Stratezik, Toronto

ChatGPT Ads Benchmarks 2026: Real Costs, CTR, and What Actually Converts

What ChatGPT Ads actually cost and convert in 2026. OpenAI publishes no official benchmarks, so we collect the real, sourced numbers: CTR, CPM, CPC, ROAS, and by-industry estimates, each labeled by source.

Shah Md. Rifat
By Shah Md. Rifat
Updated 2026-09-01
ChatGPT Ads Benchmarks 2026: Real Costs, CTR, and What Actually Converts

The honest short answer: there are no official ChatGPT Ads benchmarks. OpenAI publishes no cross-advertiser CTR, CPC, conversion, or ROAS figures. The credible public numbers come from one analyst (Similarweb) and a handful of real-spend campaigns. As of mid-2026, industry CTR sits around 0.5 to 0.7 percent and is falling as inventory grows, CPMs clear roughly $25 to $45, CPCs run a few dollars, and the one fully documented campaign returned 1.49x ROAS at a $1.72 CPC. Treat all of it as directional on a platform barely seven months old.

We run ChatGPT Ads for ourselves and clients, so this page collects what the numbers actually say, with a source and a date on every figure, and it labels the guesses as guesses. If you want the lever-by-lever playbook instead of the benchmarks, read ChatGPT Ads optimization.

Why there are no official ChatGPT Ads benchmarks

OpenAI has not released any cross-advertiser performance data. Ads began showing to logged-in US free and Go users around February 2026, and the self-serve Ads Manager opened around May 2026, so every sample is small and volatile. Anyone who tells you "the average ChatGPT Ads CTR is X" is quoting a third party, usually a single agency's account, not OpenAI.

One distinction most posts blur: OpenAI's own ChatGPT Ads Manager is not the same thing as third-party gen-AI ad networks (Kontext, Thrad, and others) that place ads across many AI apps. Some of the splashier case studies, including a widely shared "70 percent higher ROAS" result, are from those networks, not from ads bought inside ChatGPT. We only use ChatGPT-specific figures below.

The ChatGPT Ads numbers that actually have a source

MetricReported figure (2026)SourceConfidence
CTR0.68% (May) falling to ~0.50% (June); top quartile ~1%, best brands ~1.6%Similarweb, as reported by PPC.land and Search Engine RoundtableAnalyst, single dataset
CPM~$25 to $45 clearing (launched at ~$60)Practitioner reportsDirectional
CPC~$3 to $5 category floors; $8 to $15 for deep B2B; $1.72 measuredPractitioner / OpascopeDirectional
Conversion rate2.35% measured; 4% to 7% estimatedOpascope / LapisOne real case vs. estimates
ROAS1.49x blended (the only public real-spend figure)OpascopeSingle case

Two honest notes on this table. The 0.68 percent CTR is repeated everywhere, but it traces to a Similarweb report rather than the blog post most outlets link, so attribute it carefully. And the pretty 4 to 8 percent conversion rates are single-source estimates: the only campaign that published a measured rate came in at 2.35 percent, materially lower.

The one fully documented campaign

The most citable hard number in the space comes from Opascope, who published 15 days of real spend in June 2026: roughly $60,000 in spend produced about $89,000 in revenue, a 1.49x blended ROAS, a $1.72 CPC, and a 2.35 percent conversion rate, with daily ROAS swinging between 0.2x and 2.9x. The author is explicit that these are one account's results on a young platform, not universal benchmarks. We agree, and that volatility (0.2x to 2.9x day to day) is the real lesson: judge this channel over 60 to 90 days, not on a Tuesday.

Why the benchmarks are moving: the auction is filling up fast

A benchmark from May is not a benchmark for December, because the auction is getting more crowded by the week. Similarweb counted 5,171 unique advertisers by mid-2026 with the base roughly doubling every four weeks, and ad penetration of US chats rising from 14 percent in May to 26 percent in June. More advertisers compress CTR (which is exactly what the 0.68 to 0.50 percent slide shows) and lift CPMs off their early floors.

The map is expanding just as fast. After the US, ChatGPT Ads reached the UK, Mexico, Brazil, Japan, and South Korea, and then around 31 European countries in August 2026, bringing the total to roughly 40 markets, with Canada, Australia, and New Zealand named as coming next (per Search Engine Land, Euronews, and Digiday). OpenAI also began rolling out conversion-optimized (oCPC) bidding in beta and a "Maximize results" automated bid strategy in the second half of 2026. Each new market and bidding tool resets the benchmarks again. So use any number here as a starting expectation, then measure your own.

Benchmarks by industry (early estimates, clearly labeled)

There is no measured, per-vertical ChatGPT Ads dataset yet. The table below is the best structured practitioner estimate available (Lapis, August 2026), and the source itself calls these "early and emerging" figures, not measured benchmarks. We reproduce them because the ordering is plausible and useful, but do not take the exact bands as fact.

