Skip to main content
Perspectives 2026-09-23 Updated 2026-10-02 · 11 min read

OpenAI Brings ChatGPT Ads Below Answers: What Marketing Teams Should Do Next

ChatGPT Ads is available in Singapore and 62 other countries as of October 2, 2026. What buying ads changes for pricing, tracking and organic visibility.

A single dark card with a narrow indigo strip across it

ChatGPT ads are sponsored placements below a ChatGPT answer. OpenAI announced the start of its Singapore rollout on September 23, 2026. Its self-service availability list included Singapore among 63 available countries in the October 1 editorial check, reconfirmed on October 2, 2026. Buying that placement does not buy a recommendation in the answer.

Third-party tracking shows why the slot has to be measured apart from the answer above it. In Seer Interactive’s July 2026 data, an advertiser appeared inside the ChatGPT response itself on only 5.4% of the responses carrying its ad. The practical lesson is not that marketing teams must now choose between paid and organic ChatGPT visibility. It is that the answer and the ad slot are two separate contests, measured with different tools and won with different work.

Key findings

  • OpenAI says ChatGPT Ads reached a $1 billion annualized run rate in less than 200 days and serves tens of thousands of advertisers, while ads “do not influence answers”.
  • Seer Interactive found an ad on 77% of tracked ChatGPT responses by late July 2026, but a brand’s own ad on only 7.3% of responses to prompts about that brand.
  • In the same Seer dataset, which covered 3,697 unique advertisers, an advertiser was named in the organic response 5.4% of the time and cited as a source 3.3% of the time.
  • In Canlah AI’s September 7, 2026 self-audit, ChatGPT named the brand in 8.3% of open buyer answers and in 18 of 18 branded answers, a gap that no ad purchase addresses.
  • Canlah AI sampled ChatGPT through the OpenAI API, which is not the consumer ChatGPT surface where ads are served, so its archive measures the answer, not the ad slot beneath it.

What happened

OpenAI first described its advertising plans on January 16, 2026. The post set out principles that still frame the product: answer independence, conversation privacy, clear labelling, no ads for users under 18 and no ads near sensitive topics such as health, mental health and politics. Testing began in the United States on February 9, 2026 for logged-in adult users on the Free and Go tiers. OpenAI’s help centre now says ads may appear for Free and Go users, including in logged-out experiences, while Plus, Pro, Business, Enterprise and Edu accounts do not have ads. It also states that ads appear below the end of a response, are labelled as sponsored and run on systems separate from the chat model.

Buying opened in stages across 2026. Digital Applied reported on February 13, 2026 launch-partner terms of about $60 CPM with a $200,000 minimum spend. Digiday reported on April 21, 2026 that cost-per-click bidding had arrived, with suggested bids of $3 to $5 per click and CPMs as low as $25 in some cases. A self-serve Ads Manager beta followed on May 5, 2026. By August 2026 OpenAI said it had moved beyond CPM and CPC bidding to conversion optimisation and had added product feeds, geographic targeting and custom audiences; its Ads documentation describes product-feed ads that carry several product cards. On August 18, 2026 it announced expansion to 31 European countries the following week, and self-serve Ads Manager access across those markets opened on August 31, 2026.

Sources: OpenAI advertising principles; OpenAI testing announcement; OpenAI help centre; Digital Applied pricing guide; Digiday CPC report; OpenAI Ads Manager announcement; OpenAI Europe expansion; OpenAI product feeds documentation

On August 31, 2026 OpenAI said the ads business had passed $1 billion in annualized run rate and was launching across India, Europe, the Middle East and North Africa. MediaPost reported that the announcement cited more than 50 technology and measurement partners and one ecommerce advertiser with a threefold return on ad spend over 28 days. These figures come from the company announcement and should be treated as company-reported case evidence, not independent market benchmarks.

Sources: OpenAI milestone announcement; MediaPost coverage

ChatGPT ads pricing, buying and availability at a glance

ChatGPT Ads fact sheet, availability and product documentation rechecked October 2, 2026

