I analyzed 14,000 websites running ChatGPT ads. Here's who's buying.

I found OpenAI's ad pixel on 14,406 company websites. The buyers aren't a new crowd. They're the teams that already buy everything else, and they test new channels first.

• 14,406 companies with the ChatGPT ads pixel
• 45% also run 4+ other ad platforms (vs 5% of sites)
• 1 in 5 Qualified and G2 buyer-intent customers already run it
• 82% of pixels aren't in the page's HTML at all

When a company advertises on ChatGPT, it adds a small OpenAI script to its website, oaiq.min.js from bzrcdn.openai.com, to count the sign-ups and purchases its ads drive. Like every ad pixel it's public: anyone who loads the page can see it.

So I looked. Across 1.79 million company domains I found the pixel on 14,406 companies. HTTP Archive's September crawl of about 11 million homepages saw it on 37,028 sites in total, including shops and blogs outside my company list. That's about 1 in 70 company websites. Then I compared those advertisers with everyone else: which ad platforms they run, what software they use, how old and how big they are, and where they're based.

They already buy everywhere

More than four in five ChatGPT advertisers also run Google Ads, and nearly as many run a Meta pixel. That's more than double the rate of the average site, which is what you'd expect from anyone who pays for traffic. The interesting part is further down the list.

Share running each ad platform. ChatGPT advertisers compared with the 863,000 company homepages in HTTP Archive's crawl.

The biggest gaps are on the channels most marketers still call experiments. ChatGPT advertisers are 14 to 17 times as likely to run Reddit or Quora ads and 12 times as likely to run Spotify ads. These are teams with a test budget and the habit of spending it on whatever is new.

Counting ad platforms per company shows the same thing. Almost half of all websites run none. The typical ChatGPT advertiser runs three or four.

How many other ad platforms a company runs, out of 12: Google, Meta, Microsoft, LinkedIn, TikTok, Reddit, Pinterest, X, Snapchat, Criteo, Spotify, Quora.

So ChatGPT isn't pulling in a new kind of buyer. It's one more line in the plans of teams that already spread spend across five channels, and it's competing with Reddit and podcasts for the "let's see if this works" money, not with the Google budget.

B2B marketers lean in hardest

The clearest way to see who's buying is to ask: of the companies that use a given tool, how many already run ChatGPT ads? Across all company sites the answer is 1.4%. Among customers of B2B buyer-intent tools it's ten times higher. About 1 in 5 companies running Qualified or G2's buyer-intent tag have the pixel, and so do 1 in 6 6sense customers. These teams sell software, measure pipeline rather than clicks, and know their buyers now research software in AI assistants.

The second cluster is data-heavy direct-to-consumer brands. Northbeam and Triple Whale are attribution tools you only buy when you're spending real money across many channels, and their customers adopt ChatGPT ads at 7 to 15 times the average rate.

Share of each tool's customers running ChatGPT ads, among company sites in the crawl that use each tool. The dashed line is the average across all company sites (1.4%).

The bottom of that chart matters too. Only 1 in 100 WordPress sites and fewer WooCommerce stores carry the pixel. The small-business long tail that runs on themes and plugins hasn't shown up yet.

Software and marketing lead, health care lags

Marketing agencies have the highest adoption of any industry: about 50 in every 1,000 run the pixel, probably agencies testing the channel for clients before anyone else does. Software and internet companies follow. Industries that sell locally or are heavily regulated, such as hospitals, hospitality and food, sit far below average.

ChatGPT advertisers per 1,000 company sites, by industry. Industries with 80+ advertisers among 1,000+ company sites. The dashed line is the average across all company sites (14 per 1,000).

Mid-sized, younger companies move first

The sweet spot is companies with 51 to 1,000 employees: big enough to have a growth team and a test budget, small enough to move quickly. Companies with 11 to 50 people are underweight. Age matters even more. Companies founded after 2010 make up 44% of ChatGPT advertisers but only 31% of company sites, and companies founded before 2000 are underweight by almost a third.

Company size (employees).

When advertisers were founded.

Four in ten are American

The United States accounts for 38% of advertisers with a known country. The UK and Brazil follow, each with about 6%. Brazil's position is the surprise: it's ahead of India, Canada and every European market. Japan has one of the most striking clusters in the data, with several of Recruit Holdings' consumer sites, including Hot Pepper Gourmet, SUUMO, Jalan and Carsensor, all carrying the pixel.

Where ChatGPT advertisers are based. Share of the 12,168 advertisers with a known country. The other 22% are spread across more than 100 countries.

Some of the names

Among the best-known sites with the pixel installed:

• Software: Zoom, Atlassian, Figma, Notion, Datadog, Twilio, Dynatrace, Box, Mailchimp, SurveyMonkey, Typeform, Amplitude, Pendo, Optimizely, Klaviyo
• Consumer: Uber, Eventbrite, GoFundMe, Coursera, Scribd, IXL Learning, Vinted, Discogs, CurseForge
• Web infrastructure: GoDaddy, Namecheap, Hostinger, Porkbun, Wix, Bitly, Cloudinary, OneSignal
• Security: Avast, ESET, Check Point, DataDome, TrustArc
• Japan: Hot Pepper Gourmet, SUUMO, Jalan, Carsensor, ZOZO, Piccoma

Most of these pixels are invisible to a scraper

Most companies don't paste the OpenAI script into their HTML. They add it through a tag manager, which loads it after the page opens. For 40% of advertisers the pixel only appeared when a real browser loaded the page. For another 42% it was configured inside a Google Tag Manager container. Fewer than one in five had it in the page code itself, which is all a simple scraper reads. That's why counts of ChatGPT advertisers vary so much depending on who's counting.

Where the pixel was found.

How I counted

• A company counts as a ChatGPT advertiser when its website loads OpenAI's ad conversion pixel (bzrcdn.openai.com, oaiq.min.js), either in a scan of the site and its tag-manager container, or in HTTP Archive's September 2026 crawl, which loads about 11 million homepages in Chrome.
• Comparisons use the 863,000 company homepages in that crawl; 12,156 of the advertisers are among them, and the rest were found by scanning the site directly. Industry, size, country and founding year come from public company profiles and only cover companies that have them.
• A pixel shows that a company set up conversion tracking for ChatGPT ads, not how much it spends or whether a campaign is live today. Some teams install pixels before they launch.
• Only homepages and their tag containers were checked, so a pixel that loads only on a checkout page is missed. Every number here is a lower bound.
• Updated October 8, 2026 from the current store; the first version (October 7) counted 11,392 advertisers against 631,000 crawled homepages. Since then more companies were matched to the September crawl. Changes in method: the industry chart now shows industries with 80+ advertisers (it was 50+, which now covers 40 industries), duplicate profile labels are merged (for example "computer software" and "software industry", "law practice" and "legal services"), and country names are normalized to one code each. A pixel counts as tag-manager when any tag-manager container loads it, as page code when it appears in the HTML, a bundle or a script tag, and as browser-only otherwise.