OpenAI doesn’t publish benchmarks by industry, so the numbers out there come from independent tests, vendors selling the channel, or OpenAI’s own customer stories. Don’t average them. Taken together: click rates of 0.5% to 1%, $3 to $5 a click for consumer offers, consumer sign-ups that convert, small B2B tests that didn’t, and no public conversion numbers for insurance yet.
Each number below carries one of three labels. Independent: someone ran a campaign and published the numbers, or an audience-measurement firm published panel data. Vendor: an agency or tool that sells the channel reported its own result. OpenAI: a customer story OpenAI published. Use the ranges to size a test, not to forecast.
Click rate
| Figure | Who | Label |
|---|---|---|
| 0.68% overall, top quartile 1%, best brands 1.57% | Similarweb panel data, May 2026 | Independent |
| 0.73% in the US through September 16; 0.60% (New Zealand) to 1.91% (Brazil) by market | Similarweb and Graphite, relayed by Lapis | Independent, second-hand |
| 1.1% overall; 0.9% on the financial-services campaign | Workshop Digital, 69,630 impressions | Independent |
| Under 1.25% median, range 0.35% to 7.9%, 24 campaigns | PPC One compilation | Mixed |
| Google Search 6.64%, Meta 1.71%, for comparison | WordStream cross-industry averages | Independent, other platforms |
Plan on 0.5% to 1%. A higher click rate means a good card, not a good campaign. The quote form decides whether the campaign works.
Cost
| Figure | Who | Label |
|---|---|---|
| CPM about $60 at launch, $25 to $45 nine weeks in | Digiday, buyer interviews, April 2026 | Independent, trade press |
| CPM $64.34 on one B2B account; above $65 for Hostinger | Search Engine Journal roundup, September | Independent, editorial |
| Cost per click $3.14 (B2B), $3.44 on financial services | Workshop Digital | Independent |
| Cost per click about $9, after a $12 max bid; $3 bid got no impressions | Symphonic Digital, B2B | Independent |
| Median $4.41, range $0.76 to $14.80, 15 campaigns | PPC One compilation | Mixed |
| $3 to $22 across advertisers | Search Engine Journal roundup, September | Independent, editorial |
| Earlier starting-bid advice: $3 to $5 per click | OpenAI, for the old Max CPC setting | Old advice, not a measured result or a suggested Target CPC |
No public click cost exists for insurance. Workshop’s $3.44 came from a B2B agency’s campaign aimed at financial-services audiences, not insurance buyers.
Conversions
| Figure | Who | Label |
|---|---|---|
| 53 clicks, 1 consultation click, 0 leads on about $500 | Symphonic Digital, B2B | Independent |
| 739 clicks, 0 qualified leads on $2,319; Google Search 2 in the same period | Workshop Digital, B2B | Independent |
| About 40 clicks, 0 conversions on about $200 | adlibrary.com, two-week test | Independent |
| 2.35% conversion rate, $1.72 a click, 1.49x return over 15 days | Opascope, one revenue-tracked account, its own tracking | Vendor |
| Shoppers referred from AI chatbots like ChatGPT, not from ads, converted about 1.5x other referral traffic, across its US clients | Criteo, OpenAI’s ad tech partner, March 2026 | Vendor |
| Cost per customer 15.3% below paid search | WeightWatchers via DV Rockerbox | OpenAI story |
| Cost per lead down 34% after conversion bidding; return up 38% quarter on quarter | Angi | OpenAI story |
| 10%+ of clicks signed up; 60% cheaper than social, 30% cheaper than search | Jotform | OpenAI story |
| About 40% started a document, about 8% completed one | LegalNature | OpenAI story |
| 2.3x the orders last-click reporting showed; 67% of the extra purchases came from new customers | Dose, measured by WorkMagic | OpenAI story |
| 93% of visitors new | Portland Leather | OpenAI story |
| About 2x return per $10k test group; 70%+ of clicks became sessions | Stream | OpenAI story |
| 3x return, 7x in a sale | Newegg | OpenAI story |
Consumer sign-ups and purchases convert. Three small independent B2B tests found no leads. B2B software stories from OpenAI did better.
