OpenAI names 20 ChatGPT Ads measurement partners and tests image-led ads in the US

OpenAI said on 5 October 2026 that ChatGPT Ads now works with 20 named third-party measurement partners, from attribution vendors such as AppsFlyer, Adjust and Branch to incrementality specialists Haus and WorkMagic, and that DoubleVerify and Integral Ad Science are building brand suitability pilots for the product. In the same announcement OpenAI introduced an image-led ad format that it will first test inside ChatGPT's image generation, starting later in October with an initial group of US advertisers. For Thai advertisers, who gained self-serve access on 23 September, the measurement roster is the part that affects current campaigns; the visual format is a US test with no date for any other market.

What OpenAI announced on 5 October 2026

OpenAI published two linked posts on 5 October 2026: a product post on its main site titled "Building advertising for the way people use AI", and a longer entry on its Ads blog titled "More ways to measure ChatGPT Ads". The OpenAI announcement covers three things: a new visual ad format, a larger set of measurement tools and partners, and work on brand suitability. Both posts open with an audience figure. OpenAI says ChatGPT reaches 1.2 billion people each week and calls it the world's largest AI-native consumer platform.

OpenAI describes its measurement approach as supporting both in-house and partner-led measurement. That covers conversion signals sent through its Pixel and Conversions API (CAPI), attribution, incrementality and brand measurement. The company calls the firms named on 5 October its "first set of third-party partners" across those areas.

The announcement arrives about eight months after the paid US pilot began on 9 February. By OpenAI's own figures, the ad business passed a US$1 billion annualised run rate at the end of August, and the platform now spans more than 60 countries. Digiday listed unproven measurement first among five barriers to the business's growth, and most of the 5 October material is aimed at it.

The measurement partners, by category

OpenAI grouped its partners by the job each one does. The table below lists them exactly as OpenAI named them on 5 October 2026.

Several of those names matter most to app marketers. AppsFlyer, Adjust, Branch, Singular, Kochava, Airbridge and Tenjin are mobile measurement partners, so an advertiser running app-install or in-app purchase campaigns can, according to OpenAI, now connect ChatGPT Ads through the same vendor that already attributes its other paid channels. Hightouch, Tealium and LiveRamp sit one step earlier in the chain. OpenAI says they make it easier to send conversion data from systems an advertiser already runs, which in practice can mean less custom integration work.

Measured appears twice, once among the full-funnel partners and once in the incrementality group. OpenAI did not say which partners are live in which countries. It also did not say whether any integration needs a minimum spend, a separate vendor contract or approval from OpenAI before it can be switched on.

The early results, and who reported them

OpenAI's Ads blog shows five early results, four of them credited to a named partner, and OpenAI says they were validated by its partners. Read them as what they are: figures chosen by the platform for its own announcement, published with no spend levels, test dates, sample sizes or methods. The trade publication PPC Land made the same point in its coverage on 5 October.

• WeightWatchers, measured by DV Rockerbox: attributed cost per acquisition on ChatGPT Ads was 15.3% below the brand's blended paid-search benchmark.
• Dose, measured by WorkMagic: a geo-lift study found statistically significant lift, with 67% of incremental purchases coming from net-new customers, and estimated 2.3 times more incremental orders than last-click attribution captured.
• Portland Leather Goods, measured by Triple Whale: 93% of visitors arriving from ChatGPT Ads were new visitors.
• Fospha's Daily Media Mix Attribution: ChatGPT generated 83% more revenue per 1,000 impressions than the paid social average across the same advertisers.
• Incrementality across verticals: one slide states that 100% of campaigns delivered incremental conversions, with statistically significant lift in travel, fintech, software and other verticals. The slide text does not say how many campaigns were tested or which partner measured them.

OpenAI added one figure from its own analysis. In early data from global campaigns optimised for deeper-funnel outcomes, 52.7% of eligible one-day view-through conversions happened within an hour of the matched ad impression. That number describes timing. On its own it does not show that the ad caused the purchase; that question is what the incrementality studies are for.

