AI Overviews break the old link between rankings and traffic. A page can hold position one and still lose clicks, because the AI summary at the top of the results answers the question first. That means a report built on rankings and sessions alone now tells a misleading story. The fix is to report three things side by side: visibility (including in AI answers), the clicks you still earn, and the leads and revenue those visits produce.
This guide is for agencies rebuilding client reports and for marketing leaders who need to explain to their boss why traffic dropped while rankings held.
Key takeaways
– Ahrefs’ first study (April 2025) found AI Overviews were linked to a 34.5% lower click-through rate for the top-ranking page.
– In Pew Research Center’s study, only 1% of visits to results pages with an AI summary included a click on a link inside the summary.
– Zero-click search isn’t new. SparkToro’s 2024 study found 58.5% of U.S. Google searches ended without a click.
– Search Console’s Generative AI performance report now shows impressions in AI Overviews and AI Mode by page.
– Lead with outcomes (leads, revenue), then visibility, then traffic. Rankings move to the appendix.
Why do rankings and traffic no longer move together?
Because the results page changed. When an AI Overview appears, it often answers the question well enough that the searcher stops there.
Ahrefs compared click-through rates for top-ranking pages on keywords with and without AI Overviews and found a 34.5% drop in its first study. Pew tracked real browsing by 900 U.S. adults and found that people rarely clicked links inside AI summaries at all.
Add the long-running zero-click trend, and you get a familiar client conversation: “We’re ranking better than ever. Why are leads flat?”
What should you stop leading with?
- Average position across all keywords. It hides which pages matter and ignores AI answers entirely.
- Total sessions. A drop in informational traffic may not affect revenue at all.
- Keyword counts. “We rank for [X] more keywords” says nothing about business value.
Keep them, but put them in the appendix. They’re diagnostic, not headline, metrics.
What should a modern SEO report measure?
Three layers, in this order.
1. Business outcomes
- Leads, sales or booked calls from organic search (GA4 key events and CRM).
- Pipeline or revenue influenced by organic, where your CRM can track it.
- Self-reported source (“How did you hear about us?”), which often catches AI influence that analytics misses.
2. Visibility
- Search impressions for priority pages (Search Console Performance report).
- AI Overview and AI Mode impressions by page (Search Console Generative AI performance report). It shows impressions only, with no clicks or queries, and data starts May 18, 2026.
- AI mention share: how often you’re named in a fixed set of buyer questions run monthly in ChatGPT, Perplexity, Gemini and AI Mode.
- Branded search demand: impressions and clicks for queries containing your brand name.
3. Traffic quality
- Organic clicks and CTR for priority pages.
- Engagement and conversion rate by landing page.
- AI assistant referrals in GA4 (sources such as chatgpt.com and perplexity.ai).
How do you explain a traffic drop without sounding defensive?
Show the pattern, then the business impact, then the plan.
Example wording:
“Clicks to our how-to guides fell [X]% this quarter, mostly on searches where Google now shows an AI summary. Our pages appeared in those summaries [X] times last month, so we’re still visible, just not always clicked. Leads from organic search held at [X]. Next quarter we’re focusing on pages closer to a buying decision, where people still click, and adding reasons to visit on the guides that matter most.”
Honest, specific and focused on what happens next. Clients don’t need a lecture on AI. They need to know the plan.
What does a good one-page report look like?
Illustrative layout (placeholders, not client data):
– Headline: “Organic search produced [X] leads this month ([up/down] [X]% vs. last month).”
– Outcomes: leads, sales, pipeline influenced.
– Visibility: Search impressions on priority pages; AI Overview/AI Mode impressions; AI mention share ([X] of [X] prompts); branded search trend.
– Traffic quality: clicks, CTR and conversion rate on the top 10 pages.
– What we shipped: [X] items, linked.
– Next month: three to five actions, each with an owner.
– Appendix: rankings, keyword counts, technical health.
Most executives read the headline and the next-month list. Make those two lines count.
How should you handle the Generative AI report in reports?
Use it for page-level insight, not a vanity total.
- Top pages by AI impressions. Which content does Google pull into AI answers? Pair each with clicks and leads from the Performance report and GA4.
- Important pages with no AI impressions. These are content upgrade candidates.
- Never add AI impressions to Web impressions. Google says AI feature data is already included in the Performance report’s Web totals.
- Note the limits. No clicks, queries or position; AI Overviews and AI Mode combined.
What about AI assistants outside Google?
Track them separately. GA4 can show referral visits from AI assistants if you build a custom channel group using a regex on session source. Expect undercounting, because some AI visits arrive without referrer data and land in Direct.
