Products SEO & GEO Agent Paid Ads Agent Website Agent Conversion Agent
How it works Pricing FAQ Blog Book a call

Marketing attribution basics: from AI chat traffic to GA4

Updated · 8 min read

Short answer Marketing attribution decides which ad, search or post gets credit for a lead. GA4 now offers data-driven and two last-click models; first click, linear, time decay and position-based ended in November 2023. Keep data-driven. Clicks from ChatGPT, Gemini or Copilot get their own AI Assistant channel, but answers people read without clicking get credit nowhere.

Want help applying this to your business? Talk it through on a free 30-minute call.

Book a call

What marketing attribution means

Attribution is how your reports split the credit for a lead among the places the customer came from. Google defines it this way:

"Attribution is the act of assigning credit for important user actions to different ads, clicks, and factors along the user's path to completing the action." (Analytics Help: Get started with attribution)

Take one customer. She clicks your Facebook ad, later opens your site from a ChatGPT answer, then searches your name on Google and books. That's three visits and one lead. The attribution model is the rule that decides who gets the booking: the ad, ChatGPT, Google, or a share of each.

The setup that works for most small businesses, as of November 2026:

  1. Mark your leads as key events in Google Analytics 4 (GA4): a form sent, a call tap, a booking. See GA4 key events for a small business.
  2. Keep the default model, data-driven, and the default 90-day lookback window.
  3. Tag every ad link. GA4 counts a social visit as Paid Social only if the link's medium marks it paid, such as "cpc" or "paid_social" (Analytics Help: Default channel group). Without the tag, your Facebook ad's visitors are mixed in with free posts.
  4. Ask every new customer "How did you hear about us?" No model credits an AI answer nobody clicked.

Which attribution models still exist?

Fewer than most articles list. Google retired four rule-based models in 2023:

"The first click, linear, time decay, and position-based attribution models are no longer available as of November 2023." (Analytics Help: Get started with attribution)

Google Ads dropped the same four and moved conversion actions that used them to data-driven (Google Ads Help: About attribution models).

Here is how each model would credit the path above (Facebook ad, then ChatGPT link, then Google search):

ModelHow it splits creditWho gets the bookingIn GA4 today?
First clickAll to the first clickFacebook adNo
LinearEqual sharesA third eachNo
Time decayMore to recent clicksMostly Google searchNo
Position-based (U-shaped)Most to first and lastMostly Facebook and GoogleNo
Data-drivenBy how much each click raised the odds of a lead, learned from your dataA share each, in decimalsYes, the default
Paid and organic last clickAll to the last click, skipping direct visitsGoogle searchYes
Google paid channels last clickAll to the last Google ad click; with none, as the row aboveGoogle searchYes

Google Ads keeps two: data-driven, "the default attribution model for most conversion actions", and last click (Google Ads Help: About attribution models).

Which attribution model should a small business use?

Data-driven, the default. Whether your leads and clicks are tracked at all matters far more than the model.

Your situationWhat to use
Nearly all leads come from one channelAny model. They agree, so the Traffic acquisition report is enough.
Google Ads plus organic searchData-driven, in GA4 and in Google Ads
Ads on both Google and MetaData-driven in GA4 as the referee, with every Meta ad link tagged
Checking GA4 against Google AdsPaid and organic last click, for the comparison only
Many customers come from AI answers or word of mouthAny model, plus "How did you hear about us?"

Two rules apply under every GA4 model. Direct visits, from a typed address or a bookmark, get no credit "unless the path to key event consists entirely of direct visits". And data-driven credit can still shift for a week: "Conversions can be reattributed for up to 7 days after the conversion" (Analytics Help: Get started with attribution).

Where does AI chat traffic show up in GA4?

Since May 13, 2026, GA4 puts clicks from AI chatbots in their own channel, AI Assistant, with no setup (Analytics Help: What's new in Google Analytics). Google's own AI answers stay out of it:

"AI Assistant is the channel by which users arrive at your site from sources like ChatGPT, Gemini, Deepseek, Copilot, or Grok. It excludes Google’s AI Overviews and AI Mode." (Analytics Help: Default channel group)

What the customer did decides who gets the credit:

What the customer didGA4 channelCredit for the lead
Clicked a link in a ChatGPT, Gemini, Copilot or Claude answerAI AssistantAI Assistant
Clicked a source in Google's AI Overviews or AI ModeOrganic SearchOrganic Search
Read an AI answer, then searched your nameOrganic SearchOrganic Search
Read an AI answer, then typed your web addressDirectDirect, only if no other click is on the path
Read an AI answer, then phoned youNot recordedNone

A visit lands in AI Assistant only when the site it came from matches Google's list of AI assistants (Analytics Help: Default channel group). Google doesn't publish the full list. Search Console doesn't split out Google's AI answers either: they "are included in the overall search traffic" (Google Search Central: AI features and your website).

