Conversion window

A conversion window is the length of time after somebody clicks or sees an advert during which a resulting sale or inquiry may still be credited to it. Once the window closes, the same customer buying the same thing counts as something else. It is a deadline, and it is quietly responsible for a large share of the reporting disagreements businesses have with their own numbers.

In short

• How long a click stays eligible for credit after it happens.
• Set per conversion action, not once for the whole account.
• Google Ads dates conversions to the click, so old months keep changing.
• Changing the setting never repairs data already collected.

The defaults, tool by tool

Every platform ships with a different answer, and none of them announce it.

• Google Ads, clicks. Thirty days by default, settable from one to ninety. Google advises against going below seven.
• Google Ads, views. One day by default, settable up to thirty. This covers people who saw an advert and never clicked it.
• Google Analytics 4, acquisition events. Thirty days by default for the events marking a first visit or a first app open, with seven available.
• Google Analytics 4, everything else. Ninety days by default, with thirty and sixty as the alternatives. Whatever you pick also governs how sessions are attributed.
• Meta. Seven days for clicks and one day for views. The seven and twenty-eight day view windows were removed in January 2026.

Nobody chose that spread of numbers as a set, and it explains a great deal. Google Analytics 4 counting for ninety days while Google Ads counts for thirty means the two tools are answering slightly different questions about the same customer, before anything else has gone wrong.

Why last month’s number keeps moving

This is the question that brings most people to the term in the first place, and the answer is a piece of arithmetic, not a fault.

Google Ads credits a conversion back to the date of the click that produced it. A click on the twenty-eighth of the month that becomes a sale on the fourth of the next is recorded on the twenty-eighth, in a month that was reported on and closed several days earlier. The longer the window, the longer that backfilling continues, and a ninety day window can still be adding to a month a quarter later.

There is a reason for the design. Cost is recorded on the click date too, so placing the conversion there keeps cost and result on the same day and makes cost per conversion meaningful. Placing it on the day of the sale would compare this week’s revenue against last week’s spend.

The habit that follows is simple. Do not report a period until roughly one window has passed, or report it and say plainly that the figure is provisional. A business that reports on the first of the month, every month, with a thirty day window, is permanently reading an incomplete number and treating it as final.

Words you will hear

• Lookback window. The same idea, and the name Analytics uses.
• Click-through window. The period after a click.
• View-through window. The period after an impression that was not clicked.
• Engaged-view. A view counted after somebody watched a video for a qualifying length of time. Analytics defaults to three days for these.
• Conversion lag. How long your customers actually take, which is the measurement the window should be set from.
• Engage-through. Meta’s category for likes, shares and comments, which used to be counted alongside clicks and now reports separately.

Setting one that matches how people buy

The instinct is to pick a long window because it shows more conversions. That instinct produces a number nobody can act on.

A long window credits an advert for a decision it may have had very little to do with. Somebody who clicked eleven weeks ago and bought yesterday probably arrived through something else in between, and the attribution model is left dividing credit across a stretch of time in which most of the useful evidence has gone. A short window does the opposite, and simply loses the slow half of your customers, which is the expensive half in any trade where people ask for a quote before they commit.

The honest way to choose is to read your own lag rather than borrow a rule. Google Ads reports how long your customers took between click and conversion, and the shape of that distribution decides the setting. Most trades find that a large majority converts inside a few days and a thin tail continues for weeks. Set the window past the point the tail stops being worth waiting for.

Two things then matter more than the number itself. Write down what you chose and the day you chose it, because a key event whose window changed in March will produce a step in the chart that somebody will otherwise blame on the advertising. And keep it still. A window that gets adjusted whenever results disappoint stops measuring anything at all, since every comparison is then between two different definitions of the same word.