Uncategorized

How Much Should You Invest in AI Search? What Two 2026 Traffic Studies Say About Sizing the Budget

Every week brings a new claim that AI search will replace Google, or that GEO (Generative Engine Optimization) is a distraction. Neither claim helps you set a budget. Two October 2026 analyses of real analytics data suggest AI assistants still send a small share of visits. Size AI search work against measured traffic and leads, not headlines.

Key Takeaways
  • Marie Haynes analyzed 12 months of GA4 data from 100 sites and found ChatGPT sent 0.62% as many sessions as Google organic search.
  • SEO Works, a UK agency, found organic search sent 39 times more traffic than AI assistants across 95 sites from September 2025 to August 2026.
  • AI referrals often convert at a higher rate, but the gap is inconsistent: in the UK data, AI converted worse than organic on 23 of 63 sites.
  • Traffic is not the only value of being cited, so treat AI search as a measured, capped line item rather than a rebuild of your SEO plan.

Scope note: both studies come from practitioners, not peer-reviewed research, and we read the authors’ own published write-ups. Treat the figures as directional benchmarks, then check them against your own analytics.

What the 100-site analysis measured

In October 2026, SEO consultant Marie Haynes published an analysis of 12 months of Google Analytics 4 (GA4) data from 100 sites, covering more than 145 million sessions. She reported that ChatGPT sent 0.62% as many sessions as Google organic search, or 0.73% once other large language model (LLM) referrals were counted.

She also found that ChatGPT referral growth plateaued around August 2025 and has declined since. On conversions, only 55 of the 100 sites tracked them. Within those, ChatGPT visitors converted at 13.8% against 3.02% for Google organic, yet Google organic produced 964 conversions to the AI assistants’ 36.

The implication for a buyer is that a higher conversion rate on a tiny volume still produces few leads. Haynes herself calls it “my small dataset” and notes that 45 sites were not tracking any key events, which limits the conversion comparison.

What the UK benchmark measured

SEO Works, a UK agency, published a benchmark of 95 GA4 properties tracked monthly from September 2025 to August 2026. Organic search delivered 2,825,594 sessions and AI assistants delivered 72,095, which the agency summarizes as organic sending 39 times more traffic. AI assistants equalled 2.6% of organic sessions.

Its conversion comparison used 63 sites. AI referrals converted at 7.18% against 6.08% for organic, a pooled ratio of 1.18 times. However, AI converted worse than organic on 23 of those 63 sites, and for the 10 B2B product businesses the ratio was 0.62 times.

The agency also estimated that a 5% lift in organic search adds 10.3 conversions per site per month, while doubling AI referral traffic adds 6.6. For a team choosing where the next hour of effort goes, that is a direct comparison worth running on your own numbers.

The two studies side by side

ItemHaynes (October 2026)SEO Works benchmark
Sample100 sites, 145M+ sessions, 12 months95 UK GA4 properties, Sept 2025 to Aug 2026
AI traffic vs organicChatGPT at 0.62% of Google organic; 0.73% for all LLMsAI assistants at 2.6% of organic sessions
AI conversion rate13.8% vs 3.02% (55 sites tracking)7.18% vs 6.08% (63 sites compared)
Main cautionSelf-described small dataset; 45 sites tracked no key eventsAI Overviews excluded, so the AI figure is a floor

What neither study can tell you

The first limit is measurement. SEO Works notes that Google AI Overviews appear as ordinary organic sessions in GA4, so its AI figures are a floor. Some AI referrals also arrive without a referrer, for example from in-app browsers, and show up as direct traffic.

The second limit is that sessions are not influence. A buyer can see your brand in an AI answer, then search for you by name or type your address, and that visit is never credited to the assistant. Neither study measured that path, so neither can say it is small.

The third limit is time. Haynes found ChatGPT growth stalled, while SEO Works reported AI referrals rising from May 2026 and treats the cause as a working hypothesis. Both can be true, which is why a one-time benchmark should not set a two-year plan.

How to size the budget

None of the sources prescribes a budget split, so the steps below are our recommendation. They keep AI search work proportionate while preserving the option to scale up if your own data shows demand.

  1. Pull 12 months of GA4 data and calculate AI-assistant sessions as a share of organic sessions for your own site.
  2. Confirm that key events or lead forms are tracked, because a conversion comparison needs them. Our guide to tracking AI referral traffic covers the setup.
  3. Fix shared foundations first: crawlable pages, clear service descriptions and accurate business details help both search and AI answers.
  4. Cap dedicated AI-search work at a fixed share of search effort, and review it quarterly against leads, not citations alone.
  5. Add a branded-search and direct-traffic check, using the method in our post on measuring AI search visibility.

Be wary of any vendor that quotes a single AI-visibility score without showing the traffic and leads behind it. Our checklist for evaluating an agentic marketing vendor lists what to ask for. For the conversion side of this debate, see what three studies found on AI search conversion.

Frequently Asked Questions

Is GEO worth investing in if AI assistants send so little traffic?

It is worth a measured, capped investment. The two studies show small referral volumes, but they do not measure brand influence, and the shared foundations of good SEO and GEO benefit both channels.

Do AI search visitors convert better than Google visitors?

On average in both studies, yes, but not reliably. SEO Works found AI converted worse than organic on 23 of 63 sites, and B2B product sites converted at 0.62 times the organic rate.

How do I measure AI search traffic on my own site?

Segment GA4 sessions by referring domain, such as chatgpt.com, then compare them with organic sessions and conversions. Remember that AI Overviews clicks appear as Google organic, so your AI total will be a floor.

The right question is not whether AI search matters, but how much of your next quarter it deserves. At Grow with Hafsa we help US and UK teams baseline their own AI referral data and tie it to leads. That work sits within our AI marketing and agentic SEO services, so budget follows evidence instead of the news cycle.

For a worst-case stress test of click-based traffic, see what a 1,100-user forced AI Mode experiment found and how to use it in planning.

Related reading: before sizing a budget, segment your own traffic by funnel stage. This look at a Digiday survey shows why upper- and lower-funnel losses may differ.