Ready-to-use OpenAI Ads Manager connector

Ads inside ChatGPT are one half of the shift. The other half is what happens to the reporting work behind them, and that's the part most teams haven't solved yet.

Whatagraph's IQ Agents is built for exactly that: an always-on team of agents that builds each client's reports from your templates, then keeps watch on the data, catching a broken connection, a drifting metric, or an off-pace goal before your client spots it. It's in early access now, opening by cohort.

Request early access if you want to be in the next one.
 

Here they are:
 

Generate an API key in OpenAI Ads Manager under Settings, then paste it into Whatagraph when you add the source. That's the whole setup, no code and no third-party connector in the middle.

A few things worth knowing before you start:

- Each API key is scoped to one ad account, so reporting on several accounts means one key and one source per account

- Regenerating or revoking a key in Ads Manager stops the source updating, so reconnect with the new key if you rotate them

- Data starts on 9 Feb 2026, when the platform's pilot began, so earlier date ranges come back empty

Once connected, your account, campaign, ad group, and ad data refreshes on a schedule and flows into every report that uses it.
 

You get access to an audience that isn't sitting on a search results page. Ads appear inside conversations where someone's already describing a problem, so your offer shows up at the moment intent is forming rather than after someone types a keyword.

For digital marketing teams, the practical upside looks like this:

- A new inventory pool while CPCs on Google Ads and Meta Ads keep climbing

- No minimum commitment in the self-serve beta, so you can test with a small daily budget before you scale

- Familiar structure, since the campaign, ad group, and ad hierarchy is the one your PPC team already runs

- Standard performance metrics, so CTR, CPC, CPM, and cost per conversion compare cleanly with your other channels

That last point only pays off if the numbers actually sit side by side, which is where a reporting layer comes in.
 

Targeting is conversational, not keyword-based. Instead of bidding on exact-match terms or building interest audiences, you write context hints at the ad group level that describe the topics and situations where your product is relevant.

OpenAI's selection system weighs those hints alongside your ad title, ad copy, and landing page, then decides whether your ad fits the conversation. Layer country, state, DMA, and ZIP code targeting on top and you control where ads run as well as when they're relevant.

The trade-off is less direct control than a keyword list gives you, so reporting matters more. In Whatagraph, you can watch CTR and cost per conversion by ad group to see which context hints are actually pulling their weight.
 

The AI does the matching. It reads the intent in a live conversation and picks the ad that fits, which is why relevance depends on your context hints, creative, and landing page rather than on a bid on a keyword.

To improve performance, feed the system better inputs and read the results faster:

- Write specific context hints for each ad group rather than one broad set for the whole campaign

- Match your landing page to what the hint promises, since the selection system looks at it too

- Test ad formats and copy variations, then compare CTR and cost per conversion side by side

- Set up the Conversions API or pixel so conversion data, not just clicks, feeds your decisions

Whatagraph handles the reading part. Ask IQ Chat which ad groups moved this week, or let IQ Summaries write the performance narrative straight into your client report.

Here are the key differences:

Here are the key differences:

Start small, read the data often, and treat the first month as learning rather than scaling.

- Split ad groups by intent, so each set of context hints stays specific enough to judge

- Start with a daily budget to pace spend while you learn, then move to a campaign-total budget once performance holds

- Set up the Conversions API or pixel before launch, not after, so you're never optimizing on clicks alone

- Add tracking parameters to your landing page URLs, so sessions and conversions show up in GA4 next to the spend

- Give campaigns time to gather data before you judge them, since the auction is still young and benchmarks are moving

- Report weekly, not monthly, because the platform ships changes fast and your numbers move with them
 

Most of the early mistakes come from treating it like a mature search platform.

- Writing context hints like keyword lists, which gives the selection system nothing to work with

- Pointing ads at a generic homepage instead of a landing page that matches the hint

- Judging performance on CPC alone while conversion measurement sits unconfigured

- Running one broad ad group, then having no way to tell which theme worked

- Reporting OpenAI Ads in isolation, so nobody can see what it did to blended CAC or ROAS

- Assuming historical data exists before 9 Feb 2026, which it doesn't

Yes, you can easily blend data on Whatagraph. Blends join OpenAI Ads with Google Ads, Meta Ads, Criteo, LinkedIn Ads, GA4, and 60+ other sources on a shared key like date, so you get one unified dataset instead of a stack of separate tables.

From there you can build the metrics that actually matter across channels:

- Blended ROAS, CAC, and CPL with OpenAI Ads spend included

- Cost-per-click and CTR compared channel by channel in one widget

- Custom dimensions that standardize campaign naming across every platform

- Change currency setting when accounts run in different currencies
 

Not on Whatagraph. You get a governed data foundation with the full white-labeling reporting and visualization for clients and internal team members. 

Peak Seven saved 63 hours a month after moving their reporting into Whatagraph, and cut report build time from four hours to 90 minutes.

Yes. Build the report once on Whatagraph, save it as a template, and re-use it the next time you onboard a new client or account for OpenAI Ads. You can also:

- Schedule email delivery daily, weekly, or monthly

- Share a live link, password-protected and on your own domain

- Export to PDF or CSV when someone wants a file

- Push data to Looker Studio, BigQuery, or Snowflake for deeper analysis

- Add goals and alerts so Slack or email tells you when conversions drop or spend runs hot
 

Pick the destination and Whatagraph handles the transfer. Email a scheduled report, share a live link, or send raw OpenAI Ads data to Looker Studio, BigQuery, or Snowflake, then set the cadence once and leave it.

Your historical data builds up in one place too, so month-on-month comparisons don't depend on somebody remembering to save last quarter's export.