Meta Attribution Windows Explained for Agency Reporting

Your client emails on Monday asking why your report shows more leads than their own inbox does. You open Ads Manager, look at the same campaign, and see a different figure again. Nothing is broken. You are almost certainly looking at the same campaign through different attribution windows, and until you can name the window behind each number, you cannot explain the gap to anyone.

Attribution windows sit behind a lot of the “your numbers do not match ours” conversations in Meta Ads reporting, and they are among the easiest things to fix once you understand what they do. This post covers what a window is, the settings Meta offers, where the setting lives, and how to keep client reports consistent.

What an attribution window actually is

An attribution window is the answer to one question: how long after seeing or clicking an ad does a conversion still count as caused by that ad?

Meta records two kinds of interaction. A click is someone tapping your ad. A view, which Meta calls an impression, is someone seeing your ad without tapping it. If a person clicks your ad, leaves, and comes back later in the week to buy, Meta can credit that purchase to the ad, but only if your attribution window is long enough to reach that far.

That is the whole idea: a window is a length of time and a type of interaction. Change either one and the reported conversion count changes, even though the behaviour of real people did not.

The settings Meta gives you

Meta calls this the attribution setting, and it sits on the ad set. The options you see when creating or editing an ad set combine a click window and a view window:

• One day click. Only conversions that happen within a day of someone clicking. The tightest option.
• Seven day click. Conversions within a week of a click. This is Meta’s default for most objectives.
• One day view. Conversions within a day of someone seeing the ad without clicking.
• Seven day click and one day view. Both counted together. Broader, and it credits ads people saw but did not tap.

There is also an engaged-view window for video, which counts a conversion when someone watched a meaningful chunk of your video before converting.

These settings do two separate jobs. They tell Meta’s delivery system what to optimise towards, and they tell the reporting interface what to display. The optimisation half is set when the ad set is created and shapes how Meta spends your budget. The reporting half is what you argue about in a client call.

Why the same campaign shows different numbers

Here is the part that catches people out. The conversion figure in a report uses the ad set’s attribution setting by default. If some ad sets in a campaign use seven day click while others use seven day click and one day view, the campaign total mixes them. It is not wrong, it is just not a single consistent rule.

Meta also lets you compare windows directly in the reporting interface. That is useful for understanding a campaign and dangerous for reporting, because two people can open the same account, sit on different comparison views, and read out different numbers in complete good faith.

Then there is the modelling layer. Meta uses statistical modelling to fill in conversions it cannot observe directly, largely because of the privacy changes that limited tracking on iOS. Modelled figures can shift for the same date range days after the fact, which is why a number you screenshot on Monday may not match it on Friday. Windows and modelling are separate mechanisms, but they compound: a wider window gives the model more room to attribute, and the figures settle more slowly.

If your reported numbers and your Ads Manager numbers keep drifting apart, the deeper causes of number mismatches are worth reading alongside this.

Why your client’s numbers will never match exactly

Your client is counting something different from what Meta counts, and no attribution window will reconcile the two.

Meta counts a lead when the ad drove the action it was told to watch for, within the window. Your client counts a lead when a named human landed in their inbox or their own records. Those definitions diverge for real, mundane reasons:

• A person fills the same Instant Form twice, on a phone and then a laptop. Meta may record two events. Your client sees one person.
• A person clicks the ad early in the week and converts at the weekend. Inside a seven day click window Meta credits the ad. Outside it, Meta does not, and your client still got the lead.
• A person sees the ad, does not click, and searches for the brand later. With a view window that counts as ad-driven. Without one it looks like organic traffic.
• Someone submits a form with a typo in their email. The client’s system rejects it. Meta counted it.

The honest position is not “our number is right and yours is wrong”. It is that the two systems answer different questions, and the platform number is the one you use to judge whether the advertising is working, because it is the only one that knows which ads people actually interacted with.

How to keep this consistent in your reporting

The fix is discipline rather than cleverness. Four habits cover almost all of it.

Pick one window per client and write it down. Use it across every ad set in that account and record it somewhere the client can see. A default everyone agrees on beats a technically superior setting nobody remembers.

State the window on the report itself. A single line at the top saying which attribution setting the figures use converts a future argument into a footnote. It takes one sentence.

Let the numbers settle before you report them. Because of modelling and delayed conversions, very recent days keep moving. Reporting Monday to Sunday on the following Tuesday rather than late Sunday night gives the figures time to firm up.

Do not change the window mid-campaign without noting it. Changing an ad set’s attribution setting changes what Meta optimises towards, so it is a real intervention, not a reporting tweak. If you do change it, mark the date, because the before and after are not directly comparable.

If you pull these numbers manually every month, window discipline is exactly the kind of thing that gets lost. In DashOps, campaign and ad set figures come through the same connection with the same period comparison every time, so a report you send in March is built the same way as one you send in September. The scheduled client report goes out on the same cadence with the same definitions, which takes the accidental version of this problem off the table.

Explaining it to a non-technical client

Most clients do not want a lesson in attribution. They want to know whether they can trust the number. One analogy usually lands: a shop assistant hands out flyers in the street. If someone walks in that afternoon holding a flyer, the assistant clearly caused the visit. If they walk in the following week without it, having read it and remembered, the assistant probably helped, and you have to decide whether to credit them. The attribution window is that decision, made in advance and applied consistently.

Then show the trend rather than the single figure. A cost per lead that fell across several months is meaningful whichever window you picked, because the same rule was applied to every month. A read-only client view or a share link showing that trend, as DashOps offers, usually ends the conversation faster than a spreadsheet does.

What to do next

Open one client’s account and check the attribution setting on every live ad set. If they are not all the same, you now know why the campaign total looks odd. Decide on one setting for that account, note the date you standardised it, and add a line to your report template stating which window the numbers use.

If attribution changes have already forced you to rethink how you report, the wider picture on attribution shifts covers what else moved. And if the manual work of pulling these numbers is what is really costing you, the full guide to reporting outside Ads Manager walks through the alternative.