The monthly prompt check fills the gap. Pick 20–50 questions your buyers actually ask, run them in each assistant, and log whether you’re mentioned, linked or absent. AI answers vary between runs, so watch the trend over several months.
How often should you report?
- Monthly: outcomes, visibility, traffic quality, shipped work, next actions.
- Quarterly: trends, competitor comparison on AI mentions, budget and strategy decisions.
- Weekly: only for launches or problems. Weekly AI data is mostly noise.
What should businesses ask their agency to change?
If you’re the client:
- “Can the first line of the report be about leads or revenue?”
- “Which of our pages appear in AI Overviews, and are they producing anything?”
- “Are we mentioned when buyers ask ChatGPT about our category?”
- “Can rankings move to the appendix?”
- “What are the three most important things you’ll do next month?”
How do you set up the data in an afternoon?
Most of this runs on tools you already have. A one-time setup:
- Search Console: confirm the property covers all versions of the site (a domain property is simplest). Bookmark the Performance report filtered to your priority pages and the Generative AI performance report.
- Branded query filter: in the Performance report, create a query filter using a regex of your brand name and common misspellings. Save the comparison of branded vs. non-branded clicks.
- GA4 key events: make sure form submissions, calls and purchases are marked as key events, so organic and AI traffic can be tied to outcomes.
- GA4 AI channel group: create a custom channel group called “AI assistants” with a regex on session source covering chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com and claude.ai. Place it above Referral so those visits aren’t lumped in with other referrals.
- Self-reported source: add an optional “How did you hear about us?” field to your main forms, and pass it into the CRM.
- Prompt check sheet: a simple spreadsheet with 20–50 buyer questions in rows and one column per assistant per month. Log “mentioned,” “linked” or “absent.”
- Report template: one page with the layout above, plus an appendix tab for rankings and technical health.
After setup, a monthly report takes about an hour to pull and an hour to write.
How should agencies and in-house teams split the work?
Agencies should own the data pull, the analysis and the “what it means” section, because that’s where expertise shows. They should also own the next-month actions.
In-house teams should own the business data: CRM outcomes, sales feedback and the self-reported source field. They’re the only ones who can say whether organic leads are good leads.
The best reports are written together. A 15-minute call before the report goes out, where the in-house lead shares what sales heard that month, makes the “what it means” section far sharper.
What are the common mistakes?
- Celebrating rankings while leads fall. Clients notice.
- Blaming AI for everything. Some drops are seasonality, technical issues or competitors. Check before you explain.
- Double-counting AI impressions.
- Reporting AI mentions from one screenshot. Use a fixed prompt set over time.
- No next steps. Every report should end with actions.
Illustrative example: before and after
Illustrative example (placeholders, not client data):
Before: Page one shows average position improving from [X] to [X], [X] keywords in the top 10, and sessions down [X]%. The client asks why they’re paying for rankings that don’t bring leads.
After: Page one shows organic leads ([X], steady), AI Overview impressions on the top service pages ([X]), mention share in ChatGPT and AI Mode ([X] of [X] buyer questions), and three actions for next month. Rankings sit in the appendix. The conversation shifts from “Why is traffic down?” to “Which pages should we improve next?”
Frequently asked questions
Why is my traffic down if my rankings went up?
AI Overviews and other search features can answer the question on the results page, so fewer people click. Ahrefs found a 34.5% lower CTR for top-ranking pages when an AI Overview appeared in its first study.
Can I see AI Overview clicks in Search Console?
No. The Generative AI performance report shows impressions only. Clicks from AI features are counted inside the regular Performance report’s Web data but aren’t separated.
What should replace rankings in SEO reports?
Lead with leads and revenue, then visibility (search impressions, AI Overview impressions, AI mention share, branded demand), then traffic quality. Keep rankings as a diagnostic in the appendix.
How do I track ChatGPT traffic?
In GA4, filter or group sessions by sources like chatgpt.com and perplexity.ai. Some AI visits arrive without referrer data, so add a self-reported source field to forms.
How often should SEO reports include AI metrics?
Monthly for page-level AI impressions and mention share, with a quarterly review of trends and competitors.
Is zero-click search new?
No. SparkToro found 58.5% of U.S. Google searches ended without a click in 2024. AI Overviews have added to an existing trend.
Want reporting that tells the real story?
We rebuild SEO reporting for businesses and agency partners around outcomes and AI visibility, so every report ends with a clear next step.
Prefer to talk? Call (813) 397-3665 or email strategy@seooutsourcing.com.
