Key takeaways
- OpenAI oCPC campaigns optimise delivery towards one selected standard conversion event while still charging per valid click.
- The main risk for UK advertisers is poor event selection, because the chosen conversion event cannot be changed after campaign creation.
- Lead-gen advertisers should avoid optimising to shallow form fills unless those leads are qualified and commercially useful.
- Ecommerce advertisers need to check whether purchase events reflect profit, returns and stock reality before scaling product-feed oCPC.
- The right first move is a tracking audit, not a budget increase.
OpenAI oCPC campaigns put conversion tracking under the microscope. The feature is not just another bidding option. It changes the bargain advertisers make with the platform: you still pay for clicks, but OpenAI decides which clicks deserve delivery based on one selected conversion event.
That makes signal quality the commercial issue. If your event is wrong, late, duplicated or too shallow, the system optimises spend towards the wrong people. We see the same mechanism across automated bidding in mature paid media accounts. In our accounts, UK ecommerce campaigns carrying a serious tracking fault tend to report flattering CPAs, because fake or double-counted conversions deflate the number and hide keyword waste. Missing revenue does the opposite: it deflates reported ROAS and makes a decent campaign look like a dog. Both errors push automation the wrong way.
For UK PPC teams testing OpenAI Ads, the lesson is blunt. Do not treat OpenAI oCPC campaigns as a media buying shortcut. Treat them as a measurement test first, then a scaling test.
What has changed with OpenAI oCPC campaigns
OpenAI has documented conversion-optimised cost-per-click campaigns for its Ads API. The campaign uses bidding_type: conversions, optimises delivery towards one tracked conversion event, and still charges per valid click rather than per conversion.
The feature is in open beta for standard and product-feed campaigns. Advertisers need conversion tracking via the JavaScript Pixel, the Conversions API, or both. The Conversions API is positioned as the more reliable source than the pixel alone. Each campaign needs exactly one active standard conversion event, and custom events are not accepted as optimisation goals.
Two constraints matter operationally. You cannot switch an existing CPM or CPC campaign into oCPC, and you cannot change the selected conversion event after campaign creation. If the goal is wrong, you rebuild the campaign. That is not a small admin point. It forces teams to decide what the account is really optimising for before money starts moving.
Why the payment model creates the real risk
The commercial risk sits in the gap between how you pay and what the system optimises. OpenAI oCPC campaigns still spend on clicks, so a badly chosen event does not stop spend. It just teaches the system to buy more of the wrong clicks.
Here is the mechanism. If a lead-gen advertiser sets registration_completed as the optimisation event when sales quality actually depends on qualified_call_booked, the platform learns from a shallow proxy. It favours users who complete forms, not users who convert into pipeline. CPC billing keeps charging for each valid click. The dashboard looks active, conversions arrive, and the sales team gets weaker enquiries. The money has moved from buying intent to buying form-fill behaviour.
Ecommerce teams face the same issue with product-feed campaigns. If the selected event is order_created but revenue values, cancellations and stock constraints are not properly fed back into reporting, the campaign treats every tracked order as equally useful. A full-margin repeat purchase and a low-margin discounted order both become the same kind of signal if the event design is too blunt. That flattens the account’s view of profit.
Single-event optimisation raises the stakes
Google Ads teams are used to having more ways to segment goals, values and campaign types. OpenAI’s beta design is narrower. One active standard event does the steering. That makes the pre-launch decision more important than the bid.
This is where paid media teams get caught. They spend hours debating creative, product sets and daily budgets, then accept whatever event name appears in the tracking setup. That is backwards. In OpenAI oCPC campaigns, the event is the steering wheel. Creative and bids are the pedals.
The no-change rule also creates a testing cost. If you discover after launch that lead_created attracts cheap but poor leads, you cannot simply swap the campaign goal. You need a new campaign, new learning period and clean separation in reporting. Any team without naming conventions, event governance and launch notes will lose the ability to compare tests properly.
This connects directly to the tracking hygiene problems we covered in GA4 campaign diagnostics for PPC. Automation does not forgive bad measurement. It scales it. The platform will always find more people who look like your tracked converters, even when those converters are only valuable on paper.
PPC Geeks’ View
The problem advertisers will face is false confidence from cheap conversions. OpenAI oCPC campaigns make that more dangerous because the selected conversion event cannot be changed after creation. A poor event choice gets locked into the campaign architecture, not just the report.
Do not launch conversion bidding until the event has passed a quality test. If the event would not justify a budget increase in your board report, it should not train the bidding system.