Each partner figure answers a different question. A CPA comparison against paid search is an attribution result and carries the assumptions of the attribution model behind it. A share of new visitors says something about reach into new audiences and nothing about cost. A geo-lift study is the closest thing in the list to causal evidence, and it covers one brand. None of the five should be treated as a forecast for a Thai account.

View-through conversions now feed optimisation

One line in the Ads blog post changes how a campaign learns. OpenAI wrote on 5 October 2026 that advanced conversion optimisation learns from eligible click-through and view-through conversions for the goal an advertiser selects, so purchases that follow either path can inform future ad delivery.

That differs from the position Relevant Audience reported on 20 August 2026, when OpenAI's documentation added a one-day view-through metric, shown as VTA (1d), and kept it out of the main conversions total, cost per acquisition, bidding, billing and conversion optimisation. The new post does not say when the change took effect, which campaigns count as using advanced conversion optimisation, or whether view-through conversions now appear in the headline conversions column.

OpenAI also separated reporting from learning. Advertisers can pick a click-through attribution window that fits their buying cycle and add a one-day view-through window, and OpenAI states that these settings control attribution in reports but do not change how campaigns optimise. Changing the reporting window will therefore change the numbers a team sees, not the way the system delivers.

The post also points to Event Quality Score (EQS), which helps advertisers judge the quality of the conversion signals they send through the pixel and CAPI. OpenAI's argument is simple: the optimiser can only learn from the events it receives, so weak or duplicated signals limit both reporting and delivery.

Brand suitability pilots with DoubleVerify and IAS

OpenAI said it is developing brand suitability evaluation pilots with DoubleVerify (DV) and Integral Ad Science (IAS). According to the announcement, the two independent firms will assess how OpenAI's brand safety standards are applied in a controlled testing environment, without access to private user conversations. Both chief executives are quoted. Lidiane Jones of IAS said the evaluation will test how OpenAI's safeguards perform across a broad and changing range of scenarios. Mark Zagorski of DoubleVerify described ChatGPT as an environment where context is set by the conversation rather than by a page or a video.

IAS issued its own statement the same morning. As PPC Land reported, IAS called itself one of the first media quality partners supporting ChatGPT Ads, said the service runs with a select group of advertisers, and gave no date for wider availability. Neither company published a start date, a pilot size or a list of markets.

OpenAI described two further controls. Its placement guardrails assess whether a conversation is appropriate for advertising and aim to keep ads away from emotionally vulnerable, sensitive or otherwise unsuitable contexts, backed by automated review, human oversight and ongoing monitoring. For narrower requirements tied to a brand's own policies, OpenAI said Negative Phrases are now available to qualifying advertisers. It did not publish the qualifying criteria.

The new image-led ad format

The second part of the announcement is a visual ad unit. OpenAI says the format uses images showing product inspiration, product usage, or the experiences a product makes possible, so that people can discover a brand and see what makes an offering distinctive. The first test will run during image generation in ChatGPT. OpenAI says these ads will be clearly labelled and kept separate from the image the user is creating, and it repeats that advertising does not influence the answers ChatGPT gives.

Testing begins later in October 2026 in the US with an initial group of advertisers. OpenAI gave no pricing, no list of participating advertisers, no creative specifications and no date for other markets, and it did not say whether the format will be bookable through Ads Manager. PPC Land noted that when it reviewed OpenAI's Ads Manager help documentation in April, the text said ads did not appear after a user generated an image, so this test reverses that earlier rule.

The Ads blog also says OpenAI is "experimenting with new placements" and is in early-stage testing with Kantar and Cint to measure the effect of ChatGPT Ads on brand resonance. OpenAI did not link those brand studies to the visual format directly. A picture-led unit shown while someone is making an image is, however, the kind of placement usually judged on brand lift rather than on last-click sales, so the two pieces of news are likely to be read together by buyers.

