search Engine Journal reported that Google Gemini had started adding UTM parameters to some outbound links. Google’s John Mueller confirmed he could see it too. Google has not documented when or which links get tagged. For ecommerce directors, that matters because a new Gemini line can look like fresh demand when it is often a cleaner label for visits that used to land in Direct — especially on mobile, where referrers are often stripped.

This sits alongside the separate GA4 AI Assistant channel (referrer-based, medium ai-assistant), which we covered earlier. Keep that channel. Also check UTM and manual source dimensions. Annotate the date the tags first appear. Reconcile Direct against Gemini. Do not celebrate a spike until you have done that.

What changed (in plain English)

Search Engine Journal’s Roger Montti reported on 1 October 2026 that Gemini had begun adding UTM parameters to outgoing links so site owners could attribute referrals more clearly. Traffic from AI chatbots often falls into Direct in GA4 when there is no referrer or UTM data to tie the visit to a source.

Google’s John Mueller replied in the Reddit thread that sparked the report:

"Nice. I see it too. Doesn’t Gemini have a referrer too though? I thought you could already track them."

Source: Search Engine Journal — Google Gemini Adds UTM Parameters For Referral Attribution (Roger Montti, 1 October 2026).

Mueller also asked for reproducible detail on what happens to the referrer, and said it would be useful to keep the referrer as well as the UTM tagging:

"If someone has more information on what happens with the referrer (ideally with something I can reproduce), I’m happy to forward that to the team. It would be great to have these retained (in addition to the UTM-tagging)."

In plain English:

  • UTMs are a query-string label. When someone clicks a tagged link, the URL can carry campaign-style parameters that GA4 maps into manual source, medium and campaign dimensions. See Google’s notes on traffic-source dimensions and manual tagging and URL builders / UTM parameters.
  • A referrer and a UTM are not the same thing. A referrer can be stripped by in-app browsers and some mobile paths. A UTM rides on the URL, often survives those paths, and can appear in raw server logs even when the referrer does not.
  • One Reddit commenter (not Google) reported that the referrer mostly survives on desktop web, that pass-through from the Gemini mobile app runs around 9%, and that Android assistant invocations strip it entirely. Treat those figures as one commenter’s report, not a published Google measurement. The useful point is the mechanism: a large share of assistant clicks may have been landing in Direct.
  • Google has not documented the rules. SEJ notes the implementation is undocumented. ChatGPT, by comparison, only tags grounded web-source links. Do not invent exact utm_source / utm_medium values. Read the actual values from your own server logs and GA4 first.
  • This is separate from the GA4 AI Assistant channel. That channel groups traffic by referrer with medium ai-assistant. Keep it. UTM tagging is an extra signal in manual / campaign dimensions. Use both.

Industry coverage such as Technical SEO News frames this as a re-baseline: mark the date tags appear, treat the step-up as a measurement change rather than demand, and expect Direct to give back roughly what the new Gemini line gains.

For the earlier AI Assistant channel context, see our post GA4’s new AI Assistant channel: what ecommerce directors should do.

Why it matters for the P&L

  • A spike can look like growth when it is a label change. If last month’s Gemini visits sat in Direct, this month’s Gemini line is not new shoppers. Celebrating it without checking Direct risks over-funding the wrong channel.
  • Direct is often your “unknown” bucket. On mobile and in-app browsers, referrers get stripped. Assistant clicks that used to look like typed URLs may now arrive tagged. Direct can fall as Gemini rises — often good hygiene, not lost demand.
  • Budget decisions need a stable series. A month-over-month Gemini comparison that crosses the tagging start date compares two different measurements. Annotate the break before anyone uses the chart for spend.
  • Keep two views. Referrer-based AI Assistant traffic and UTM-tagged Gemini clicks can diverge by device. Directors need both, not a single slide number.
  • Undocumented means it can change. Exact parameter values are not published. Build reports from your logs.

A worked example

Imagine a UK homewares retailer. (An illustrative scenario, not a real client.) In early October, the analytics lead notices a new Gemini-related line in GA4 that was barely visible before. Paid social and email are flat. Direct sessions are down by a similar amount to the Gemini rise. Someone asks whether to shift content budget toward “Gemini SEO”.

