Validation:
dark traffic, incrementality and accuracy
This pillar has 4 lessons. Each one answers the questions buyers ask on one topic, with a visual, the numbers behind it and a checklist you can use this week.
Hidden influence: dark AI, zero-click
45%average fall in ChatGPT sessions in GA4 after the tag broke on Oct 8, 2026What is dark AI traffic?
Lesson 7.2 · 10 questionsTest if AI really drives revenue
0.70% vs 0.52%organic click rate for brands cited vs not cited in an AI OverviewHow accurate is AI search attribution?
Lesson 7.3 · 10 questionsHow attribution models work
23xhigher conversion rate for AI search visitors than organic search visitors on one siteFirst-touch attribution for AI search
Lesson 7.4 · 10 questionsSelf-reported and survey attribution
45%of B2B buyers used generative AI in a recent purchaseHow did you hear about us ChatGPT
What this pillar covers
Start with what analytics cannot see. Lesson 7.1 explains dark AI traffic, the AI visits that arrive with no source, and the zero-click answers that never send a visit at all. Lesson 7.2 shows how to test whether AI really lifts revenue, with a test group of pages and a control group.
Then check how you give credit. Lesson 7.3 compares attribution models and shows how much credit each one gives AI search. Lesson 7.4 adds a "how did you hear about us" question, which catches the buyers who read an AI answer and never clicked.
Hidden AI influence and lift, in four numbers.
average fall in ChatGPT sessions in GA4 after the tag broke on Oct 8, 2026
Relevant Audience, Oct 16, 2026pages that lost ChatGPT traffic also saw direct traffic rise
Adobe data shared by Cedric Huesler, Oct 2026organic click rate for brands cited vs not cited in an AI Overview
Seer Interactive, Sep 2025higher conversion rate for AI search visitors than organic search visitors on one site
Ahrefs, single site, last 30 days to Jun 2025Four steps to check what AI really drives.
- 1
Read direct traffic next to AI every week
Chart direct traffic beside your AI channel. When AI falls and direct rises on the same pages, some AI visits have lost their source. Lesson 7.1 shows the pattern.
- 2
Run a test on a group of pages
Change a test group of pages and leave a control group alone for 6 to 8 weeks, then compare the two. Lesson 7.2 has the setup.
- 3
Run two attribution models side by side
Run last click and first touch together for 90 days, so one model cannot hide the share AI earns. Lesson 7.3 explains why.
- 4
Ask buyers how they found you
Add a "how did you hear about us" question with AI tools as options and a free text field. Lesson 7.4 shows how to code the answers.
Common questions about dark AI traffic, lift and accuracy, answered.
Can you measure dark AI traffic exactly?
No. Dark AI traffic arrives with no source, so you can only estimate it. Read the direct channel next to your AI channel, check survey answers and label the result as an estimate.
How accurate is AI search attribution?
Report it as a range, not one number. Tagged and referred visits are counted, while AI-influenced revenue is estimated. Lesson 7.2 shows how to label each kind of evidence by strength.
Which attribution model is fair to AI search?
First touch plus a survey answer is the simplest fair setup. Last click usually gives AI search little credit, because buyers often meet an AI answer first and convert later through branded search or direct. Lesson 7.3 compares the models.