What is new compared with RA's earlier ChatGPT Ads coverage

Relevant Audience has followed this product since its pilot. Three earlier posts set the baseline for what changed on 5 October 2026:

• On 24 September 2026 RA covered the Thai launch. OpenAI began rolling out ChatGPT Ads in Thailand on 23 September, legal entities based in Thailand can buy through self-serve Ads Manager, and ads show only to Free and Go users aged 18 and over. Nothing in the 5 October posts changes those terms.
• On 1 October 2026 RA summarised Digiday's reporting that OpenAI's ad business had reached a US$1 billion annualised run rate, with unproven measurement listed first among five barriers to its growth goals. The partner roster is aimed squarely at that barrier.
• On 20 August 2026 RA reported the one-day view-through metric as reporting-only. The 5 October post says view-through conversions now inform advanced conversion optimisation.

The new elements are the named partner list, the first incrementality and brand-lift programmes, the DoubleVerify and IAS pilots, Negative Phrases, the optimisation change and the visual format. Some names on the list are not new either: PPC Land notes that LiveRamp connected its CAPI Hub in June and that AppsFlyer and Adjust went live for app attribution in seven markets in July. The Conversions API itself is older; PPC Land notes the pixel and CAPI arrived with self-serve Ads Manager on 5 May 2026.

What this means for Thai marketers

For Thai advertisers already testing ChatGPT Ads, the first practical question is whether the account's existing measurement vendor is on OpenAI's list. A brand that attributes app installs through AppsFlyer, Adjust, Branch, Singular, Kochava, Airbridge or Tenjin, or that sends customer data through Tealium, Hightouch or LiveRamp, should ask that vendor whether its ChatGPT Ads integration is enabled for Thai accounts, because OpenAI did not publish availability by market.

Second, check the conversion signal before reading any result. By OpenAI's own logic, Event Quality Score and correct deduplication between pixel and CAPI decide what the optimiser learns from, and that matters more now that view-through conversions feed advanced optimisation. Third, treat the partner case studies as illustrations. The 15.3% lower CPA against paid search came from one brand measured with one vendor's model; a Thai account needs its own comparison, ideally against its Google Ads search campaigns on the same conversion definition. Teams that use GA4 as their source of truth should also confirm how ChatGPT traffic is tagged and attributed there, and a GA4 setup review is a sensible place to start.

The image-led format is not available in Thailand, and OpenAI has given no date. Thai brands can still prepare by deciding which product imagery would make sense next to a picture a user has just generated, since that is where the format will appear first.

FAQ

Is the new ChatGPT image ad format available in Thailand?

No, OpenAI said testing starts later in October 2026 in the US only, with an initial group of advertisers. It gave no date for Thailand or any other market and did not say whether the format will reach self-serve Ads Manager.

Which measurement partners work with ChatGPT Ads?

OpenAI named Hightouch, Tealium and LiveRamp for sending conversion data; AppsFlyer, Triple Whale, Adjust, DV Rockerbox, Northbeam, Branch, Singular, Kochava, Airbridge and Tenjin for attribution; Fospha, Measured and INCRMNTAL for full-funnel measurement; Haus, Measured and WorkMagic for geo-based incrementality; and Kantar and Cint for brand resonance tests. OpenAI did not say which integrations are enabled for Thai accounts, so confirm with the vendor.

Are the reported results independent?

They are partner-reported figures that OpenAI chose to publish, not independent audits. OpenAI says the partners validated them, but no spend, dates, sample sizes or methods were released alongside.

Do view-through conversions affect optimisation now?

Yes, for advanced conversion optimisation, according to OpenAI's 5 October 2026 Ads blog post, which says the system learns from eligible click-through and view-through conversions. The reporting window settings, by contrast, only change what appears in reports.

Do advertisers have to change anything?

OpenAI did not announce any required action. Sensible steps are to check whether your measurement vendor supports ChatGPT Ads, review Event Quality Score, and confirm that pixel and CAPI events are deduplicated.

ChatGPT Ads now has much of the measurement plumbing buyers asked for, while the evidence that it pays off is still vendor-supplied. If you want a second pair of eyes on tracking or on the design of a first test, Relevant Audience's ChatGPT Ads service can review the setup with you.