The Head of Digital asks for three checks before any budget moves. First, annotate the date UTM-tagged Gemini clicks first appear in server logs and in GA4’s manual source / medium dimensions. Second, plot Direct next to Gemini for the same window — they move in opposite directions by roughly the same volume. Third, keep the AI Assistant channel report and add a UTM / manual view using the actual parameter values from their own logs — not guessed names.

The board update then says: Gemini visibility improved because outbound links started carrying UTMs; this looks like re-labelling from Direct, not a sudden demand spike. Content work continues; spend decisions wait for more stable tagged data.

Director checklist

  1. Do not treat a Gemini spike as new demand until you check Direct. Plot Gemini (or the tagged source you actually see) against Direct for the same dates.
  2. Annotate the date tags first appear. Mark it on your GA4 exploration and any board chart that crosses that date.
  3. Read the real UTM values from your own logs. Google has not published the exact parameters. Build filters from what your server and GA4 show.
  4. Keep the AI Assistant channel and add a UTM / manual view. Referrer-based AI Assistant (medium ai-assistant) and UTM tagging are complementary. Do not switch to UTMs alone.
  5. Reconcile sessions and revenue, not just sessions. If Gemini rises and Direct falls, check whether purchase and revenue totals still match your backend.
  6. Say the break out loud in reports. Any month-over-month Gemini comparison that crosses the tagging start date should note that attribution changed.
  7. Watch mobile vs desktop. Referrer loss is worse on mobile and in-app paths; UTMs matter most there.
  8. Do not rebuild your whole AI channel model yet. The tagging is undocumented. Prefer a short annotation and a dual view until Google publishes the rules.

Caveats

  • Undocumented by Google. When and which Gemini links get UTMs is not published.
  • Commenter figures are not official. Desktop vs mobile-app vs Android assistant pass-through rates in industry coverage come from a Reddit commenter, not a Google study.
  • ChatGPT’s rules are not Gemini’s. ChatGPT only tags grounded web-source links. Do not assume Gemini behaves the same way.
  • A Gemini line is a floor, not a total. Untagged assistant clicks may still land in Direct.
  • UTMs do not measure citations without clicks. Tagging describes humans who click through, not how often Gemini cites you without sending traffic.
  • Keep the AI Assistant channel. Referrer-based grouping still matters.

Final thoughts

#15/100 — When Gemini suddenly shows up in your traffic report, check Direct first. It may be a label change, not new demand.

Clearer attribution from Gemini is useful. It is not, by itself, proof of a new growth channel. The director’s job this week is simple: annotate the date tags appear, read the real parameter values from your logs, keep the AI Assistant channel, add a UTM view, and reconcile Direct before anyone celebrates a spike.

Not sure how Gemini and Direct are moving in your GA4 property? Run the free GA4 and GTM tracking scan at https://audit.taggurus.co.uk/, email info@taggurus.co.uk, or book a meeting today to discuss.

FAQ

Q: What changed with Gemini and UTMs?
A: Industry reporting in early October 2026 (Search Engine Journal) says Gemini began adding UTM parameters to some outbound links. John Mueller confirmed he could see it. Google has not documented the rules.

Q: Does a Gemini spike mean more demand?
A: Not necessarily. Many assistant clicks previously landed in Direct when the referrer was stripped. UTMs can re-label that traffic. Check Direct for the same window before treating the rise as growth.

Q: How is this different from the GA4 AI Assistant channel?
A: The AI Assistant channel is referrer-based (medium ai-assistant). UTM tagging is a query-string signal in manual source / medium / campaign dimensions. Keep both.

Q: What UTM values should we filter on?
A: Do not guess. Read the actual values from your server logs and GA4. Google has not published a public parameter list for Gemini.

Q: Are the mobile referrer pass-through numbers official?
A: No. Figures such as ~9% pass-through from the Gemini mobile app come from a Reddit commenter quoted in industry coverage, not from a published Google measurement.

Q: What should we tell the board?
A: That attribution improved for some Gemini clicks; annotate the start date; reconcile Direct; and avoid month-over-month Gemini claims that cross the break without a note.

Sources

Digital Team
Next

That ‘Optimize’ Banner in Google Tag Manager: What Ecommerce Directors Should Do Before They Click