ChatGPT Ads Earned $0.89 a Visit vs $2.45 for Google Search

What happened

A working paper by Maximilian Kaiser and Christian Schulze, covered by Search Engine Journal, compares paid ChatGPT traffic with paid Google search traffic in the US. A ChatGPT ad visit produced $0.89 in revenue. A paid Google search visit produced $2.45.

Conversion followed the same pattern: 0.37% for ChatGPT ad visits against 1.17% for paid search. The gap held when the authors compared both channels on the same websites and controlled for device and week.

The paid-ad data covers 102 advertiser sites with full sales tracking. Of the 87 advertisers that bought both channels in the US, 84 earned less per ChatGPT visit than per search visit.

The results are concentrated

The numbers are lopsided. Among those 87 advertisers, 73 got zero sales from ChatGPT ads, across 18,726 visits. The other 14 generated all 570 ChatGPT ad purchases.

One advertiser accounted for 66% of all purchases across the 102 sites. Remove that advertiser and the gaps get wider. So the average looks better than the typical experience.

What a ChatGPT click is worth

The authors use a "parity price": the cost per click at which a ChatGPT click would return the same revenue per dollar as a paid search click. Using 16 accounts with a median search CPC of $0.72, they estimate about $0.26 per ChatGPT click. Because one advertiser dominates the revenue, resampling gives a range of $0.05 to $0.57.

Compare that with the publicly known $1.72 CPC, reported by Opascope from one account's 15 days of spend. OpenAI's own guidance suggests starting CPC campaigns at a maximum bid of $3 to $5. On this evidence, those bids look expensive.

Organic ChatGPT referrals dipped

The paper also looked at unpaid referrals from ChatGPT answers to 1,898 sites, as a rough test of whether ads made the product less useful. Over the first 13 weeks, organic referrals fell an estimated 5.9% relative to the EU, where there were no ads. The range runs from -12.6% to +1.2%, so zero is possible. After adjusting for faster ChatGPT adoption in North America, the estimate becomes -7.2% to -8.4%. The authors describe a moderate decline of roughly 6% to 8%.

Conversion on the organic visits that did arrive did not change in any detectable way.

Read it with caution

This is a working paper under review, and it has real limits:

• Advertisers chose to buy ads, so the results show what they got, not proof the ads caused it.
• It uses last-click data, with no view of prompts, impressions or subscription tiers.
• The pre-launch trend test was shaky. Four of five rollout groups failed it individually.
• The data ends Aug. 22. OpenAI has since expanded to 89 countries and says more than 100,000 advertisers have run campaigns.

Other reported results look better, but they use different metrics. OpenAI cited an ecommerce advertiser with a 3x return on ad spend over 28 days. DoubleVerify reported a Weight Watchers cost per acquisition 15.3% below its benchmark.

What to do now

Calculate your own parity price. Divide ChatGPT ad revenue per visit by paid search revenue per visit, then multiply by your search CPC. That gives you a ceiling for ChatGPT bids on tracked revenue. It ignores margins and later purchases.

Clean your data first. OpenAI says ad-review crawlers can visit landing pages with tracking parameters, and analytics tools may count them as paid sessions. Filter them out before judging performance.

Test small and set a stop rule. With zero sales on most accounts, set a spend cap and a visit threshold before you start, and don't begin at $3 to $5 because a help page suggests it.

Keep working on organic visibility. The organic dip is modest and uncertain, but ChatGPT referrals are not guaranteed to grow now that ads are in the product.

Watch the ad format. Ads currently appear as labeled cards separate from answers. OpenAI says it is exploring more native ways for businesses to interact with consumers. Native placement could raise click value, but it could also raise user distrust.