This guide shows you how to test ChatGPT ads without paying search prices for weaker tracked sales. An October 2026 working paper by Maximilian Kaiser and Christian Schulze, in PPC Land’s report on the study, found US paid ChatGPT visits returned $0.89 in revenue per session versus $2.45 for paid Google Search. Conversion in that comparison was 0.37% versus 1.17%, so a maximum bid tied to your own numbers matters before budget leaves campaigns that already convert.
Updated November 6, 2026
Key Takeaways
- US paid ChatGPT visits returned $0.89 versus $2.45 on paid search.
- Parity price is about $0.26; a known account paid $1.72.
- Paid ChatGPT converted at 0.37% versus 1.17% for search.
- Of 87 dual-channel buyers, 73 recorded zero ChatGPT ad sales.
What should you check before you test ChatGPT ads?
The paper followed 1,898 shops from June 29, 2025 to August 22, 2026. Paid results come from 102 sites with complete sales tracking in data from Grips Intelligence. Ads were labeled sponsored cards below the answer, shown to logged-in adults on Free and Go plans, not Plus, Pro, Business, Enterprise, or Education.
Of 87 US advertisers that bought both channels, 84 earned less per ChatGPT visit than on paid search. Tech Help Canada, citing the authors, says 98% converted worse. Seventy-three had zero sales across 18,726 visits; 14 advertisers made all 570 purchases, and one account drove 66%.
How do you price a test from the parity gap?
A parity price is the CPC at which a ChatGPT click matches paid search on tracked revenue per ad dollar. In 16 accounts with at least 80% US search traffic, median search CPC was $0.72, scaled to about $0.26. One advertiser held 76% of US paid ChatGPT revenue, so resampling spanned $0.05 to $0.57.
The largest known account paid $1.72 over 15 days reported by Opascope, more than six times parity. OpenAI’s ads basics page recommends a starting maximum bid of $3 to $5.
| Measure | Paid ChatGPT, US | Paid Google Search, US |
|---|---|---|
| Revenue per visit | $0.89 | $2.45 |
| Conversion rate | 0.37% | 1.17% |
Use the formula Search Engine Journal draws from the paper before you raise a maximum bid:
- Divide ChatGPT revenue per visit by paid-search revenue per visit on the same sites.
- Multiply by your search cost per click to get parity on tracked revenue per dollar.
- Compare that figure with the CPC you pay, with about $0.26, with $1.72, and with $3 to $5.
- Assume similar profit margins and count only the tracked visit.
The paper’s fact sheet warns, “ChatGPT’s ads don’t pay. Fixing them could cost ChatGPT its users.”
What should your measurement checklist cover?
- Last click. The data does not reveal prompts or impressions. Assisted conversions and later purchases may be missed.
- Device and week. On the same websites, conversion and revenue gaps remained after those controls.
- Review crawlers. OpenAI’s basics page says ad-review crawlers can visit landing pages with tracking parameters and be flagged as paid in third-party analytics.
- Halo traffic. There was no detectable increase versus similar sites that had not bought an ad, though only large lifts are ruled out.
- Attribution. Digiday reports that Accuracast’s group CEO Farhad Divecha said, “We continue to see discrepancies there.”
Keep unpaid visits separate. The unadjusted volume change was -5.9%, and that interval included zero. The authors call the 6% to 8% headline “a range across assumptions, not an interval.”
Conversion shifted by -0.02 percentage points from a 2.21% baseline, with no detectable change.
When does a small discovery budget still make sense?
A small test still makes sense if you label it research spend, not a substitute for search. Advertisers chose to buy, and the window closed August 22, 2026, before EU ads on August 24. A later footprint of 89 countries and more than 100,000 advertisers does not erase that early gap.
Do not mix other published scores into the session comparison. On August 31, OpenAI posted a three-times return for one ecommerce advertiser over 28 days, and DoubleVerify reported Weight Watchers’ attributed cost per acquisition 15.3% below its benchmark. If your revenue per visit and conversion do not approach search, keep the spend small, using the same caution as judging whether PPC deserves more than SEO or reviving a PPC campaign that is not converting.
Frequently asked questions
What revenue did paid ChatGPT visits produce versus paid search?
In US traffic, a paid ChatGPT visit returned $0.89 versus $2.45 for a paid Google Search visit, from 102 sites with full sales tracking. Conversion was 0.37% versus 1.17%.
How is the ChatGPT ads parity price calculated?
Divide ChatGPT revenue per visit by paid-search revenue per visit, then multiply by search CPC. From a $0.72 median search CPC, the authors estimated about $0.26, resampled from $0.05 to $0.57.
Is OpenAI’s $3 to $5 starting bid a match for the study?
The ads basics page recommends that range as a starting maximum for CPC campaigns. It sits well above the study’s about $0.26 parity price and the $1.72 paid by the largest known account.
Did organic ChatGPT referrals fall after ads launched?
The unadjusted volume change was -5.9%, and that interval included zero. The authors call the “6 to 8%” headline a range across assumptions, not an interval, while conversion showed no detectable change.
