Sponsored ads showed up in 47% of ChatGPT conversations about video games and consoles in June, up from just 3% in March. Toys and hobbies chats went from 5% ad presence to 24% over the same stretch. If you sell toys, games, or hobby products and you're staring at a self-serve Ads Manager wondering whether to put real budget behind it, those numbers look like a green light. They might not be.
The figures come from Comscore's Q2 2026 AI Intelligence Report, released September 22. They track something narrower than "ad opportunity." Comscore measured the share of U.S. PC-based ChatGPT conversations, classified into retail categories, that included at least one sponsored ad — not the number of users reached, not impressions, and not conversions (Search Engine Journal). A conversation counts the same whether it had one passing ad mention or five.
Comscore hasn't published how many conversations sit behind each monthly percentage. A jump from 18% to 47% in gaming could reflect a real surge in ad delivery, or it could reflect a small sample swinging hard month to month. Without conversation counts, you can't tell which.
Context: What the Panel Can and Can't Tell You
Comscore's own source note flags two more issues worth sitting with before you read 47% as "nearly half of gaming shoppers see ads." First, the data comes from an opt-in panel limited to U.S. PC activity, which skews toward a specific type of desktop user and doesn't capture mobile or app-based ChatGPT use at all.
Second, it's unclear whether ad-free plan conversations are mixed into the denominator. OpenAI's ads currently show to Free and Go tier users and logged-out visitors; Plus, Pro, Business, Enterprise, and Edu accounts don't see ads at all (Search Engine Journal). If Comscore's panel includes any ad-free accounts in its base, the real ad rate among conversations actually eligible for ads is higher than the headline number suggests.
There's a timing wrinkle too. OpenAI began testing ChatGPT ads on February 9 for logged-in Free and Go users, and Comscore's series starts in March, just weeks into that test. On May 5, OpenAI rolled out a beta self-serve Ads Manager with CPC bidding alongside the original CPM option.
June, the month showing 47% in gaming and 24% in toys, was the first full month after that rollout went live. Comscore doesn't connect the two directly, and the report can't distinguish whether the spike reflects more advertisers buying in, more inventory being shown, or something else entirely. Search Engine Journal's own read on the data lines up with that caution: until Comscore releases detailed conversation counts, the June increase should be treated as a preliminary observation, not a confirmed trend.
For comparison, an SE Ranking analysis covered in August looked at 50,006 commercial prompts collected on a single day in July and found ads appeared in 25.94% of them. That's a different methodology — prompt-level ad frequency on one day, rather than conversation-level tracking over months. The two data sets aren't directly comparable, but both point the same direction: ad presence in commercial ChatGPT queries is real and growing, even if nobody has pinned down exactly how big. For more on this, see more on ai search data sources: what matters by business type.
Implications: What This Means for Your Test Budget
The Worked Example
Say you run a mid-size toy retailer with $50,000 in monthly ad spend across Google and Meta. You see the 24% toy and hobby ad-density figure and wonder if it justifies a $500 to $1,000 test on ChatGPT's Ads Manager. This pairs well with go deeper on will ai replace seo jobs? what pew's data really shows.
The honest answer: the density figure alone doesn't justify anything. It measures how often ads appeared in conversations Comscore tracked, not how many toy shoppers are having those conversations or what they do afterward.
What actually justifies the test is something else Comscore's chart didn't measure: OpenAI's Shopify integration, announced September 16, lets Shopify merchants launch ads through their existing product catalog with no separate feed setup, and expansion to more markets started September 23 (Search Engine Journal). Low setup friction plus CPC bidding — where you pay only for clicks, not impressions — makes a $500 to $1,000 test genuinely low-risk regardless of what the density chart shows. Run the test because entry costs are low, not because a category statistic promises reach.
The Verification Checklist
Before you act on any third-party AI ad-presence report, check three things:
- Does it disclose sample size or conversation counts?
- Is the panel opt-in and platform-limited in ways that skew the audience?
- Does the methodology account for ad-free tiers being excluded from, or mixed into, the base?
Also read: see how to audit ga4 for real vs. inferred ai traffic
Comscore's report fails the first check outright and leaves the third ambiguous. That doesn't make the data useless — it's a real signal that ad inventory exists and is growing. It just means you shouldn't treat it as a market-sizing tool.
Limitations to Keep in Mind
This is preliminary, self-reported data from one research firm, without published conversation counts to back up the percentages. It tells you ad presence is rising in retail-adjacent chats; it doesn't tell you how many shoppers you'd reach, what they'd click, or what they'd buy.
If you do run a test, track your own numbers instead of leaning on category averages: click-through rate against your Shopify catalog listings, cost per click relative to your Google Shopping benchmarks, and whether ChatGPT traffic converts at a comparable rate to your other channels. Given the platform's newness, watch attribution quality too, since ChatGPT referral tracking in analytics platforms is still maturing.
What to watch next: whether Comscore publishes conversation counts in a future report, and whether OpenAI's Sponsored Agents feature — still in limited testing — changes how ad interactions get counted at all.
