Between April and August, the count of advertisers on ChatGPT ads multiplied 21-fold, yet monthly spending stopped rising after May. That gap points to a crowded auction on a fixed shelf rather than a booming one. For search teams weighing a paid test inside ChatGPT, the shape of the market matters as much as the headline growth.
The analysis, published by Search Engine Land on November 10, 2026, comes from Ronnie Malewski, Managing Director at NP Digital. It relies on AdClarity data for U.S. ChatGPT ads from April 13 to August 31. AdClarity counted 174,700 campaigns and 346,900 ads, over 8 billion impressions, and roughly $419 million in estimated spend. Nearly 90 percent of global ChatGPT ad spending came from the U.S. in that window.
Monthly active advertisers rose from 839 in April to 2,702 in May and 17,934 in August. Weekly impressions held at about 390 million. More bidders chasing roughly the same inventory is the likely reason the typical advertiser’s budget shrank by about 85 percent. By August, a third of the advertisers that had tried the platform had stopped running ads.
Churn alone does not show that campaigns failed. The analysis notes that tracking sales and conversions was limited at launch, so some may have left because they could not tell whether ads worked. That is a measurement problem, and it is separate from a performance problem.
Spend is concentrated. Software and services for work companies hold 57 of the top 100 advertiser slots and 8 of the top 10, and they account for more than 69 percent of spend among that group. Retail and consumer brands make up about 7.6 percent of total spending. The top 100 advertisers take 43 percent of spend while making up under half a percent of all advertisers.
The tail is long. Across more than 26,000 advertisers, the top 10 take 15 percent of spend and the top 1,000 take 79 percent, leaving about 25,000 advertisers to split 21 percent. Just 62 advertisers reached $1 million, 144 reached $500,000, and 592 reached $100,000. The data does not say whether smaller advertisers get better or worse results than the big ones.
Commerce signals were weak in the same period. Instant Checkout was pulled back by OpenAI, only about 30 Shopify merchants launched on it, and Walmart saw in-chat purchases convert at roughly a third of the rate of visits to its own site. That context may explain why retail is thin.
OpenAI is also rolling out image-based ads, which it first tried inside image generation for U.S. users of its two lower tiers, Free and Go. Measurement integrations are coming too: conversion data via Hightouch and LiveRamp, attribution through AppsFlyer and Adjust, and incrementality studies with Haus and WorkMagic, among other partners.
Two cautions apply. AdClarity derives spend by pricing the impressions it observes at assumed ad rates, so the dollar figures are not official numbers from any company. The author is an agency executive, which is worth weighing when reading the conclusions.
The practical advice from the analysis is to set up measurement before spending: a website pixel, Conversions API, and UTMs. It also warns that last-touch reporting inside ChatGPT may miss some interactions. Teams that test ChatGPT ads over the next quarter should cap the budget, tag every click, and compare results against branded search before scaling, because flat inventory with a rising bidder count leaves little room for guesswork.
Based on analysis by Ronnie Malewski in Search Engine Land, published November 10, 2026, using AdClarity estimates.
