What ChatGPT ads can be used for today: discovery and product recommendation campaigns, with direct response still limited
Updated 5 min read
In short
ChatGPT ads are currently best suited to consumer discovery and retail-style product recommendation, especially for advertisers that can use product feeds. As of October 2026, OpenAI reports that ads created from product feeds have been among the strongest-performing formats in its program, and these are designed for ecommerce-style catalogs. The audience is structurally consumer-skewed, and delivery optimizes on conversational context while users explore, compare, and decide, which favors upper- and mid-funnel influence more than strict direct response.
Key takeaways
As of October 2026, OpenAI states that product-feed-based ads have been among the strongest-performing formats in its ChatGPT ads program.
ChatGPT ads skew toward consumer discovery because only Free and Go users see ads, and delivery keys on conversation context while users explore, compare, and decide.
Launch categories focus on lifestyle goods, local services, travel, and digital products, with regulated verticals like finance and healthcare allowed only case-by-case in the US.
The channel has structural limits such as no keyword bidding, a fixed reporting set, and billing that never charges per conversion even in conversion-optimized campaigns.
Retail advertisers that qualify can start with a product-feed campaign and a minimum daily budget of $25, while others typically begin with a context-driven awareness test.
ChatGPT ads are emerging as a context-aware, consumer-skewed channel that is strongest today where it can recommend concrete products from feeds and support discovery while people are actively comparing options. For the full capability picture you can cross-reference ChatGPT ads capabilities, and ecommerce teams can go deeper in the separate retail playbook at ChatGPT ads for ecommerce.
Documented strengths of ChatGPT ads today
Per OpenAI's documentation as of October 2026, ads created from product feeds "have been among the strongest-performing ads" in the ChatGPT ads program to date. This is the clearest performance signal OpenAI has published, and it is specific to feed-based formats rather than all campaigns. Feed-based campaigns are designed for retail-style advertisers that have broad or frequently changing catalogs, and the ad units can show images, titles, star ratings, prices, sale prices, and brand information.
This design makes the format well suited to product recommendation and comparison use cases, where a user is asking ChatGPT for options and the ad can surface a relevant item from the catalog. Because the system can pull structured fields like price and rating directly from the feed, it can present a compact, decision-ready summary inside the conversation. OpenAI also provides a limited feed beta for hotels, which extends this structured recommendation pattern into travel inventory.
These documented strengths do not mean ChatGPT ads can replace search ads for capturing intent. There is no keyword bidding in ChatGPT ads, and per OpenAI's help content, delivery instead keys on conversational context and intent. That is a different mechanism from query capture in search auctions, and it shapes where the channel fits in a media mix.
Why the channel skews toward consumer discovery
OpenAI states that ads in ChatGPT are shown only to Free and Go plan users, while paid professional tiers and under-18 accounts are excluded. This construction makes the reachable audience structurally consumer-skewed, because Plus, Pro, Business, Enterprise, and Edu users do not see ads at all. Any expectation that the channel reliably reaches professional B2B buyers at work is therefore misleading, since many of those users are on paid tiers that are excluded from ad delivery.
Delivery also keys on the context and intent of the current conversation. OpenAI describes ads as appearing while users "explore, compare, and decide," which is a discovery-oriented moment rather than a traditional search results page. The system looks at what the user is talking about and their apparent goal, then decides whether to show an ad that fits that context.
This combination of consumer-heavy reach and exploration-focused delivery makes the channel naturally suited to consumer discovery. Users are often asking for ideas, options, or recommendations, and the ad can participate in that exploration by suggesting a product, service, or experience that matches the conversation. That structural design is why the channel aligns more with discovery and mid-funnel influence than with strict last-click direct response.
Business types and formats that fit today
OpenAI's ad policies, as of version 1.6 in September 2026, define which categories can use ChatGPT ads. Allowed launch categories include lifestyle and household goods, local services, travel and experiences, and digital products and education. These are the areas where a consumer-skewed, recommendation-oriented channel can most easily match user conversations with relevant offers.