Why AI answers get less credit than they earn

Every model splits credit only among the visits GA4 recorded. An AI answer someone reads, then acts on later, isn't one of them. OpenAI counts a local business as mentioned "even if the customer doesn’t open the listing or visit the website" (OpenAI Help Center: Setting up your business on ChatGPT).

Three ways to see that part:

  • Ask. Put "How did you hear about us?" on your booking form and at the start of every first call, with "ChatGPT or another AI assistant" as a choice. Log the answer with each lead.
  • Compare monthly. Set the answers beside GA4's channel counts. When more customers name an AI tool than GA4 credits to AI Assistant, the rest sit in Organic Search, Direct or nowhere.
  • Claim your ChatGPT listing. Business Hub shows mentions and website clicks "over the past 28 days", and OpenAI warns "Counts may be incomplete." Availability depends on location and business type; local stores and restaurants "can currently claim only an existing U.S. listing" (OpenAI Help Center: Setting up your business on ChatGPT).

How far back does credit reach?

The lookback window, which Google Ads calls the conversion window, is how many days after a click that click can still earn credit. Pick one longer than your customers take to decide.

Tool and settingChoicesDefault
GA4, most key events30, 60 or 90 days90 days
GA4, first visits (first_visit, first_open)7 or 30 days30 days
Google Ads, click-through conversions1 to 90 days30 days for Search and Display

So GA4 has no 7-day window for an ordinary lead. In both tools, a change applies only from the day you make it (Analytics Help: Select attribution settings; Google Ads Help: About conversion windows). Google Ads recommends "windows of at least 7 days". For most service businesses, leave GA4 at 90 days.

Why Google Ads, Meta and GA4 add up to more leads than you got

Each ad platform credits its own ads by its own rules, so their totals overlap. Google Ads counts leads after clicks, video engagements and views of Google ads (Google Ads Help: About conversion windows). Since March 2026, Meta counts only link clicks as click-through for website conversions, and files shares, saves and other non-link actions as "engage-through" (Meta for Business). Meta itself names the gap:

"Many third-party platforms, on the other hand, primarily attribute only to website link clicks." (Meta for Business: Simplifying ad measurement for a social-first world)

GA4 sees visits to your site, not someone scrolling past a Meta ad. Count each lead once, in your own records, and use GA4 to split them by channel. Never add the platforms' totals together.

Example

Illustrative cases with invented numbers.

A US roofing company on Google and Meta (US$)

It spends $3,000 a month. Google Ads reports 20 leads and Meta 12, a total of 32. The office logged 25 calls and forms. GA4's data-driven split of the same leads: Paid Search about 14, Organic Search 5, Paid Social 4, AI Assistant 2. The owner judges cost per lead on the office's 25, split GA4's way, not on 32.

A Toronto accounting firm and ChatGPT (C$)

It pays C$1,500 a month for SEO and listings. GA4 shows 1 lead from AI Assistant in March. On the intake form, 6 new clients said ChatGPT: 4 reached the site by searching the firm's name, 1 typed the address, 1 phoned. The intake answers, not GA4, show what the listing work is worth.

FAQ

Is GA4 enough, or do I need attribution software?

For a business with a few channels, GA4 is enough: Google offers it "free of charge" with data-driven attribution built in (Google Marketing Platform: Google Analytics). No software sees an AI answer nobody clicked, so you'd still need to ask customers.

Does Google's AI Overviews traffic count as AI Assistant?

No. Clicks from AI Overviews and AI Mode count as Organic Search in GA4, and as ordinary web search traffic in Search Console.

Do cookie banners break attribution?

Visitors who decline analytics cookies drop out of your paths. GA4 can estimate them only with at least 1,000 daily consenting users on 7 of the previous 28 days, plus other conditions (Analytics Help: Behavioral modeling for consent mode). A site with a few hundred visitors a day won't qualify.

Is Perplexity in the AI Assistant channel?

Google's descriptions of the channel name ChatGPT, Gemini, Claude, Deepseek, Copilot and Grok, not Perplexity. In Traffic acquisition, view visits by session source: an AI tool listed under Referral or Unassigned isn't being counted as AI Assistant.

Different for you?

Keep learning

NextMarketing analytics for small business: tools, metrics, habitsMarketing analytics for a small business: the free tools, the numbers that show which marketing brings paying customers, and a 30-minute monthly review.

Ask us one question

One line is enough. We reply by email.

Sources

  1. Analytics Help: Get started with attribution
  2. Analytics Help: Select attribution settings
  3. Analytics Help: Default channel group
  4. Analytics Help: What's new in Google Analytics (May 13, 2026: AI Assistant traffic)
  5. Analytics Help: Behavioral modeling for consent mode
  6. Google Ads Help: About attribution models
  7. Google Ads Help: About conversion windows
  8. Google Search Central: AI features and your website
  9. Google Marketing Platform: Google Analytics
  10. OpenAI Help Center: Setting up your business on ChatGPT
  11. Meta for Business: Simplifying ad measurement for a social-first world (March 3, 2026)