OpenAI’s Conversions API Is Live — WooCommerce Has No Plugin Yet

ChatGPT Ads Now Has Real Measurement Infrastructure

OpenAI moved from experimental reach campaigns to a full performance advertising stack in under six months.

ChatGPT now serves over 900 million weekly active users, and OpenAI’s ads pilot reached $100 million in annualised revenue within six weeks of its January 2026 launch. That’s not an experiment anymore. In May 2026, OpenAI opened self-serve Ads Manager access to all US businesses. By June 5, any advertiser who had configured the pixel or Conversions API gained access to conversion-optimised campaigns — the kind that optimise toward purchases and leads rather than impressions and clicks.

The shift from reach to conversion optimisation is structural, not cosmetic. Conversion campaigns require a closed attribution loop. The advertiser must send post-click behavioural data back to OpenAI before the algorithm can optimise toward purchase outcomes. That loop is now available through two tools: a JavaScript pixel that fires in the browser, and a server-side Conversions API that sends events directly from your backend.

OpenAI’s Conversions API endpoint at bzr.openai.com accepts up to 1,000 server-side events per batch, mirroring Meta’s CAPI architecture with SHA-256 hashed user data and event-ID deduplication.

Technology partners are already building into the ecosystem. Criteo reported more than 1,000 brands running active campaigns through its API connection as of May 2026. LiveRamp became the first independent ad tech company to pipe conversion data into the API in June 2026. Stape, Tealium, and Singular have all shipped connectors. The infrastructure is real. The question for WooCommerce store owners isn’t whether ChatGPT Ads will matter — it’s whether their tracking is ready to use it.

How the Conversions API Works

The pattern mirrors Meta’s CAPI — same dual-track logic, same deduplication, different endpoint.

If you’ve implemented Meta’s Conversions API, you already understand the architecture. OpenAI’s version follows the same dual-track pattern: a browser pixel for speed, a server-side API for reliability, and deduplication via shared event IDs. The server endpoint accepts a POST request to bzr.openai.com/v1/events?pid=<PIXEL-ID> with a Bearer API key for authentication.

Every event requires four fields: id, type, timestamp_ms, and data. The timestamp must fall within the last seven days and no more than ten minutes ahead of the current time. One request accepts up to 1,000 events, but a single failing event in the batch rejects the entire batch — so pre-validation with the validate_only flag matters before you go live.

The API supports six action sources out of the box: web, mobile app, offline, physical store, phone call, and email. That means CRM and offline-conversion use cases are technically possible from day one, though OpenAI hasn’t yet published native first-party connectors for major CRM or CDP platforms.

User matching relies on hashed identifiers. All personally identifiable data — email, phone number, external user ID — must be SHA-256 hashed before transmission. The requirement is strict: lowercase the input, trim whitespace, then hash. Never send raw PII. The match rate depends on how many signals you pass with each event. Email gives the highest match rate. Client IP address and user agent string are available on every server-side request and should always be included.

You may be interested in: The Cumulative Tracking Loss Stack: When Ad Blockers, ITP, Consent, and App Browsers Hide 50% of Your Customers

The WooCommerce Integration Gap

Every major ad platform has a WooCommerce integration path. OpenAI doesn’t — yet.

Meta has official WordPress plugins. Google Ads has Enhanced Conversions with WooCommerce extensions. TikTok, Pinterest, and Snapchat all ship WordPress or WooCommerce connectors. OpenAI has none. As of July 2026, WooCommerce has no native plugin for the OpenAI Conversions API or pixel.

The only existing pre-built connector is Stape’s GTM server-side tag, released in May 2026. It works well — it maps GA4 event names automatically, supports all standard OpenAI event types, handles the oppref cookie, and lets you fire asynchronously for faster response times. But it requires a Google Tag Manager server container. If you’re not running sGTM, the tag is unusable.

WooCommerce has no native OpenAI Conversions API plugin as of July 2026; the only existing connector is Stape’s GTM server-side tag, which requires a Google Tag Manager server container.

For WooCommerce stores without server-side GTM, the options are narrower. You can build a custom integration using the woocommerce_payment_complete hook to fire a server-side POST to OpenAI’s endpoint on every confirmed order. You can use a middleware platform to bridge the event. Or you can wait for the ecosystem to catch up — which, based on the Meta CAPI timeline, could take six to twelve months.

