ChatGPT Ads Attribution Windows Are Undocumented — Test This First

What attribution window do ChatGPT Ads use?

OpenAI has not publicly documented the click-through attribution window for ChatGPT Ads — store owners are scaling spend without knowing the lookback period their conversions are measured against (Agentic Commerce, 2026). That’s the situation as of Q3 2026. You can set up the pixel, configure oCPC bidding, send conversion events, and scale your budget — but you cannot verify the window against which OpenAI credits those conversions to your campaign.

This matters because attribution windows fundamentally shape ROAS calculations. A 7-day window captures fewer conversions than a 30-day window for the same spend. If OpenAI uses a longer window than you assume, your reported ROAS looks better than it really is. If it uses a shorter one, you may be under-counting and pulling budget from a channel that’s actually performing.

Every other major ad platform publishes this. Google Ads documents its 30-day click-through default. Meta publishes its 7-day click / 1-day view default. TikTok documents its 7-day click / 1-day view. ChatGPT Ads? Nothing. The oCPC bidding documentation covers event setup and pixel installation but stops short of the attribution mechanics.

OpenAI has not publicly documented the click-through attribution window for ChatGPT Ads — stores are scaling oCPC spend without knowing the lookback period (Agentic Commerce, 2026).

How does the oppref cookie attribute ChatGPT Ads clicks?

The oppref cookie stores the ad-click ID that OpenAI uses to match a conversion back to the campaign — if the cookie expires before the purchase, the conversion goes unattributed (Ridoway Pixel, 2026). This is the technical mechanism underneath ChatGPT Ads attribution, and understanding it is the first step toward testing the window.

When a user clicks a ChatGPT Ads placement, OpenAI redirects through a tracking URL that sets the oppref cookie in the user’s browser. That cookie contains a unique identifier tying the click to the campaign, ad group, and creative. When the OpenAI pixel fires a conversion event (typically on the WooCommerce thank-you page), the pixel reads the oppref cookie and sends the click ID alongside the conversion data.

The critical unknown is the cookie’s expiry. First-party cookies set by JavaScript are subject to Safari ITP’s 7-day cap. Third-party cookies are blocked entirely in Safari and Firefox. If oppref is a first-party cookie set via JavaScript, Safari users who buy more than 7 days after clicking will be unattributed regardless of OpenAI’s intended window. If it’s set server-side with proper headers, it could survive longer — but OpenAI hasn’t documented which method they use.

The oppref cookie carries the ad-click attribution ID for ChatGPT Ads; if it expires before purchase, the conversion goes unattributed (Ridoway Pixel, 2026).

Related: What Is the OAIQ JavaScript Pixel and How Does ChatGPT Ads Conversion Tracking Work for WooCommerce?

How do ChatGPT Ads attribution windows compare to Meta and Google?

Meta defaults to a 7-day click / 1-day view window; Google Ads uses a 30-day click-through — ChatGPT Ads’ window is undocumented, making cross-platform ROAS comparisons unreliable (Segwise, 2026). When you compare ROAS across platforms, you’re implicitly assuming the attribution windows are comparable. They aren’t — even between documented platforms.

Google’s 30-day window captures conversions that Meta’s 7-day window misses. A customer who clicks a Google ad and buys 15 days later is attributed to Google. The same customer clicking a Meta ad and buying 15 days later is not attributed to Meta. Same customer, same behaviour, different ROAS numbers — purely because of the window.

Platform	Default click window	Default view window	Documented?
Google Ads	30 days	N/A	Yes
Meta Ads	7 days	1 day	Yes
TikTok Ads	7 days	1 day	Yes
ChatGPT Ads	Unknown	Unknown	No

Now add ChatGPT Ads with an unknown window. If OpenAI uses a 30-day window, ChatGPT Ads ROAS is inflated compared to Meta. If it uses a 1-day window, it’s deflated compared to Google. You don’t know which comparison is valid — and neither does your oCPC algorithm.

Related: Before You Spend a Dollar on ChatGPT Ads: The WooCommerce Tracking-Health Checklist

How should WooCommerce stores test ChatGPT Ads attribution?

Run a controlled test with a dedicated UTM campaign, compare WooCommerce orders against OpenAI’s reported conversions at 1, 7, 14, and 30-day intervals to reverse-engineer the window (StoreLeads, 2026). The documentation gap is fixable with data you already have.

Here’s the test protocol. Create a ChatGPT Ads campaign with a unique UTM parameter (e.g., utm_campaign=chatgpt_attribution_test). Run it for 30 days with consistent daily spend. Then pull two datasets: WooCommerce orders that carry the UTM parameter (from your order metadata or GA4), and OpenAI’s reported conversions for the same campaign.

Compare the counts at four intervals. At day 1: how many WooCommerce orders with the UTM vs. how many conversions OpenAI reports? At day 7? At day 14? At day 30? The point where OpenAI’s count stops growing is the attribution window. If OpenAI reports 50 conversions after 7 days and still 50 after 30 days, the window is roughly 7 days. If the count keeps climbing through day 14 but plateaus by day 30, the window is somewhere between 14 and 30 days.

The UTM parameter is crucial. Without it, you can’t independently verify which WooCommerce orders came from ChatGPT Ads. The pixel and cookie might attribute them — but you need the UTM in your own data to cross-reference.

