What’s inside
- The new channel marketers can’t measure with old tools
- What shipped in Extract: Reporting + Conversions APIs
- Why it matters for your stack
- How it works
- What early customers are seeing
- How to get started
For the first time, marketers can buy ads inside the world’s most-used AI assistant. With OpenAI’s launch of self-serve Ads Manager, CPC bidding, a measurement pixel, and a Conversions API [CAPI], ChatGPT has gone from controlled pilot to a real performance channel in a matter of months.
The harder question: how do you actually know how your ChatGPT campaigns are performing, and how do you make them work better over time?
ChatGPT is now another ad platform, like Google, TikTok, or Meta, with cost and performance data flowing automatically into your platform of record, and clean conversion signal flowing back to OpenAI to power its bidding algorithms.
That’s what Extract has built. We’ve shipped a full integration with OpenAI’s ChatGPT Ads platform, and it’s live for every Extract customer today.
“ChatGPT is the first genuinely new ad channel in years, and it’s emerging right as AI is reshaping how every buy-side algorithm makes decisions. We’re proud that Singular customers are the first to seamlessly integrate their conversions for maximal performance and analyze their ChatGPT performance metrics directly in their BI. The advertisers who win in this new channel will be the ones who can both measure ChatGPT performance alongside every other channel and feed clean, structured conversion signal back into OpenAI’s system from day one. That’s exactly what we built.”
Eran Friedman, CTO, Singular & Extract
What Extract shipped
Two integrations, both available now:
Automated spend and performance reporting from ChatGPT Ads.
Granular spend, impression, click flowing into Singular alongside every other channel you measure, broken out at the account, campaign, ad group, and ad level.
Automated Conversions API for ChatGPT Ads.
Send conversion events from your BI stack or data warehouse (powered by Singular), to OpenAI to power CPC optimization and tighten the feedback loop between spend and outcome.
Together, they close the loop: clean data in, clean signal out, all in one platform.
- Data trust erosion: Analysts lose confidence when visualizations or KPIs break.
- Operational risk: Teams spend hours debugging instead of building insights.
- Hidden cost: Reprocessing large datasets can multiply compute costs.
Why this matters for your stack
ChatGPT is a categorically new ad environment, and advertisers naturally want to start advertising on ChatGPT. Users arrive with explicit intent, comparing options, weighing tradeoffs, and deciding what to do next. That’s why OpenAI’s shift to CPC matters. It’s also why early benchmark data suggests why advertisers can’t afford to fly blind on this channel.
For growth teams running cross-channel, the stakes are sharper. Every new channel that lands outside your measurement stack is a hidden tax on growth. You can’t compare ROAS. You can’t trust the numbers in your weekly review. And AI-driven optimization on the buy side, including OpenAI’s own bidding algorithms, only works as well as the signal you feed back into it. Bad data in, bad decisions out, at speed.
Here’s what each stakeholder gets:
- For data and BI teams:
Clean, normalized data flows into your warehouse via ETL alongside the rest of your marketing stack. No bolt-on dashboards. No separate pipeline. And when your team is ready to send conversion events back to ChatGPT, you can do it straight from the warehouse. - For OpenAI’s algorithms:
Higher-fidelity conversion signal via CAPI, which means better bidding decisions, better relevance, and better performance over time.
How it works
Key capabilities
- Conversions API from your warehouse or BI.
Push first-party conversion events from your BI stack or data warehouse through Extract to OpenAI. Useful for offline conversions, downstream LTV events, or any signal that lives outside Singular’s attribution graph but matters for OpenAI’s optimization. - Unified cross-channel ROI.
Because ChatGPT’s spend lands in the same schema as your other 1,000+ cost sources, ROAS, eCPI, and LTV reporting work the moment you connect the integration. No new dashboards. No manual reconciliation.
Both paths respect OpenAI’s privacy model. Aggregated, attributed performance data flows back to optimize delivery.
For full implementation steps, event mapping guidance, and configuration details, see the Extract help center
How to get started
Setting up the OpenAI Ads Conversions API connection in Extract
Make sure your ChatGPT ad account is active
Confirm your ChatGPT Ads Manager account is active and your campaigns are running – you’ll need your API key and Pixel ID.
Setup your destination
Log into Extract data destinations from the Singular platform and go to Extract’s Catalog.
- Add OpenAI Ads Conversions API as a reverse ETL destination.
- Add your Conversions API key and your Pixel ID
Configure your source
Configure your warehouse source [Redshift, BigQuery, Snowflake, etc.] and define the event data you want to be sent
Create your connection
Go to Connections and “Add New Connection.” Select the source you just configured and select the OpenAI Ads Conversions API destination you created.
Configure your sync frequency.
Map your data
Go to the Streams tab and map your dataset fields.
The bigger picture
Every measurement era is defined by the channels that emerge inside it. SKAdNetwork reshaped iOS measurement. CTV reshaped cross-device. Conversational AI is the next one, and it’s moving faster than either of them did.
Extract’s job is to make sure every new channel, every new signal, and every new optimization surface is available to advertisers the moment it matters, in the same platform of record they already trust. ChatGPT cost reporting and CAPI are the latest examples. They won’t be the last.