— Siobhain McConnell, Senior Client Manager, PPC Geeks
In real campaign work, we see this most often in lead-gen accounts running automated bidding against form submissions, with no offline qualification imported back into the ad platform. The account looks efficient. The sales team says lead quality is soft. The PPC report says CPA is down. That disagreement is not a reporting nuisance. It is the signal that automation is optimising against the wrong business outcome.
For ecommerce, the practical observation is slightly different. Retailers often have purchase tracking in place, but margins, refunds, returns and repeat customer value sit outside the ad platform. That means a campaign optimised for purchases rewards volume before profit. Our view is simple: if value quality is missing, run conservative budgets until the data is fit for bidding.
This is exactly the type of issue we look for in a free Google Ads audit, especially where automation, tracking or campaign structure is affecting performance. The platform name changes, but the failure pattern is familiar: weak signals create strong-looking waste.
What advertisers should do next
- Pick the event before you build the campaign. Write down the commercial reason for choosing order_created, lead_created or registration_completed. If the event does not represent a real business outcome, do not use it for OpenAI oCPC campaigns.
- Check the event source. Use the Conversions API wherever your technical setup allows it, then compare it against pixel activity. A pixel-only setup leaves too much room for browser loss, consent gaps and delayed firing. If the two sources disagree, fix the source before increasing bids.
- Build the campaign paused first. Create the campaign, ad groups and ads in paused status, then check event IDs, billing_event_type, product_feed_id and product filters before activation. This matters because the selected conversion event is fixed after creation.
- Separate test goals into separate campaigns. Do not blend lead_created and registration_completed logic into one budget decision. If the business wants to compare them, create separate campaigns with clear naming and controlled budgets.
- Treat the CPA bid as an optimisation input. The ad group max_bid_micros functions as the CPA bid in oCPC, but billing remains click-based. Set it from expected value and close rate, not from the lowest CPA your board wants to see.
- Set a minimum data threshold before judging performance. Do not make large bid, budget or creative changes after a handful of conversions. Pull impressions, clicks, spend, CTR, CPC and conversions together, then calculate cost per conversion from spend divided by tracked conversions.
- Read the setup rules line by line. The official OpenAI conversion optimised campaign documentation confirms the beta scope, the one-event rule, product-feed support and click-based billing. Use it as your launch checklist, not as background reading.
If you already run target-based bidding elsewhere, reuse the discipline, not the settings. Our piece on the Target CPA and ROAS split explains the same core point for Google Ads: bid strategy changes only work when campaign goals, value rules and reporting match the decision the system is being asked to make. If you would rather hand the setup to a team that does this daily, our Google Ads agency team can run the test with you.
What this means for your campaigns
OpenAI oCPC campaigns are a serious signal that ads inside AI-led environments are moving quickly towards conversion-based delivery. The mistake is to assume the platform has solved efficiency for you. It has not. It has given you a new way to spend click budget against a conversion model, and that model is only as good as the event behind it.
The immediate priority for UK advertisers is not bigger tests. It is cleaner tests. Define the event, verify the source, document the campaign goal, and keep budgets tight until performance data proves the campaign is attracting the right customers rather than easy converters.
If your tracking is weak, OpenAI oCPC campaigns will expose it. If your tracking is clean, they give you a controlled route to test intent in a new ad environment without switching to pay-per-conversion economics.
If you would like a second pair of eyes on how this affects your account, our team offers a free Google Ads audit, with no strings. For wider context on where UK spend is heading, the IAB reports that the UK digital ad market reached £40.5bn, and OpenAI’s own Ads API developer documentation sets out the wider feature set behind this beta.
Frequently asked questions
What are OpenAI oCPC campaigns?
OpenAI oCPC campaigns are conversion-optimised cost-per-click campaigns in the OpenAI Ads API. They optimise delivery towards one tracked conversion event, but advertisers still pay for valid clicks rather than paying per conversion.
Do OpenAI oCPC campaigns charge per conversion?
No. Billing remains click-based. The selected conversion event guides delivery and bidding, but the advertiser is charged when a valid click occurs.
Can I change the conversion event after launch?
No. OpenAI states that the selected conversion event and campaign goal cannot be changed after campaign creation. To optimise for a different event, you need to create a new campaign.
Which advertisers should be most careful with oCPC?
Lead-gen advertisers with weak lead qualification and ecommerce advertisers with incomplete purchase value data should be most careful. In both cases, the campaign will optimise towards the tracked event, even when that event does not reflect revenue quality.
What should UK advertisers fix before testing OpenAI oCPC?
Fix the conversion event first. Confirm the event source, test the Conversions API or pixel setup, document the commercial goal, and launch with controlled budgets until the conversion data proves it is reliable.