VerticalEstimated conversion rateEstimated CTR
Education and instruction6% to 11%0.7% to 1.3%
Local and personal services6% to 10%0.8% to 1.5%
Health-adjacent5% to 9%not reported
SaaS and B2B4% to 8%0.6% to 1.2%
Finance and insurance4% to 7%not reported
E-commerce and retail3% to 6%0.4% to 1.0%

Estimated CPC sits around $3 to $5 across most verticals. The pattern practitioners give for the ranking: research-heavy, high-consideration decisions (education, local and personal services, B2B) benefit most, because the conversation itself adds value before the click. Price-sensitive, comparison-shopping categories like general e-commerce lag. Remember the reality check: the one measured campaign converted at 2.35 percent, below every band above.

Two things this table does not tell you. First, some categories are currently restricted for policy reasons rather than performance ones (financial, legal, healthcare, and data-brokerage advertisers face limits), so "can you run it" comes before "does it convert." Second, there is no reliable ChatGPT-specific data yet for travel, beauty, or CPG, so we will not print numbers we cannot source.

How ChatGPT Ads compare to Google and Meta

The base rates for the incumbents are well established: Google Search converts around 2 to 4 percent (WordStream, across 13,000-plus campaigns) and Meta around 1 to 3 percent (Littledata, across 2,800-plus Shopify stores). Practitioners argue ChatGPT clicks convert higher because the user self-qualifies mid-conversation, but there is no matched, apples-to-apples study yet, and the one real campaign placed ChatGPT "between search and paid social." So the fair read is: conversion quality looks promising but unproven, while the clear, measurable edge today is cost. CPMs on ChatGPT run a fraction of competitive Google categories, on arguably the highest-intent surface in advertising. Buy the cheap, high-intent clicks now; do not assume a CVR advantage you cannot yet measure.

What actually moves the numbers

The industry-wide consensus on optimization is short and boring, which is a good sign: one conversation or theme per ad group written from the buyer's perspective, CPC bidding for performance and CPM for awareness, dedicated landing pages matched to each ad group's intent, and judging the channel on cost per qualified action rather than CTR. Give each test about 100 clicks before you read it, and the channel 60 to 90 days before you judge it. Track with the OpenAI pixel plus the server-side Conversions API, because the platform dashboard alone cannot tell you conversion value.

That is the summary. The full lever-by-lever version, including bid-floor testing and the tracking stack, is in our ChatGPT Ads optimization playbook. For where the cheap inventory is by category, see the Toronto ChatGPT Ads Index.

How we compiled this

Every figure on this page is attributed to a named source with a date, and labeled by type: analyst (Similarweb), practitioner case study (Opascope), or practitioner estimate (Lapis, Just Global). OpenAI publishes no official benchmarks, so we do not claim any. Where a number is a single-source estimate, we say so; where there is no reliable data, we leave it blank rather than invent a figure. Because the platform is young and expanding monthly, confirm current numbers before you set budgets. We do not accept payment for coverage, and we run this channel ourselves.

FAQ

What is the average ChatGPT Ads CTR in 2026? There is no official figure. The most-cited third-party number is roughly 0.68 percent in May 2026, falling to about 0.50 percent by June as more advertisers entered (Similarweb, via PPC.land and Search Engine Roundtable). CTR is structurally lower than search because the ad appears after ChatGPT finishes answering.

How much do ChatGPT Ads cost? Practitioner reports put clearing CPMs around $25 to $45 (down from a roughly $60 launch level) and CPCs at a few dollars, with category floors near $3 to $5 and higher for deep B2B audiences. The one campaign that published real spend measured a $1.72 CPC. OpenAI does not publish cost data.

What is a good ROAS on ChatGPT Ads? The only public real-spend figure is 1.49x blended over 15 days (Opascope, June 2026), with daily ROAS swinging from 0.2x to 2.9x. Treat that as one data point, not a target, and judge your own account over 60 to 90 days.

Does OpenAI publish official ChatGPT Ads benchmarks? No. OpenAI has released no cross-advertiser CTR, CPC, conversion, or ROAS benchmarks. Any figure you see is third-party, usually a single agency's account, so weigh the source and the date.

Which industries perform best on ChatGPT Ads? Practitioner estimates rank research-heavy, high-consideration categories highest: education, local and personal services, and B2B SaaS, with e-commerce lagging on price sensitivity. These are single-source early estimates, not measured data, and some verticals (finance, legal, healthcare) are policy-restricted regardless of performance.

Are ChatGPT Ads worth it in 2026? For most advertisers, as a funded experiment rather than a budget migration: the auction is still uncrowded, CPMs are low, and early accounts build data before conversion bidding matures. The risk is asymmetric and the numbers are unproven, so instrument tracking first and judge on your own pipeline, not on someone else's benchmark.

Want the optimization playbook, not just the platform overview? Get our free ChatGPT Ads cheat sheet — context hints, bid-floor tests, industry readiness, and the measurement stack from practitioners running real budgets.

Quick answers

Shah Md. Rifat

Shah Md. Rifat
Content Strategist · Stratezik · Toronto, ON · LinkedIn