A click attribution window is the fixed period after someone clicks your ad during which a resulting conversion is credited back to that click. It differs from a view-through window, which credits conversions that follow an ad impression nobody clicked.
When you set this window, you decide how much of your ad's delayed impact you are willing to see, since many buyers return to convert days after the first click. Choose a window that is too short and you undercount conversions, starve your bidding of signal, and quietly judge a working channel as a failure.
The click attribution window measures the gap between an ad click and a conversion, then decides whether that conversion still counts toward the ad. In ChatGPT Ads you configure this window at 7, 14, or 30 days, all documented in OpenAI's Ads Manager guide. Any conversion that lands after the window closes is no longer tied to the click, even when the ad clearly started the journey.
The window has three moving parts: the click event that starts the clock, the conversion event your pixel or Conversions API reports, and the matching logic that joins them if the conversion falls inside the window. ChatGPT Ads uses last-touch logic, so when both a click and an impression qualify, the click takes the credit.
The click attribution window sits beside the view-through window, which OpenAI fixes at one day and keeps out of your main conversions total. Together they define what your reported numbers actually contain before any analysis begins. AirOps helps teams connect ChatGPT ad performance to the same market data driving their organic AI search work, so paid and organic read from one source of truth.
Resources: See how to track ChatGPT ad conversions with the pixel and a holdout test
A click attribution window runs as a sequence that starts the moment someone clicks and ends when the matching deadline passes. Here is the order it actually follows.
Click recorded. ChatGPT Ads appends an oppref identifier to your landing URL, marking the click that starts the clock.
Window set. The length you chose in Ads Manager determines how long that click stays eligible to receive credit for a later conversion.
Conversion fires. Your measurement pixel and Conversions API report the purchase, signup, or lead, each carrying a shared event ID.
Match made. OpenAI checks whether the conversion falls inside the window and, if it does, ties it to the original click using last-touch logic.
Report built. The conversion appears in your click-through column, typically a day or two after it happened.
The window tells you how many conversions the platform could connect to a click within the deadline you set. It does not tell you about buyers who converted later through untracked paths, so a short window hides real demand your ads created.
The window you pick shapes the return on ad spend you report, and that number decides whether ChatGPT Ads keeps its budget next quarter. Get it badly wrong and you make a budgeting decision on numbers that describe your setting more than your channel.
It governs how much credit the channel earns: a longer window captures delayed conversions, so the shortest setting can make a campaign look weaker than a longer one with no change in actual performance.
It feeds or starves your optimization: ChatGPT's conversion-optimized bidding trains on click-attributed events, so a window too short to catch your real sales cycle gives the system thin signal and worse targeting.
It creates a silent failure mode: when buyers convert after the window closes, those sales vanish from the ad's record, and a profitable channel reads as a loser you are tempted to cut.
SEO managers use the click attribution window to compare ChatGPT Ads against organic AI search on a matched timeframe before reallocating budget.
Growth marketers use the click attribution window to set a window that matches their sales cycle so paid reporting reflects true pipeline instead of only same-session buyers.
Demand gen leads use the click attribution window to defend ChatGPT Ads spend to finance with conversion numbers that account for delayed, considered purchases.
A click attribution window only works alongside an attribution model that decides which touch earns the credit, and ChatGPT Ads applies last-touch logic, so the final qualifying click wins the conversion whenever several clicks or impressions compete for the same sale.
Conversion lag is the time that passes between the click and the completed purchase, and it is the single factor that should set your window length, because any window shorter than your audience's typical lag silently drops real conversions out of the report.
The click window operates independently of the view-through window, so extending the number of days you allow a click to earn credit never changes the fixed view-through figure that OpenAI calculates and reports as a separate column.
Credit delayed conversions that a same-session view would miss entirely.
Match your reporting window to platforms like Google and Meta so cross-channel comparisons stay honest.
Feed ChatGPT's conversion-optimized bidding enough signal to target better.
Pick a window length that fits short or considered buying cycles.
Separate click-driven sales from view-through credit in your reporting.
Defend ChatGPT Ads budget with numbers finance will trust.
Set your window to match your real conversion lag, so the report captures the buyers who actually came back to purchase.
Compare ChatGPT Ads to Google and Meta on the same window length, so no channel gets an unfair reporting advantage.
Keep the window consistent across campaigns, so quarter-over-quarter trends mean something.
Capture the oppref identifier server-side, so delayed conversions survive browser cookie limits that erase the attribution link.
Send conversions through both the pixel and the Conversions API with a shared event ID, so the match holds even when a browser blocks the pixel.
Revisit your window whenever your sales cycle shifts, so the setting keeps tracking reality.
Avoid picking the shortest window just because it reports conversions fastest. It flatters your same-day numbers while hiding the delayed sales that make the channel worth running, and a channel judged on a day or two of data gets cut before it ever proves itself.
