ChatGPT Ads Adds Conversion Bidding, App Attribution, and Competitive Intelligence in Rapid Measurement Build-Out

OpenAI shipped eight new capabilities to its ChatGPT Ads Manager on July 24, spanning bidding, measurement, budget management, and campaign operations. The update includes conversion-optimized cost-per-click bidding, integrations with mobile measurement partners AppsFlyer and Adjust, and automatic advanced matching for improved conversion attribution. A day earlier, Dentsu and Adthena announced the first competitive intelligence product built specifically for ChatGPT's ad auction, filling an analytics gap that left advertisers blind to why their CPCs were swinging by as much as 278%.

Taken together, the announcements mark the most significant week for ChatGPT's advertising measurement infrastructure since OpenAI launched its self-serve Ads Manager on May 5 with a conversion pixel and Conversions API. ChatGPT ads now appear in roughly 51% of US replies to free-tier users and are live in seven countries. The platform has gone from zero measurement to conversion-optimized bidding in under six months — a pace that puts pressure on measurement teams to evaluate a channel that is scaling faster than the tools to audit it.

Conversion-Optimized Bidding Closes the Performance Gap

The most consequential update is conversion-optimized CPC, or oCPC. Advertisers can now select a Conversions objective when building campaigns, which pushes delivery toward clicks more likely to produce a conversion event while billing continues on a per-click basis. This is the same outcome-optimization model that Google Ads and Meta have operated for years — it uses machine learning to predict which impressions are most likely to drive downstream actions and bids accordingly.

Until now, ChatGPT Ads operated on a simpler model: CPM bidding (launched in February) and flat CPC bidding (added in May). Advertisers could pay for impressions or clicks, but the platform had no mechanism to optimize delivery toward actual conversions. That meant advertisers were effectively buying reach without the algorithmic targeting that makes performance marketing work on Google and Meta.

The oCPC update relies on conversion signals flowing back to OpenAI through its OAIQ pixel or Conversions API. Advertisers who have not implemented conversion tracking will not benefit — the algorithm needs data to optimize against. This creates a measurement prerequisite: the more conversion data OpenAI receives, the better oCPC performs, which in turn drives more advertiser adoption of its tracking tools.

App Attribution Arrives via AppsFlyer and Adjust

OpenAI also connected ChatGPT Ads to AppsFlyer and Adjust, the two largest mobile measurement partners, giving app advertisers install and in-app event attribution for the first time since the ad pilot began in February. The integrations are live in seven markets: the US, Canada, the UK, Australia, New Zealand, Japan, and South Korea.

The timing is notable. On June 22, Google, Meta, Moloco, and Unity invested more than $1 billion in AppsFlyer at a $2.7 billion valuation in a deal explicitly structured to preserve measurement neutrality — all four investors contractually committed to non-preferential treatment in AppsFlyer's attribution logic. Adding OpenAI as an integrated network one month later extends that neutrality claim to a platform none of the four investors control.

For app advertisers already running campaigns on Google, Meta, and TikTok, the practical effect is straightforward: ChatGPT install and in-app conversion data now flows through the same attribution pipeline, using the same rules, as every other channel. Advertisers can generate attribution links through AppsFlyer or Adjust, use those links in ChatGPT campaigns, and see conversion events in both the measurement partner's dashboard and OpenAI's Ads Manager.

Dentsu and Adthena Build the Competitive Layer OpenAI Did Not

While OpenAI focused on first-party measurement, Dentsu and Adthena addressed a different gap: competitive intelligence. ChatGPT's Ads Manager shows advertisers their own performance metrics — CPC, impressions, click-through rate — but provides no visibility into the competitive landscape. Advertisers have no native way to see who else is bidding in their categories, when a new competitor entered the auction, or why costs suddenly changed.

That blind spot proved expensive as the UK auction opened. According to Adthena, CPCs in the UK ChatGPT auction swung by 39% to 278% with no explanation available to advertisers inside the platform. Dentsu, as one of OpenAI's four UK agency launch partners with live campaigns for brands including Hilton, saw the problem firsthand — three of its clients are currently among the nine largest ChatGPT advertisers in the UK.