Question Current public answer Source What it changes for planning
Does ChatGPT have ads? Yes, below the response, labelled as sponsored, for Free and Go users, including in logged-out experiences OpenAI help centre The answer and the ad are read in one screen, so both need an owner
Who does not see ads? Plus, Pro, Business, Enterprise and Edu users; users identified as under 18 OpenAI help centre These ad-free tiers cannot be reached through ChatGPT ad placements
Can you advertise on ChatGPT? Yes, through Ads Manager beta or the Advertiser API, in supported countries only OpenAI Ads docs Account country decides access before budget does
Which markets can buy? 63 countries marked Available on October 2, 2026, including Singapore OpenAI help centre The advertising and billing legal entity must be based in an available country
Is Singapore available? Yes; rollout announced September 23, 2026 and self-service availability reconfirmed October 2, 2026 OpenAI announcement; help centre Check the account country and billing entity before opening a campaign
ChatGPT ads pricing Billed per impression or per click; reported CPM fell from about $60 at launch to $25 in some cases Digiday; OpenAI Ads docs Reported ranges, so size the first buy as a test, not from a rate card
Bid strategies Fixed bid, maximize clicks, maximize conversions; conversion campaigns bill per click OpenAI Ads docs Validate conversion tracking and attach an active standard conversion goal before launch
Targeting Country and region, platform (iOS, Android, web), custom audiences (at least 25,000 matched users to include or bid up) and context hints OpenAI Ads docs A customer list under 25,000 matched users cannot be used for inclusion targeting or bid multipliers; exclusions have no minimum
ChatGPT ads tracking Measurement Pixel, Conversions API, image tag; default 30-day click and one-day view attribution windows OpenAI Ads docs Selected attribution windows affect reported conversions, CPA and ROAS; compare like-for-like settings

Source: availability and OpenAI product documentation rechecked October 2, 2026. Historical pricing figures retain their original report dates and are not OpenAI rate cards. Recheck availability on publication day and before purchase.

The September 23 announcement describes a rollout starting, not a guaranteed self-service purchase date for every Singapore account. Account creation and access are separate.

Context hints supplement targeting and are not exact-match keywords, with up to 2,000 hints per ad group. They describe relevant products, use cases and needs. Singapore availability is now documented by the rollout announcement and Ads Manager list, rather than inferred from hiring.

Why the ad slot does not buy the answer

OpenAI’s own position is that advertisers cannot shape, rank or influence what ChatGPT says. Seer Interactive’s tracking shows what that separation looks like in practice. Across roughly two weeks of data from its clients’ AI tracking workspaces, updated July 28, 2026, ads appeared in 96 of 97 workspaces and on 77% of tracked responses. Evaluation prompts accounted for 38.7% of ad impressions, awareness prompts 24.3%, advice prompts 18.8% and comparison prompts 16.5%.

Sources: Seer Interactive ChatGPT ads study

The overlap between buying and being answered was small. The advertiser was named inside the ChatGPT response 5.4% of the time and cited as a source 3.3% of the time. On prompts naming a specific brand, only 7.3% of responses carried that brand’s own ad, and Seer recorded a competitor running “Looking Beyond [Brand]? Try [Competitor]” under a branded answer. This does not establish how ads perform; it suggests that the organic answer and the paid slot are now read in one glance and won separately.

That creates two distinct risks. A brand absent from the answer can buy a line underneath it, but the answer above still recommends someone else. A brand that the answer already recommends can find a rival’s ad directly below its best moment. Neither risk appears in a report that only counts organic mentions.

Canlah AI evidence: recognised by name, absent from the category answer

Canlah AI ran its own audit pipeline on canlah.ai on September 7, 2026, for Singapore in English. Thirty-nine buyer questions were sent three times each to ChatGPT through the OpenAI API with web search and to Gemini, and Google AI Overviews were sampled as browser-rendered results for the Singapore market. Only grounded answers were counted, and branded questions were never scored with open ones.

Canlah AI self-audit, Singapore, September 7, 2026

Surface Open-question answers Canlah AI named Mention rate Branded answers naming Canlah AI What the number does not show
ChatGPT (OpenAI API, web search) 84 7 8.3% 18 of 18 The API is not the consumer ChatGPT surface where ads are served, so nothing is known about ads under these answers
Gemini (API, grounded) 93 0 0.0% 21 of 21 Organic answers only; no ad data was collected for this surface
Google AI Overviews (rendered, Singapore market) 51 0 0.0% 11 of 11 An Overview appeared on 62 of 65 renders; the three blank renders are outside these counts

Source: Canlah AI evidence archive, September 7, 2026. Open questions do not name Canlah; branded questions do and are disclosed, never scored.

ChatGPT knew the brand every time a question named it and treated it as a category answer in seven of 84 open-question answers. That is the shape Seer’s branded-prompt finding warns about. A branded question is where a competitor’s ad can sit under a correct answer, and an open question is where the brand is missing before any ad is shown.

The citations explain the open-question gap. Across all 39 questions, 382 domains were cited in ChatGPT and Gemini API answers combined, led by agency pages and “best agency in Singapore” listicles. canlah.ai drew six citations on open questions and ranked 46th. A ChatGPT ad could place a sponsored line under those answers in markets where ads are sold. It would not change which domains the answers drew on.