AirOps: connects your ChatGPT Ads performance, including spend, CTR, and CPC, to the same market data behind your organic AI search, so you read paid and organic results side by side.
OpenAI Ads Manager: sets the click attribution window per campaign and reports click-through conversions in the column you configure.
Google Analytics 4: captures the oppref-tagged landing traffic server-side so you can see delayed conversions your ad platform's window misses.
Find your lag. Pull your last quarter of conversions and measure how many days typically pass between first click and purchase. You can do this this week with existing analytics and no new budget.
Set the window. In Ads Manager, choose the click window that covers most of that lag, under Edit columns then Conversion Events then Attribution windows.
Verify tracking. Confirm the oppref identifier is captured server-side and that your pixel and Conversions API share one event ID, so conversions inside the window actually register in the click-through column.
Align comparisons. Set Google Ads and Meta to read on the same window length so your cross-channel numbers compare like with like.
Review and adjust. After a full cycle, check whether conversions cluster near the deadline, and lengthen the window if too many sales arrive just as it closes, which signals your lag outran the setting.
A click attribution window is the set period after an ad click in which a later conversion still counts toward that click.
You configure it in Ads Manager and let your pixel and Conversions API report the conversions it matches.
The window only reflects conversions the platform can link to a click before the deadline passes.
Set it shorter than your buying cycle and you undercount sales, misread performance, and risk cutting a channel that works.
The leverage is matching the window to your real conversion lag, which turns reported numbers into a fair read of the channel.
They credit different events. A click attribution window credits a conversion that follows a click within the number of days you set, while a view-through window credits a conversion that follows an ad impression nobody clicked. In ChatGPT Ads the click window is configurable while the view-through window is fixed and cannot be adjusted, so the two never move together. The split matters because click-through conversions flow into your main conversions total and drive bidding, while view-through conversions sit in a separate column and stay out of your cost-per-acquisition math. When a user both sees and clicks an ad before converting, the click takes the credit under last-touch logic. Treat the click window as your core performance read and the view-through number as supplemental context, because mixing them inflates your totals and hides which conversions had a real click behind them.
Match it to your conversion lag, which usually means the window should cover the time most of your buyers take to purchase after clicking. If you sell a low-cost product people buy on impulse, a short window captures nearly everything. If you sell considered purchases with research cycles that run a week or more, a longer window keeps those delayed sales attached to the ad that started them. The practical method is to pull a quarter of historical conversions, measure the days between click and purchase, and pick the setting that covers the bulk of that distribution. Then hold it steady, because changing the window every month makes your trend lines meaningless. Revisit only when your product mix or sales cycle genuinely shifts. A window chosen this way reflects how your audience actually buys, so the reported numbers stop punishing campaigns for the lag that is normal in your category.
Because conversions do not all happen at the moment of the click. A share of buyers convert immediately, and the rest return hours, days, or weeks later, so where you draw the deadline decides how many of those later conversions the platform still credits to the ad. A short window counts only the fast converters and discards everyone who took longer, which makes the same campaign look less productive. A longer window pulls more of the delayed conversions back into the report. The size of the swing depends entirely on your category: impulse purchases barely move when you extend the window, while considered purchases with long research cycles can shift dramatically. Browser cookie limits add another source of variance, because delayed conversions tracked only client-side can expire before they are counted. This is why two advertisers with identical spend can report very different results purely from how each one set the window.
Yes, within fixed limits. You choose the click window yourself in Ads Manager under the conversion event settings, so the length is a direct and deliberate decision you own. What you cannot do is invent an arbitrary length: you select from the options OpenAI offers for the click window, and the view-through window stays locked. You also cannot change how a conversion is matched, since the platform applies last-touch credit and awards the click when a click and an impression both qualify. The bigger lever sits in your own tracking: capturing the click identifier server-side and sending conversions through both the pixel and the Conversions API determines whether a conversion inside your window actually gets recorded. So you control two things that matter, the window length and the reliability of your measurement, while the matching rules and the view-through behavior remain OpenAI's to set.
There is no single right number, so the honest benchmark is a window that captures the large majority of your conversions before it closes. A practical target is to cover the span in which most of your historical conversions land, which for many ecommerce and SaaS advertisers falls within a week of the click, while longer consideration cycles need more room. Judge the setting by one test: pull your conversion distribution and check what share arrives after the window closes. If a meaningful chunk lands outside it, your window is too short and you are undercounting. A good window also holds steady over time and matches the window you use on other paid channels, so comparisons stay fair. Do not chase a competitor's number or a platform default as a benchmark, because the only figure that matters is how your own buyers behave. Set it to fit that behavior and the benchmark takes care of itself.