Related: OpenAI Just Launched a Conversions API — How to Send WooCommerce Purchase Events to ChatGPT Ads

Can the Conversions API fix attribution window issues?

The Conversions API sends events in real time alongside the browser pixel — it does not change the attribution window, but it ensures the conversion data reaches OpenAI even when the pixel is blocked (Ridoway Pixel, 2026). This is an important distinction. The CAPI solves data delivery. It does not solve data interpretation.

When a browser blocks the OpenAI pixel (ad blockers, Safari ITP, Firefox ETP), the conversion event never fires client-side. The CAPI fires the same event server-side, bypassing the browser entirely. This means more conversions reach OpenAI — but they’re still subject to whatever attribution window OpenAI applies. A conversion that arrives via CAPI 15 days after the click is attributed only if the window extends past 15 days.

The CAPI does add one useful piece: deduplication via event_id. When both the pixel and the CAPI fire for the same conversion, the event_id prevents double-counting. This means your test data (comparing WooCommerce orders against OpenAI reports) is cleaner — one order equals one reported conversion, regardless of delivery method.

What happens when the oppref cookie expires before a WooCommerce purchase?

If the oppref cookie is cleared or expires before the order completes, OpenAI cannot credit the campaign — the conversion appears as organic or direct in your analytics (Ridoway Pixel, 2026). This is the same cookie-loss pattern that plagues Meta and Google, but with an added twist: you don’t know when the cookie is supposed to expire.

With Meta, you know the window is 7 days. If a customer buys on day 8, you expect the loss. With ChatGPT Ads, you don’t know if day 3 or day 30 is the cutoff. Every unattributed conversion is ambiguous — is it genuinely organic, or did the oppref cookie expire inside an unknown window?

WooCommerce stores with longer purchase cycles — B2B suppliers, high-ticket items, stores where customers research for weeks — are most exposed. If the attribution window is shorter than the typical time-to-purchase, a systematic share of conversions will always go unattributed. You can’t fix that with better pixel implementation. The fix is server-side event capture that preserves the click ID independently of the browser cookie. Transmute Engine captures attribution parameters at event time and stores them server-side, so even if the cookie expires, the click-to-conversion mapping survives in your own data for independent verification against OpenAI’s reports.

Related: Running the ChatGPT Ads Pixel and Conversions API Together? One Wrong event_id and You’re Double-Counting Every WooCommerce Sale

How many WooCommerce stores are affected by undocumented attribution?

Of the 5,618 stores running the OpenAI pixel (July 2026), every one using oCPC bidding is optimising against an attribution window they cannot verify (StoreLeads, 2026). The adoption curve is steep — 614% quarter-over-quarter growth — and accelerating as ChatGPT Ads expands to more countries.

The oCPC bidding model makes this particularly urgent. With manual bidding, an unknown attribution window costs you reporting accuracy. With oCPC, it costs you budget efficiency. The algorithm is optimising toward conversions counted within a window you don’t know. If that window is shorter than your actual purchase cycle, oCPC will deprioritise audiences that convert — they just convert outside the window. If the window is longer, oCPC may over-value audiences whose conversions would have happened anyway.

The stores most affected are those spending the most. A store running £100/month on ChatGPT Ads can absorb attribution inaccuracy. A store at £5,000/month with oCPC bidding is making material budget allocation decisions based on numbers it can’t verify against a documented standard.

Will OpenAI document attribution windows as the ad platform matures?

The Conversions API documentation is now public, suggesting attribution window documentation will follow as oCPC scales (GlobalDev SignalBridge, 2026). OpenAI’s ad platform is on a rapid release cycle — pixel, oCPC, CAPI, and country expansion have shipped in quick succession. Attribution window documentation is a when, not an if.

Until then, store owners have two practical options. First, run the 1/7/14/30-day test described above to reverse-engineer the current window. This gives you a working number even without official documentation. Second, implement server-side event capture to build your own attribution dataset. When OpenAI does publish window documentation, you’ll have the historical data to validate it against your own records — and to see whether the window changed between when you tested and when they documented.

The combination of an undocumented attribution window and server-side event capture that preserves click IDs independently of the browser cookie means the stores that invest in tracking infrastructure now will be the only ones with clean data when the documentation arrives. Everyone else will be starting from zero.

Key Takeaways

• ChatGPT Ads attribution windows are undocumented — no public specification exists for the click-through lookback period as of Q3 2026.
• The oppref cookie is the attribution mechanism — if it expires before purchase, the conversion is lost to OpenAI’s reporting.
• 5,618 stores are affected — 614% QoQ growth in OpenAI pixel adoption with every oCPC user optimising blind.
• Cross-platform ROAS comparisons are unreliable — you can’t compare a known 7-day or 30-day window against an unknown one.
• Test at four intervals — compare WooCommerce orders against OpenAI reports at 1, 7, 14, and 30 days to reverse-engineer the window.
• Server-side capture preserves the data — store click IDs independently of browser cookies for verification when documentation arrives.

References

1. Agentic Commerce for WooCommerce. (2026). Plugin readme.
2. StoreLeads. (2026). OpenAI Tracking Pixel adoption report.
3. Ridoway Pixel. (2026). ChatGPT Ads pixel for WooCommerce.
4. Segwise. (2026). Meta Pixel and Conversions API updates.
5. GlobalDev SignalBridge. (2026). OpenAI Conversions API integration.