ChatGPT ad click-through tops 1% in four markets, while the U.S. sits at 0.73%
According to The State of ChatGPT Ads, a report by growth agency Graphite produced with Similarweb, click-through on ChatGPT ads among Free and Go users is above 1% in Brazil, Mexico, Japan and South Korea. The U.S. rate is 0.73%. Because every advertiser is pooled together, these numbers should not be read as a target for any single campaign.
Nine markets, two distinct click-through groups
Graphite's research page carries the title The State of ChatGPT Ads and a September 28, 2026 dateline, while Matt G. Southern's write-up in Search Engine Journal says the report was published on September 29. Graphite ran the research with Similarweb: Similarweb's user panel supplies reach, click-through rate (CTR) and ad placement data across the nine markets Graphite describes as live, and Graphite pairs that with its own review of U.S. ad copy and landing pages. SEJ adds a caveat on that scope: none of the 31 European countries that OpenAI said on August 18 would get ChatGPT Ads show up in the report's per-market charts. The full report is sent as a PDF once a form on the page is filled in. SEJ also flags the commercial context: Similarweb sells the ChatGPT ad tracking tool the panel data comes from, and Graphite is a growth agency focused on SEO and paid search.
As SEJ reads the report, every market gets an all-time CTR for Free and Go plan users, running from the week it entered the panel up to September 16. Brazil leads at 1.91%, followed by Mexico at 1.55%, Japan at 1.53% and South Korea at 1.21%. The five English-language markets sit close together: Canada 0.75%, the U.K. 0.74%, the U.S. 0.73%, Australia 0.67% and New Zealand 0.6%. Coverage is uneven, though: each market has somewhere between 5 and 25 weeks of data behind its rate.
CTR is counted per reply, and category data is impression share
The method is what makes the numbers readable. Per SEJ, CTR is calculated as clicks per ad-bearing exchange, where an exchange is one user message plus ChatGPT's response, and clicks are pooled across all advertisers in each market. Graphite's methodology note says Similarweb provides only percentages and shares and keeps panel volumes undisclosed. The ad-side data is Graphite's own: 95k+ U.S. ChatGPT ads collected from 500+ advertisers between April 1 and August 16, 2026, plus 8k+ landing pages crawled and classified.
Monday.com led a B2B SaaS field of 1,210 advertisers that won 19.9% of U.S. ChatGPT ad impressions between March 30 and August 4. In second place, Home Depot headed 1,510 retail and consumer brand advertisers with a combined 10.4%. Graphite says retail and ecommerce advertisers have mostly stayed away, and puts that down to product feeds that did not suit ChatGPT's conversational format. The leaderboard keeps moving: across the 20 weeks from May 4 to September 14, Jotform and Resume.io were the only advertisers to hold a worldwide top-10 spot by impression share every single week. Among the 100 largest U.S. advertisers, the report files Zapier, SAP, Aikido Security and The Wall Street Journal under gone: their impressions climbed and then faded before August 17.
Benchmarks are scarce, so the table is useful, but it is not a campaign goal
SEJ points out that, by OpenAI's own help center account, ChatGPT Ads has no performance benchmarks yet, whether by advertiser, industry or campaign type. In that gap, a market-wide CTR comparison gives advertisers a reference line. Because clicks from every advertiser are pooled and counted per reply, Southern writes that the result is "a different measurement than the CTR of a single campaign." With different time windows per market and no underlying counts, a direct comparison with a campaign's own CTR is hard. On reach, Graphite says 43% of ad-eligible U.S. users see ads after six months; in the reach chart on its page the U.K. is highest at 45.7% and Mexico lowest at 27.8%.
Graphite's playbook is cautious. In its words: "CTR has doubled since launch, but the advertiser base hasn’t settled." It advises keeping spend experimental, and it finds that early turns in a conversation draw the highest CTR and the largest impression share. The copy of top advertisers is 98% new, and feature, pricing and product pages are the most used landing page type. Because Google Search still carries far more ad volume than ChatGPT or Meta, Graphite recommends budgeting ChatGPT ads as an addition rather than a swap. Timing matters too: the U.S. category data ends on August 4, before Amazon Ads started its pilot on September 10 and before OpenAI named Shopify its first ecommerce partner on September 16. Per SEJ, OpenAI started bringing ChatGPT Ads to Indonesia, Malaysia, the Philippines, Singapore, Thailand, Vietnam and Taiwan on September 23, and it now puts availability at more than 60 countries.
The sources say nothing about Türkiye, so this section is commentary
Neither source mentions Türkiye, and it is not among the nine markets with data: Brazil, Mexico, Japan, South Korea, Canada, the U.K., the U.S., Australia and New Zealand. SEJ reports OpenAI's statement that ChatGPT Ads is now available in more than 60 countries but does not give the full list, so we make no claim either way about whether the ads are open to advertisers in Türkiye. What follows is not drawn from the sources; it is explicitly UNALSOFT commentary.
The unknowns: whether ChatGPT Ads is open to advertisers in Türkiye, CTR and reach for users in Türkiye, how Turkish-language ads perform, and the cost side. Neither the report's public page nor the SEJ article includes cost per click, cost per thousand impressions or conversion rates. Our reading is that what travels to Türkiye is not the numbers but the method caveats. Not setting a pooled CTR as a campaign target, writing copy for the channel, testing the product and pricing pages you already have, and ring-fencing the budget as an experiment hold wherever the channel opens.
UNALSOFT's take
We read this report less as a scorecard and more as a lesson in measurement literacy. The table offers a frame for what a good click-through rate looks like on a new channel, but because it pools every advertiser per reply, it is not a campaign target. That is why, on the ad management side, we set up a new channel with copy written for it, a small landing page set built from existing product and pricing pages, and a separately tracked test budget that does not eat into Google Search. Because the sources carry no data about Türkiye, this article proposes no local performance expectation.
Set up the measurement before you test a new ad channel
A short conversation is enough to settle which metric, which budget and which landing page your test should run on.