Regulated verticals such as financial services, healthcare, and legal are not broadly open. OpenAI states that these categories are handled case-by-case, US-only, with manual review and possible requirements to prove licensure. This means that any assumption that "any business that wants to test can" is false, because many verticals are either excluded or tightly restricted by policy.
Within the allowed categories, format fit matters. Retail and ecommerce advertisers get access to a dedicated product-feed format, currently in beta, where they can upload a catalog and have the system generate product ads with images, titles, ratings, and prices. Hotels have access to a limited feed beta as well, which supports structured travel inventory. Other allowed categories, such as local services or digital education products, can still participate but will not use the retail product-feed format; they rely on standard conversational ad placements keyed to context.
Structural limits of ChatGPT ads as of October 2026
Several structural characteristics limit how ChatGPT ads can be used today. First, there is no keyword bidding. OpenAI states that context hints guide matching, but they do not guarantee placement in the way a keyword auction does. This means advertisers cannot build classic search-style intent capture strategies that depend on bidding on specific queries.
Second, reporting is a fixed set. OpenAI documents that advertisers see impressions, clicks, spend, click-through rate (CTR), average cost per click (CPC), average cost per thousand impressions (CPM), and conversions, with spend data lagged by 7 to 8 hours. This reporting set supports basic performance monitoring but does not provide deeper breakdowns beyond what is listed.
Third, conversion-optimized campaigns do not change how billing works. OpenAI states that these campaigns bill per valid click or per 1,000 impressions, never per conversion. The misconception that "conversion campaigns mean you pay for conversions" is therefore incorrect in this channel; optimization can focus on conversions, but charging is still click- or impression-based.
Audience construction has its own constraints. Custom audience inclusion requires at least 25,000 matched users, which limits list-based prospecting for advertisers with smaller databases. In addition, OpenAI's ad policies and availability have been changing monthly, with versions moving from 1.0 in March 2026 to 1.6 in September 2026, so any plan needs to account for policy evolution over time.
What a qualifying business should run first
For advertisers that qualify under OpenAI's category rules and want to test ChatGPT ads, the first decision is whether you can use product feeds. Retail advertisers can run feed campaigns by uploading a catalog via CSV file, a hosted URL, or SFTP, and OpenAI's own documentation identifies feed-based ads as among the strongest-performing formats so far. If you have a broad or frequently changing catalog, starting with a product-feed campaign aligns your test with the format that has the clearest documented strength.
OpenAI states that the minimum daily budget is $25 USD, and recommends starting with a daily budget to monitor delivery, along with a starting maximum CPC of $3 to $5. This frames the scale of an initial test and sets expectations on the floor you must meet to run. For businesses that cannot use product feeds, such as many local services or digital education providers, a first campaign will typically be a context-driven awareness or consideration test that targets relevant conversations within the allowed categories.
Misconceptions about what a first campaign can do should be cleared before launch. ChatGPT ads cannot replace search ads for capturing intent because there is no keyword bidding and matching is driven by conversation context instead. Conversion campaigns do not change billing to a per-conversion model, and not every business can participate due to category and geography rules, especially in regulated sectors that are US-only and require manual review.
Temper And Forge's tested read on use cases, based on early client testing as of October 2026, is that campaigns have converted for ecommerce but with variability too high to extrapolate performance at full maturity and spend volume. The firm has seen the most prominent fit in local service, B2C, and general awareness, and currently positions ChatGPT ads as an awareness recommendation for service-based offerings while continuing to test ecommerce, using small-batch testing in the $20 to $40 per day range for SMBs with the platform floor at $25, rather than treating the channel as a flagship recommendation.
Frequently asked questions
ChatGPT ads are structurally limited for B2B because ads only show to Free and Go users, while paid professional tiers such as Plus, Pro, Business, Enterprise, and Edu never see ads. This skews reach toward consumers rather than professional users at work. You can still reach some business owners or professionals on free accounts, but the platform is not designed as a dedicated B2B lead-generation channel.
Sources
This resource is for educational purposes. Ad platforms change their policies, features, and pricing frequently; verify details against the platform's current documentation before acting on them.
