PlaybookRU

Playbook9 min read

How to measure a blog when AI answers stop people from clicking

Judge a blog by four readings together: AI assistant visits in GA4, AI impressions in Search Console, branded demand and the leads who say where they found you. Organic clicks alone now undercount the work, because most searches end without a visit.

Worried crane holds its head beside a falling traffic chart
A falling traffic line tells you less than it did, so add three other readings

Page views alone no longer show whether a blog works. About 58.5% of US Google searches end without a click to the open webmarket, and an AI Overview cuts the click-through rate of the top result by about 58%market. Measure a blog with four readings together: AI assistant visits, AI impressions, branded demand and lead questions.

Why do page views stop telling you whether a blog works?

Because the answer now appears on the results page, so the reader often never visits. Pew tracked 68,879 Google searches from 900 US adults and found that 8% of visits with an AI summary included a click on a regular result, against 15% on pages without onemarket. Ahrefs, working from a different dataset, reports about 58% fewer clicks on the first result when an AI Overview is presentmarket.

The decline is not limited to AI summaries. SparkToro and Datos measured 58.5% zero-click searches in the USmarket, and Bain's survey puts the share of searches that end without another destination near 60%market. A blog can therefore be read, quoted and remembered while its session count stays flat. Mentions on other sites matter for the same reason, and we looked at whether guest posts get cited by AI.

This changes what a report must contain. A traffic chart on its own cannot separate a blog that stopped working from a blog that works where nobody clicks. A reader who needs a vendor to produce that report can filter out the rest of the agencies by asking which of the four readings below they will deliver.

What can you still measure in Google Analytics?

You can count visits from AI assistants, though the count has a hole in it. In May 2026 Google Analytics added a default AI Assistant channel that sets the medium to ai-assistant for recognized chatbots such as ChatGPT, Gemini and Claudemarket. Those sources no longer need a manual regex.

The gap is the referrer. Google Analytics can only classify a visit when the click carries a referrer, and visits from copied links, mobile apps and in-app browsers can still land in Directmarket. Google has not published the full list of recognized assistants either, so Perplexity and Copilot may still need a custom channel groupmarket.

Treat the AI Assistant channel as a floor. A count of 40 sessions in a monthestimate means at least 40 visits came through a tracked link, and the true number is higher. A paid AI visibility tracker can add a second count, though the free reports cover the first benchmark.

Businessman reviewing printed charts at a desk in an office
The AI channel is a minimum count, so read it beside other signals

How big is AI traffic, and does it convert?

AI referral traffic is small and converts well. SE Ranking, which analysed 101,574 websites, puts it at 0.32% of all visits in 2026market. Orbit Media's study of 97 B2B sites and 29 million visits measured 0.5%market.

Quality is the other half of the picture. Orbit Media found a 1.91% lead conversion rate for AI visitors against 0.71% for organic searchmarket. Seer Interactive measured 15.9% for Perplexity visitors against a 9.77% organic baselinemarket.

ReadingWhere it livesWhat it showsWhat it misses
AI Assistant sessionsGoogle Analytics default channelVisits that carried a referrerCopied links and app visits
AI impressionsSearch Console AI reportHow often your URLs appearedClicks and queries
Branded searchesSearch ConsoleDemand for your nameThe cause of the demand
Lead questionForm or call scriptWhere a buyer says they found youBuyers who forget or guess

Track a small channel that converts well per lead. Five leads from assistants can outweigh five hundred organic sessionsestimate, and a scorecard sorted by sessions would hide them.

What does the Search Console AI report show?

It shows how often your pages appeared in AI answers. Google announced the generative AI performance reports on June 3, 2026, with impressions in AI Overviews and AI Mode broken down by page, country, device and datemarket.

The report lists impressions and leaves out clicks, click-through rate, queries and average positionmarket. Search Engine Journal calls it a diagnostic lens, not a scoreboard. An AI impression also counts differently from a classic one, because the supporting link may sit among many citations or stay hidden until the answer is expanded.

Read the report page by page. Pages with rising AI impressions and flat clicks are the ones assistants lean on, and those deserve the first update when you refresh old posts. A brief that gets usable drafts keeps that rewrite cheap.

The same logic applies to category pages on a shop, where one page can earn more citations than a dozen posts.

How do you count the leads that arrive without a referrer?

Ask them. Add one required free-text field to the contact form, "How did you hear about us?", with no dropdown, and read the answers each month. Refine Labs used exactly this field on its strategy-call form and compared the answers with its CRM attributionmarket.

Add the same question to the first call script, so leads who phone or message you are counted too. When the writing is outsourced, run one paid test first and ask for this scorecard from it. A tally of answers that name ChatGPT, Perplexity, Google's AI answer or a blog post gives you the number that Google Analytics cannot.