Adthena's Decision Intelligence product monitors paid placements across the ChatGPT ecosystem, identifies which competitors have entered specific prompt categories and when, quantifies the effect on auction prices, and converts that into recommended actions: bid caps on inflated categories, budget redirection to uncontested prompt groups, and creative strategy for high-value placements. It is, in effect, the auction analytics layer that Google Ads has provided natively through Auction Insights for over a decade but that ChatGPT's nascent platform lacks.

Advanced Matching and Operational Upgrades

The July 24 update also includes automatic advanced matching, which uses hashed customer data — such as email addresses — to improve conversion attribution by matching more user interactions to downstream events. This is the same mechanism Meta introduced years ago with Advanced Matching in its pixel: when a user interacts with an ad but does not click through in a trackable way, hashed first-party data can still connect the impression to a conversion.

Other operational updates include average daily budgets calculated over a rolling seven-day period, geo exclusions for targeting refinement, and a Bulk API for managing larger campaign volumes. These are not measurement features per se, but they reduce the operational friction that has kept performance marketers from testing ChatGPT at scale. A channel that cannot be managed efficiently at volume will not attract the budgets needed to generate statistically meaningful conversion data.

What Is Still Missing

Despite the rapid build-out, ChatGPT Ads still lacks the third-party verification ecosystem that Google and Meta have spent years constructing. OpenAI has confirmed that third-party measurement is in development but has named neither partners nor a timeline. There is no independent viewability measurement. There are no brand safety verification partners integrated into the platform. And OpenAI has not formally published its attribution window lengths — advertisers are optimizing against conversion data without documentation of how long the platform's lookback window extends.

The absence of third-party verification is not a theoretical concern. When LiveRamp became the first independent measurement partner in June via its CAPI Hub, it addressed server-side conversion matching but not the broader verification stack. Advertisers relying on DoubleVerify, IAS, or MOAT for independent ad quality measurement across their other channels have no equivalent for ChatGPT.

The ad penetration data itself illustrates the platform's volatility: ChatGPT ads showed in 26.5% of US replies in late May, collapsed to 0.05% in mid-June, and climbed past 51% within weeks. That kind of variance makes campaign forecasting unreliable and complicates any attempt to benchmark performance against stable channels.

What Measurement Teams Should Do

Implement conversion tracking before testing oCPC. The conversion-optimized bidding update is only useful if you are sending conversion signals back to OpenAI. If you have not deployed the OAIQ pixel or Conversions API, do that first. Without conversion data, oCPC cannot optimize and you are effectively running flat CPC campaigns under a different label.

Use AppsFlyer or Adjust as your attribution source of truth, not OpenAI's dashboard. The MMP integrations give you independent attribution data for app campaigns. Compare what AppsFlyer or Adjust reports against what OpenAI's Ads Manager shows. Discrepancies between the two will tell you how aggressively OpenAI's self-attribution model counts conversions relative to your MMP's rules.

Build competitive monitoring now if you are spending in ChatGPT. Whether you use Adthena's Decision Intelligence or build your own tracking, CPC volatility of 39% to 278% without visibility into the cause is an unacceptable measurement environment for scaled spend. At minimum, log your CPCs by prompt category daily and flag anomalies before they eat through budget.

Do not scale ChatGPT spend past pilot levels until third-party verification arrives. OpenAI has not named a single viewability or brand safety partner. Until independent verification is available, you have no way to confirm that your ads were actually viewable, served in a brand-safe context, or reached a human. That is a risk that increases linearly with spend.

The speed at which OpenAI is building ChatGPT's measurement stack is impressive by any standard — from zero infrastructure to conversion-optimized bidding in six months. But speed and completeness are different things. The platform now has enough measurement capability for performance-oriented pilot campaigns, and the AppsFlyer and Adjust integrations give app advertisers a credible attribution pathway. What it does not yet have is the independent verification infrastructure that separates a channel you test from a channel you trust with material budget. The advertisers watching that gap most closely will be the ones best positioned when it closes.