Translation: if you’re running a WooCommerce store and want to test ChatGPT Ads with proper conversion tracking, you need either a server-side event pipeline or a GTM server container. There’s no install-and-go path today.

The oppref Click Identifier

OpenAI’s privacy-preserving click ID works like fbclid or gclid — but you have to capture it yourself.

When a user clicks a ChatGPT ad, OpenAI appends the oppref parameter to your landing page URL. This is the privacy-preserving click identifier — analogous to Meta’s fbclid or Google’s gclid. It’s the signal that connects a server-side conversion event back to the ad impression that started the journey.

The oppref must be captured client-side, stored through the session, and passed back with the server-side conversion event. If you lose it between the landing page and the order confirmation, OpenAI can’t attribute the sale to the ad. Your Conversions API event will still fire, but the match rate drops to near-zero without at least one user identifier linking the event to an impression.

Stape’s GTM tag handles oppref automatically — it reads the parameter from the URL, stores it as a __oppref cookie, and includes it in every subsequent server-side event. If you’re building a custom integration, you need to replicate this: read the query parameter on landing, persist it through the checkout flow, and include it in the POST body when you fire the conversion event.

The pattern is identical to what stores already do with fbclid — but if your current setup doesn’t capture click IDs from the URL and persist them server-side, you’ll need to add that layer before OpenAI’s attribution loop works.

Pixel vs CAPI: What Each Tracking Method Captures

The gap between browser-only and server-side tracking isn’t theoretical — it’s measurable and growing.

The difference between pixel-only and CAPI tracking is the difference between hoping the browser cooperates and knowing the event fired. The table below shows what each method can and cannot capture across common WooCommerce scenarios in 2026.

Tracking Capability	Pixel Only (Browser)	Conversions API (Server)	Pixel + CAPI (Hybrid)
Desktop Chrome purchase	Captured (~95%)	Captured (100%)	Deduplicated, full signal
Mobile Safari purchase	Often lost (ITP caps cookies at 7 days)	Captured (100%)	Server event fills the gap
Ad blocker active	Lost (pixel blocked entirely)	Captured (100%)	Server event fills the gap
Consent rejected (GDPR/ePrivacy)	Lost (pixel suppressed)	Captured (legitimate interest basis varies)	Server event may still fire
In-app browser (Instagram, TikTok)	Unreliable (referrer stripped)	Captured (100%)	Server event fills the gap
oppref click ID persistence	Stored as JS cookie (vulnerable to ITP)	Read from server session (durable)	Both paths available
Estimated event capture rate (2026)	70–80% of actual conversions	95–100% of actual conversions	95–100% with deduplication

The estimated capture rates are drawn from industry-wide Meta CAPI benchmarks. Browser-side pixels now miss 20–30% of conversion events due to the compounding effects of ad blockers, ITP, consent mechanisms, and in-app browsers. On a new ad platform with smaller event volumes, every missing conversion has an outsized impact on algorithmic optimisation.

Why Server-Side Matters More Than You Think

Browser-only tracking misses 20–30% of conversions. On a new ad platform, that gap is the difference between usable data and noise.

Browser-based pixels are subject to ad blockers, Safari’s Intelligent Tracking Prevention, cookie consent rejection, and browser crashes. In 2026, those factors combine to suppress 20–30% of pixel events across the industry. On Meta, that loss is well-documented. On a new platform like ChatGPT Ads, the loss is identical — but the dataset is smaller, which means every missing event has a larger proportional impact on the algorithm’s ability to optimise.

A conversion-optimised campaign with incomplete data doesn’t just undercount sales — it trains the algorithm on a biased sample. The purchases that survive browser-side tracking skew toward desktop Chrome users who haven’t installed ad blockers and don’t reject cookie consent. Mobile Safari users, privacy-conscious shoppers, and in-app browser visitors vanish from the signal entirely.

The Conversions API bypasses all of this. The event fires from your server, not the visitor’s browser. No ad blocker can intercept it. No ITP policy can cap the cookie. No consent banner can suppress the HTTP request. The event fires reliably, every time.