Conversion bidding arrived in June, after Symphonic’s test ran. Workshop does not say which objective it used. OpenAI’s Angi story covers the switch from clicks to conversions.
What exists for insurance
No independent conversion number has been published for insurance yet. Four public sources come close, and none of them measures conversions.
- An Australian vendor, Somantra, reported ads on 71% of 34,279 insurance queries run on two days in August, on desktop. That is an ad-frequency figure from one market, not a performance figure.
- Nakora, citing Lapis, gives finance and insurance a click rate of 0.5% to 1.1%, second lowest of seven verticals. These are early-advertiser estimates with no data shown.
- First Page Sage’s table of conversion rates by industry (commercial insurance 1.9%, financial services 0.2%) is a projection. Several guides quote it as measured. It is not.
- Workshop’s financial-services campaign: 0.9% click rate, $3.44 a click, no leads, B2B.
Until insurance numbers are published, plan a 14-day, $350 test on one line in one state, and grow it slowly if it pays back.
Planning assumptions for a test
For a US consumer line in late 2026, drawn from the figures above:
| Measure | Plan on | Worry if |
|---|---|---|
| Impressions per $1 | 25 to 40 | Under 15 after day three (check what impressions cost) |
| Click rate | 0.5% to 1% | Under 0.3% after day seven (card or hint mismatch) |
| Cost per click | $3 to $5 | Over $8 (B2B pricing on a consumer line) |
| Clicks that become sessions | 65% to 80% | Under 50% (page speed, bots, tracking gaps) |
| Clicks that start a quote | 2% to 5% | Under 1% after day 14 (page, not ad) |
| Cost per started quote | Roughly $60 to $250 | Over $250 after day 14, or above your Google cost per started quote |
At $3 to $5 a click, $350 buys 70 to 117 clicks. At 2% to 5% starting a quote, that is 1 to 6 quotes. That is enough to prove the pixel, the server events, the real cost per click and the quote form work, but not enough to pick a winning ad. To compare two ads you want about 30 started quotes each, which at these rates means 600 to 1,500 clicks per ad. Fix the plumbing first, then fund the comparison.
How to grow a test that works
- Do not make big changes to a campaign that pays back. OpenAI advises waiting for enough results before large changes to bids, budgets or creative. To change a campaign’s objective or conversion event, you have to start a new campaign. Test new ads in a separate ad group.
- Raise the budget about 20% a month, not double. Before increasing it, check whether the campaign is spending what it already has. If it isn’t, review the target, conversion data and available chances to show the ad first. More budget may not fix those limits. After each increase, compare spend, started quotes and cost per started quote to see whether the extra spend is paying back.
- Expect costs to move. CPMs fell from about $60 at launch to $25 to $45 by April, and some advertisers paid $64 or more in September. Recheck cost per started quote every month.
- Read each month against its season. Insurance demand moves with the calendar: renewals, moving season, travel season. ChatGPT ads have no year-ago data, so use last year’s same weeks in your Google Ads and quote-form numbers as the yardstick.
FAQs
Why not average the numbers into one benchmark?
Because a Similarweb panel figure, an agency’s $500 test and an OpenAI customer story measure different things on different dates. An average of them is a number nobody observed.
Sources, checked October 27, 2026:
- Search Engine Roundtable, ChatGPT ads click-through rates (Similarweb, May 2026)
- Lapis, ChatGPT ads conversion rate benchmarks (Similarweb and Graphite figures; First Page Sage projections, labeled as such)
- Workshop Digital, first B2B test
- Symphonic Digital, field report
- PPC One, ChatGPT ads results
- Digiday, ChatGPT ads are getting cheaper
- Search Engine Journal, six months into ChatGPT ads
- Opascope, ChatGPT ads benchmarks
- Nakora, ChatGPT ads conversion rate benchmarks (the finance and insurance click rate, citing Lapis)
- Criteo, joins OpenAI advertising pilot in ChatGPT (March 2026; AI chatbot referrals)
- OpenAI customer stories
- OpenAI ads blog (WeightWatchers, Dose, Portland Leather measurement results)
Bidding source, checked November 5, 2026:


