Your competitors are already paying to sit under ChatGPT's answers

Somewhere in the page source of Twilio, HubSpot and Gusto sits a small JavaScript file served from bzrcdn.openai.com. It is the measurement pixel for ChatGPT Ads, and Bloomberry found it on roughly 15,000 company websites when it scanned the 2 million largest employers on LinkedIn. If you sell B2B software and spent the past year optimising to get mentioned in ChatGPT's answers for free, some of your competitors were wiring up attribution so they could pay to appear right underneath them.

I went into the data expecting a crowd of tiny AI startups throwing $25 a day at a shiny new toy. That is not who turned up. Companies with 11 to 50 staff show up about as often as you would expect by chance. Firms with 51 to 200 employees are 2.4 times overrepresented, and those with 1,001 to 5,000 employees 2.8 times. Series B companies over-index by 2.5 times and Series C by 3.1 times. Software and SaaS firms make up 17.0% of adopters against 3.3% of the wider company base. These are the companies with performance marketing teams, attribution in place and quarterly pipeline targets, which is to say the companies you lose deals to.

The headcount figure deserves some suspicion. Bloomberry's own public list page says it detected 7,224 companies, and a pixel on a site proves somebody set up tracking, not that money is flowing. Take the lower number if you like. I would still read it as several thousand businesses, heavily weighted towards software, that have decided ChatGPT is a place to buy attention. And OpenAI says someone is paying: on 31 August it announced a $1 billion annualised revenue run rate in under 200 days, a pace PYMNTS reports took Google four years.

The usual reassurance from the organic camp is that OpenAI keeps church and state apart. Ads appear beneath the response, are labelled, and according to OpenAI do not influence what ChatGPT says. I take OpenAI at its word, and I think that makes the organic-only strategy weaker. You can do everything right, get your product named in the answer, and then watch a rival's sponsored unit land directly below it. OpenAI's help pages say the ad system weighs the context and intent of the current conversation when choosing what to show. A prospect asking which payroll tool suits a 200-person company is exactly the intent a paying rival wants to sit against.

The companies doing the organic work already know this. Of 1,892 firms in Bloomberry's data using ChatGPT prompt links, 186 also run ChatGPT Ads, or 9.8%. Among firms without prompt links the figure is 0.13%. The marketers most serious about appearing in answers are about 75 times more likely to also be paying for the slot below them. They are running organic and paid together, the way search teams have run SEO and paid search side by side for twenty years.

There is a fair objection from the people actually running campaigns. MediaPost reported that agencies were complaining on LinkedIn about weak ChatGPT Ads performance just days before OpenAI's billion-dollar announcement. I believe some of those campaigns flopped. But a channel that disappoints agencies in its first months is one where a competitor's test is cheap, and OpenAI has every reason to make it work. It dropped the $50,000 minimum, added CPC bidding alongside the launch price of roughly $60 CPM, and Amazon now runs ChatGPT campaigns as a managed service. Neil Patel's reporting says the first $100 million of run rate came while fewer than 20% of eligible users saw ads daily. In my view OpenAI will turn that dial up, more bidders will crowd the same high-intent B2B queries, and the early entrants will have months of conversion data that latecomers lack.

The nasty part for a marketing lead is visibility. OpenAI does not publish its advertiser list. A competitor's Google ad shows up when you search your own category; a ChatGPT ad appears inside someone else's conversation, in the free and Go tiers, which your team probably does not use for work. You could be outbid on your most valuable buying questions for a quarter and never notice.

Reddit's sales team should be uneasy too. Nearly one in ten Reddit advertisers in the dataset also carries the ChatGPT pixel, and ChatGPT advertisers are 7.3 times more likely than other digital advertisers to buy Reddit ads. The same growth teams are shopping both platforms for the same budget.

This afternoon, open the homepage source of your five closest competitors and search for bzrcdn.openai.com. Then log into a free ChatGPT account, type the ten questions your buyers ask before a demo, and screenshot every sponsored unit under the answers. If a rival's name appears in either check, take a line out of next quarter's budget and start bidding.

Prompted by Who’s buying ChatGPT ads? I analyzed 15K advertisers, Bloomberry.