Waftr.com
Analyst reading an AI search analytics report on a tablet by a window
Case study · US SaaS · AI search analytics

AI search analytics case study: from 30-40 to ~350 AI visits a day

In this AI search analytics case study, a US SaaS company's organic clicks fell for branded and non-branded terms while impressions held. Search Console and GA4 analysis pointed to AI Overviews, and after about 3 months of work, visits from AI platforms reached about 350 a day, up from 30 to 40.

Two measured pointsAI platform visits a day, GA4
Start30-40/day
After about 3 months~350/day
Offer: AI visibility analyticsA Tracking Care module
The measurement timeline

Step by step: what we set up, and when each signal became visible

Five steps in the order they happened. Two carry a measured number: the starting point and the reading after about 3 months. Nothing in between was published, and nothing in between is drawn here.

  1. StartSearch Console

    Branded and non-branded queries, split apart

    What was set up

    Queries grouped into branded and non-branded in Search Console, with clicks and impressions read side by side for each group.

    What became visible

    Clicks falling sharply in both groups while impressions held. The company was still being shown in search; fewer people clicked.

  2. StartGA4

    AI platforms as their own traffic source

    What was set up

    Visits referred by AI platforms such as ChatGPT, Gemini and Claude separated from ordinary referral traffic in GA4. Our guide to checking AI traffic in GA4 walks through the method.

    What became visible
    30 to 40AI platform visits a day: the starting point
  3. DiagnosisSearch ConsoleGA4AhrefsSemrush

    AI Overviews as the likely cause

    What was set up

    The Search Console and GA4 views read against Ahrefs and Semrush data, with branded and non-branded terms kept apart.

    What became visible

    Searchers were reading the answer on the results page. Google AI Overviews were the likely cause, an effect seen across the industry, and the goal became citations in AI answers as well as clicks.

  4. DeliveryWeb teamSEO teamBrand team

    The roadmap, shipped by the client's teams

    What was set up

    Short answers and schema on about 120 pages, crawl fixes, a rewritten sitemap and robots file, and an llms.txt file. The content team wrote to a structure we suggested.

    What became visible

    Each change could be read against the same Search Console and GA4 views from steps 1 and 2.

  5. After about 3 monthsGA4Google results

    The second reading

    What was set up

    The same GA4 view as the starting point, read again once the roadmap had shipped.

    What became visible
    ~350AI platform visits a day, up from 30 to 40

    The company also appeared in AI Overview citations, and organic search traffic improved slightly.

Division of work

Who shipped what on the AI search roadmap?

We did the analysis, wrote the roadmap and measured the effect in Search Console and GA4. The client's web, SEO and brand teams made the changes to the site.

Waftr

Analysis and measurement

  • Search Console split of branded and non-branded queries
  • AI platform visits separated in GA4
  • The diagnosis, and the roadmap the teams worked from
  • The second reading after about 3 months
Client teams

Delivered on the site

  1. Short answers and schemaOn about 120 pages
  2. Crawl fixesSo bots reach the pages that matter
  3. Sitemap and robots fileRewritten after the crawl analysis
  4. llms.txtA plain text guide to key pages

“Thanks to Waftr, we are completely on top of what is happening on Google and on the AI LLM websites. They did all the analytical work and did a great job bringing LLM traffic to our site.”

VP of MarketingUS-based SaaS company (name withheld)
Short answers

What teams ask about AI search traffic

AI traffic in GA4, AI Overviews, llms.txt and timing, answered from this case.

How do you see AI traffic in GA4?
Create a custom channel group in GA4 and add a channel whose rule matches AI platform referrers, such as chatgpt.com, gemini.google.com and claude.ai, then move it above Referral. Visits from those sources then report as their own channel. Clicks from AI apps that pass no referrer still land in Direct.
Do AI Overviews reduce organic clicks?
They can. When Google answers a query in an AI Overview, fewer searchers click through, and Search Console shows impressions holding while clicks fall. This company saw that pattern for branded and non-branded queries alike, which is why the work aimed at citations in AI answers as well as clicks.
What is llms.txt?
llms.txt is a plain text file at the root of a site that lists its key pages with short descriptions for large language models. It is a proposed convention with no confirmed effect on rankings. Here it was one roadmap item, next to schema, crawl fixes, a sitemap and a robots file.
How long does AI search traffic take to grow?
In this case about 3 months passed between the starting point, 30 to 40 AI platform visits a day, and the reading of about 350 a day. Timing depends on how fast your teams ship changes and how often AI platforms recrawl the pages, and one case sets no schedule for another site.
Reading the number

What the 350-a-day figure covers

The published figure is AI platform visits measured in GA4. Knowing its edges tells you how to read your own AI traffic before anyone compares it with this case.

Counted

Sessions that reach GA4 with an AI platform as the referrer: ChatGPT, Gemini, Claude and others. Both published numbers come from this series.

Missed

Clicks from AI apps and browsers that pass no referrer. GA4 files them under Direct, which makes the reported count a floor for real AI visits.

Never in GA4

Answers that cite or name the company without sending a click. Those surface as AI Overview citations and in manual prompt checks, never as sessions.

One company, one window: a measured result, not a forecast for another site.How AI visibility analytics tracks all three
Next step

Get a read on your own AI search traffic

Send the site and a line on what Search Console shows. We reply within one business day with how we would separate AI traffic in your GA4 and which views to watch first.

Enter your name.
That email looks incomplete.
Prefer to talk it through? Book 30 minutes