Attribution Windows: Why 1 Day View and 7 Day Click Never Agree
Two platforms reporting the same campaign with different windows will produce different numbers and both can be internally correct. How to compare them without pretending one is the truth.
Change a window and you change the result. A campaign credited with 400 conversions at a 7 day click and 1 day view setting may show 260 at 1 day click, and neither figure is a lie. The longer window admits conversions where the ad was one touch among many, days before a purchase the shopper may have completed for reasons that had nothing to do with it. Comparing two platforms, or the same platform before and after a settings change, without aligning windows is the most common way teams manufacture a performance drop that never happened.
An attribution window has two parts that are often conflated: the lookback period for the touch, and which touches count, click or view. View through is the contentious half. A person who was served an impression and later bought is a weaker claim than a person who clicked, and platforms differ in how prominently they include view through in their headline number. Know which one you are reading before you act on it.
Align before you compare
Write down, for every source you report, the window, the conversion event, the timezone, the currency, and whether the figure is modelled or observed. That dull list prevents most attribution arguments. Google Analytics, ad platforms, and a store's own order system differ on all five, so their totals should not match. Agreement is not the goal. A stable, understood difference is.
When a platform changes a default, treat your trend line as broken at that date and annotate it. Teams routinely spend a month investigating a decline that begins exactly on the day a window default changed in an interface update. Annotate settings changes the way you annotate site changes and promotions. Attribution is a measurement instrument, and instrument changes belong in the record.
Choosing a window for the decision at hand
Match the window to the buying cycle and the decision. A short window is the stricter test and suits budget shifts between similar campaigns, where you want credit only for purchases plausibly driven by the ad. A longer window suits considered purchases, furniture, appliances, anything researched over weeks, where a 1 day window would credit the last search ad for work done by earlier touches.
What does not work is choosing the window that makes a favoured channel look best and then using it for everything. Decide the window by product category, publish it in the report, and keep it fixed for the period you are judging. If a channel only performs under the most generous setting, that is information about the channel, not a reason to adopt the setting.
Reporting to people who want one number
Give them one number and its definition, then the sensitivity. Headline: conversions at 7 day click, 1 day view. Underneath: the same campaign at 1 day click, and the share that is view through only. That second line changes conversations. A channel whose value is mostly view through gets a different budget discussion from one whose value is mostly clicks, even when their headline totals are identical.
Be explicit about what no window can fix. Cross device journeys, consent losses, and browser restrictions remove touches from every system. Attribution redistributes the credit that is visible. It does not recover the credit that was never recorded, which is why incrementality tests and blended store level figures belong next to platform attribution rather than underneath it.
Related reading: Attribution Playbook, Answers, Glossary.
Frequently asked questions
Which attribution window is correct?
There is no correct window in the abstract. There is a window appropriate to the product's buying cycle and the decision being made. Shorter windows are stricter and favour channels close to purchase. Longer windows credit earlier influence. Pick one deliberately, document it, and hold it steady for comparisons.
Why did my conversions drop after changing nothing?
Check the instrument first. A changed attribution setting, a consent banner change, a tag change, or a platform default update can all reduce reported conversions while orders stay flat. Compare against store orders for the same period before concluding demand fell.
Should view through conversions be reported?
Report them, labelled as view through, separately from click through. Hiding them flatters nothing and including them silently in a headline inflates channels that buy cheap impressions. The split between the two is often the most useful part of the report.
Can I compare Meta and Google numbers directly?
Not as raw totals. They use different windows, different modelling, different event definitions, and both claim credit for overlapping conversions, so adding them double counts. Align definitions as far as possible, expect a gap, and judge each channel within its own consistent series.