Most generative engine optimization (GEO) pitches treat Google's AI Overviews and ChatGPT as one channel. Each hits your lead flow differently. AI Overviews change how many people click through from a results page where you already rank, since more of the answer now sits on Google instead of your site. ChatGPT, by contrast, sends a smaller number of visits directly to your site. Measure each before you fund either, whether that means GEO work or paid placements such as Google's AI Max and ChatGPT ads.
Do AI Overviews cost you clicks?
Researchers at the Indian School of Business and Carnegie Mellon tracked 1,065 U.S. searchers for two weeks. When Google showed an AI Overview, clicks to outside sites fell by close to 40% compared with searchers who had Overviews hidden (Search Engine Journal). A later revision put the figure at 39.8% (PPC Land). It's a working paper, not peer reviewed, and it covers desktop Chrome users only, so treat the 40% as a range, not a forecast.
So what do you do with that? Ask two things about your own account: which of your lead-driving queries now trigger an Overview, and have leads from those landing pages moved since? Rank trackers report position, and position can hold while clicks drop.
Do ChatGPT referrals convert better?
The studies disagree. Visibility Labs looked at 94 ecommerce brands in 2025 and found ChatGPT referrals converting at 1.81%, against 1.39% for non-branded organic (Search Engine Land). A peer-reviewed study of 973 ecommerce sites in Marketing Science found organic had 13% higher odds of converting, and some of the authors' own robustness checks closed the gap.
To be clear, both studied ecommerce, where a conversion is typically a purchase, and ChatGPT referrals were a small slice of traffic in each. We feel this is important since lead gen has two conversions, the form completion (lead) and the close (sale), and a conversion rate says nothing about which one you're buying. A referral source can win on form completions and lose on closed revenue, or the reverse, and neither study can tell you which.
Here is where we feel the reports miss the target. Visibility Labs notes that people who get a brand recommended in ChatGPT often search for that brand on Google next, so the credit lands on branded organic. Last-click attribution undercounts ChatGPT by an amount nobody has measured yet (which is the best argument for measuring your own).
How do you measure both before you spend?
- Measure Overview exposure on your lead-driving queries. Flag the queries that trigger an Overview using your rank tracker's SERP feature data, then compare landing page leads for flagged queries against unflagged ones.
- Split ChatGPT, Perplexity, Gemini and Copilot referrals out in GA4. Build a custom channel group from those referral sources.
- Judge those visits on closed revenue, not solely form completions. Pass the channel into your CRM and compare lead-to-opportunity rate against branded organic. A cheap, unqualified lead that doesn't close proves nothing.
- Watch branded search and direct. If ChatGPT recommendations are working, some of the credit shows up there, even when the referral report doesn't show it.
- Set the threshold before you look. Decide what share of qualified leads would justify more investment, then check against it after a quarter.
Where does GEO fit into this?
Once you know what these visits are worth, you can size a GEO or paid placement budget instead of guessing. Start with the pages that answer the questions your buyers ask before they talk to sales.
Bottom line: measure first, then fund. We set up the measurement and tell you plainly whether GEO is worth funding for your business. If you want to see a report on your own URL, email us or schedule a call.