For WooCommerce stores testing ChatGPT Ads, this isn’t optional infrastructure — it’s the minimum viable measurement layer. Running pixel-only tracking on a new channel with a small event volume is asking the algorithm to optimise on a fraction of a fraction of your actual conversions.

You may be interested in: WooCommerce 10.9 and AI Agents: Every Sale They Complete Never Fires a Tracking Tag

What WooCommerce Stores Should Do Right Now

You don’t need to wait for a native plugin. The infrastructure patterns already exist.

The server-side event pipeline required for OpenAI’s Conversions API is the same pattern WooCommerce stores already use for Meta CAPI, Google Enhanced Conversions, and TikTok Events API. If you’ve built one server-side integration, the second is incremental work — a new endpoint, a new event schema, a new click ID to capture.

The stores that will benefit most from ChatGPT Ads are the ones that already have clean server-side event infrastructure. They can add OpenAI as a destination in their event pipeline without rebuilding anything. One order event, routed to Meta, Google, and OpenAI simultaneously, with platform-specific formatting applied at the destination layer.

If you don’t have server-side infrastructure yet, the arrival of a fourth major ad platform with a Conversions API should accelerate the decision. The build-once-route-everywhere pattern is no longer a nice-to-have. It’s how modern WooCommerce measurement works. Transmute Engine™ is built on exactly this principle — a single server-side event captured at the WordPress layer, transformed and routed to every platform that needs it through outPIPE.

The practical steps are clear. Install the OpenAI pixel today for immediate baseline tracking. Register your conversion events in Ads Manager. Then build or connect a server-side path — whether through sGTM with Stape’s tag, a custom integration on woocommerce_payment_complete, or an event pipeline that already handles your other platforms. The sooner the API receives real conversion data, the sooner conversion-optimised campaigns become available to your account.

Key Takeaways

• New ad platform, familiar architecture: OpenAI’s Conversions API at bzr.openai.com mirrors Meta’s CAPI — server-side POST, SHA-256 hashed user data, event-ID deduplication, and a click identifier (oppref) that must be captured and passed back.
• No WooCommerce plugin exists yet: As of July 2026, the only pre-built connector is Stape’s GTM server-side tag. Stores without sGTM need a custom integration or a server-side event pipeline.
• Browser-only tracking is insufficient: Pixel events miss 20–30% of conversions due to ad blockers, ITP, and consent rejection. On a new channel with small event volumes, that gap makes optimisation unreliable.
• AI-referred traffic is high-value: Visitors from ChatGPT convert at 4–5x the rate of organic search. The measurement infrastructure you build now determines whether you can capture and optimise for that value.
• Server-side pipelines pay compound returns: The same infrastructure that routes events to Meta CAPI and Google Enhanced Conversions can add OpenAI as a destination — one build, multiple platforms.

References

OpenAI Developers — Ads: Conversions Pixel and Conversions API documentation. developers.openai.com/ads (May 2026). Search Engine Land — “OpenAI confirms conversion-focused ads are coming to ChatGPT.” searchengineland.com (May 2026). PPC.Land — “OpenAI’s ChatGPT ads are getting conversion optimization.” ppc.land (May 2026). FatJoe — “ChatGPT Stats July 2026: 900M weekly active users, $100M ARR ads pilot.” fatjoe.com (July 2026). Pixis — “Why AI Search Traffic Converts at 4–5x.” pixis.ai (April 2026). TapClicks — “How to Track AI Referral Traffic: 70.6% arrives without referrer headers.” tapclicks.com (July 2026). Benly.ai — “Conversions API (CAPI): Server-Side Tracking Guide 2026.” benly.ai (March 2026). Stape — “OpenAI Ads Conversions API Tag for GTM Server-Side.” github.com/stape-io (2026). Digiday — “OpenAI’s ChatGPT ads get first conversion API partner in LiveRamp.” digiday.com (June 2026).

If your WooCommerce store is running conversion campaigns on Meta and Google but hasn’t started building toward ChatGPT Ads measurement, the infrastructure gap is widening. Talk to Seresa about connecting your store’s server-side event pipeline to every platform that matters — including the ones that launched last month.