ChatGPT ads tracking: what the reporting can and cannot see

Ads Manager and the Insights API report impressions, clicks, spend, CTR, CPC, CPM, conversions, cost per action, post-click conversion rate, attributed sales and return on ad spend. Conversions arrive through the Measurement Pixel, Conversions API or a no-JavaScript image tag. OpenAI reporting defaults to a 30-day click window and a one-day view window; setting the view window to zero disables view attribution. Total conversions include click-attributed and view-attributed conversions. The selected windows affect conversion reporting, including CPA, attributed sales and ROAS, so record these settings when comparing campaigns.

Sources: OpenAI Ads reporting documentation; OpenAI conversion tracking documentation

None of those metrics describe the answer above the ad. Ads reporting shows what the ad did, not whether ChatGPT named the brand, which sources it cited or whether a rival appeared in the same response. Organic answer measurement needs its own fixed question set, engine by engine, with branded and open questions kept apart. Referral traffic from organic ChatGPT answers and clicks from ChatGPT ads should also be separated in analytics, usually by static UTM parameters on ad landing pages.

What marketing teams should do next

  1. Check eligibility before planning budget: Singapore is on the self-service list checked October 1 and reconfirmed October 2, 2026. Verify the ad account country and advertising/billing legal entity, then check campaign targeting and platform settings separately.
  2. Measure the organic answer first: record, on a fixed set of buyer questions, whether ChatGPT names the brand, recommends it and cites its pages. That baseline shows whether an ad would fill a gap or sit under an answer the brand already wins.
  3. Decide on branded prompts deliberately: Seer’s data suggests most branded prompts are undefended. Where a competitor ad appears under the brand’s own answer, a defensive campaign may be worth a small test.
  4. Write context hints from real questions: use the needs and use cases buyers actually type, taken from sales calls and support logs, rather than broad category words.
  5. Separate paid and organic in reporting: tag ad traffic, keep ads metrics in Ads Manager and keep answer-level metrics in a probe archive. A combined “ChatGPT visibility” number hides both risks.
  6. Keep improving the evidence ChatGPT cites: comparison pages, third-party listicles, reviews and original data shape the answer. No ad changes that set of sources.

Every figure above is dated and re-checkable. Before a budget is committed, the availability list, the bid guidance and the brand’s own answer baseline should be pulled again on the day of the decision, because all three moved during 2026.

Where Canlah AI fits

Canlah AI works on the organic side of that split. Canlah AI is a Singapore-based SEO + GEO agency that measures and improves brand visibility inside ChatGPT, Gemini, Google AI Overviews and Google AI Mode, reporting AI visibility as re-verifiable ranges with timestamped evidence. Its role is to fix the buyer question set, record which sources each engine cites and coordinate the on-site and off-site work that puts a brand into the answer. Canlah AI’s published service pages describe SEO and GEO work, not paid-media buying, and its current probes read ChatGPT through the API, so they cannot show which ad ran beneath a given answer.

Methodology and limitations

OpenAI product details come from its help centre, Ads documentation and announcements through September 23, 2026. Availability and current product documentation were rechecked from official pages on October 2, 2026. The October 1 availability check is recorded in the editorial issue; the October 2 count is a fresh count of rows marked Available. Historical reporting keeps its original dates. Pricing figures are press-reported, not published rate cards. Seer Interactive’s figures cover about two weeks of its own clients’ tracked prompts and may not represent all ChatGPT traffic. Canlah AI figures come from a first-party audit of canlah.ai on September 7, 2026: one site, one market, English only, three API runs per question, 216 grounded API answers and 65 completed AI Overview renders, 62 of which returned an Overview. It did not observe ads.

Frequently asked questions

Can a Singapore business buy ChatGPT ads today?

Yes, subject to eligibility. OpenAI announced the Singapore rollout on September 23, 2026; Singapore was marked Available in the October 1 editorial check of the self-service list, reconfirmed October 2. The legal entity advertising and being billed must be based in an available country. Recheck before purchase.

Does buying ChatGPT ads make ChatGPT recommend the brand?

No. OpenAI states that ads run on separate systems and do not influence ChatGPT’s answers. Seer Interactive found the advertiser inside the organic response on only 5.4% of responses that carried its ad, so paid placement and organic recommendation have to be earned separately.

Does ChatGPT show ads to every user?

No. Ads may appear for Free and Go users in supported markets, including in logged-out experiences. Plus, Pro, Business, Enterprise and Edu accounts do not have ads, and OpenAI does not show ads to accounts it identifies as under 18 or next to sensitive topics such as personal health and politics.

References

Related articles

For the organic side, see Canlah AI’s Singapore GEO services and website audit.

← All articles Run a free website audit →

FREE WHITEPAPER

Marketing in the Agent Era

All 13 chapters public — no email wall. Includes an original dataset on the agent-readiness of 50 cross-border DTC brands.

Read it free →

KEEP READING