Pair the answers with branded demand. Search Console can filter branded queries, so a month in which the AI impressions rise and the branded impressions follow is a pattern worth recordingestimate. Keep the claim modest in the report, since two lines rising together prove a correlation only.

The monthly routine takes less than an hour:

  1. Export AI Assistant sessions and conversions from Google Analytics.
  2. Export AI impressions by page from Search Console.
  3. Export branded impressions and clicks from Search Console.
  4. Tally the form and call answers.
  5. Write one sentence per reading, and one decision.
Person reviewing analytics on a laptop outdoors in winter
One person, one hour, four readings is enough for a monthly review

Frequently asked questions

Is organic traffic still a good measure of blog performance?

Only as one reading among several. Pew recorded 8% of visits clicking a regular result when an AI summary appeared, against 15% without onemarket, so flat sessions can hide growing visibility.

How much of my traffic comes from AI assistants?

Expect well under one visit in a hundred. SE Ranking measured 0.32% across 101,574 sitesmarket, and Orbit Media measured 0.5% across 97 B2B sitesmarket. Your own AI Assistant channel in Google Analytics gives the exact figure for your site.

Do AI impressions in Search Console count as traffic?

No. They count how often your URLs appeared in an AI answer, and Google's report carries no clicksmarket. Use them as a visibility signal, then check conversions in Google Analytics for the visit itself.

Do AI visitors convert better than organic visitors?

In the studies we found, yes. Orbit Media measured 1.91% against 0.71%market, and Seer Interactive measured 15.9% against 9.77%market. Both samples lean towards business-to-business sites, so a local service may see different ratios.

Why do some AI-driven leads show up as Direct?

Because Google Analytics needs a referrer to label a visit. Copied links, mobile apps and in-app browsers can strip it, and those visits land in Directmarket. A lead-source question on the form recovers part of what the referrer lost.

Sources

  1. Pew Research Center, Google users are less likely to click on links when an AI summary appears: 68,879 searches from 900 US adults, 8% versus 15% clicks, March 2025 data, July 22, 2025. https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/
  2. Ahrefs, AI Overviews reduce clicks, update: 300,000 keywords, about 58% lower position-one click-through rate, December 2025 data, February 2026. https://ahrefs.com/blog/ai-overviews-reduce-clicks-update/
  3. SparkToro and Datos, 2024 zero-click search study: 58.5% zero-click searches in the US, September 2022 to May 2024. https://sparktoro.com/blog/2024-zero-click-search-study-for-every-1000-us-google-searches-only-374-clicks-go-to-the-open-web-in-the-eu-its-360/
  4. Bain & Company, Goodbye clicks, hello AI: about 60% of searches end without the user moving to another destination, 2025. https://www.bain.com/insights/goodbye-clicks-hello-ai-zero-click-search-redefines-marketing/
  5. Search Engine Journal, Google Analytics adds AI Assistant as default channel group: medium ai-assistant, ChatGPT, Gemini and Claude, referrer limits, May 14, 2026. https://www.searchenginejournal.com/google-analytics-adds-ai-assistant-as-default-channel-group/574974/
  6. MarTech, GA4 now tracks AI chatbot traffic automatically: AI Assistant channel, unpublished referrer list, Direct masking, May 19, 2026. https://martech.org/ga4-now-tracks-ai-chatbot-traffic-automatically/
  7. Orbit Media, AI traffic conversion rates: 97 B2B websites, 29 million visits, 0.5% AI share, 1.91% versus 0.71% conversion, GA4 classification limits, July 2025 to June 2026. https://www.orbitmedia.com/blog/conversion-rates-ai-search/
  8. SE Ranking, AI traffic research study: 101,574 websites, AI referrals 0.32% of visits in 2026. https://seranking.com/blog/ai-traffic-research-study/
  9. Seer Interactive, The AI search landscape beyond the SEO vs GEO hype: Perplexity 15.9% versus organic 9.77% conversion, September 22, 2025. https://www.seerinteractive.com/insights/study-the-ai-search-landscape-beyond-the-seo-vs-geo-hype
  10. Google Search Central blog, Introducing Search Generative AI performance reports in Search Console: impressions, pages, countries, devices and dates, June 3, 2026. https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports
  11. Search Engine Journal, Google reports AI search impressions and how to read them: no clicks, queries or position, impression definitions differ, August 4, 2026. https://searchenginejournal.com/google-reports-ai-search-impressions-how-to-read-them/582824
  12. Refine Labs, Attribution mirage: a required free-text "How did you hear about us?" field compared with software attribution, 620 conversions over 12 months. https://www.refinelabs.com/article/attribution-mirage

Check your site in a minute

Paste your address: we'll show your score out of 100, whether an AI assistant names you, and which searches your site has no page for. Free.