Two people can look at the same campaign, in the same account, on the same day, and disagree about whether it worked. Usually neither of them has checked the attribution window. At The Growth Bully, a Malta performance marketing agency, we treat the window as the first thing to agree and the last thing to change, because it decides which channel looks good before anybody has looked at the performance.
Arguing about channel performance without agreeing the window first is arguing about nothing.
What is an attribution window?
It is the period after seeing or clicking an advert within which a conversion still gets credited to it. A seven day window credits a purchase made six days after the click and ignores one made eight days after. The window is a choice, not a measurement.
That last point is the one worth sitting with. Nothing about the buyer changed between day six and day eight. Only the setting did. Two accounts running identical campaigns with different windows will produce different return on ad spend figures and both will be reporting honestly.
What does seven day click, one day view actually mean?
Two windows in one setting. Credit a conversion if it happened within seven days of somebody clicking the advert, or within one day of them merely seeing it without clicking. The click side is the longer, more defensible half. The view side is where the arguments start.
A click is a deliberate act by a person who chose to engage. A view is an impression that may have scrolled past at speed in a feed. Both sides of that setting feed the same headline total, which is why two campaigns with the same reported conversions can be built from completely different evidence.
Should you count view through conversions?
Report them separately, never inside your headline number. A view through conversion means somebody saw an advert and later converted without touching it. Some of that is real influence. Much of it is coincidence, and it flatters upper funnel and broad reach campaigns far more than it flatters anything else.
The mechanism is simple arithmetic rather than bad faith. A cheap broad reach campaign puts impressions in front of an enormous number of people, including most of the people who were going to buy from you anyway. Give it credit for a share of those sales and it will look like the best performing thing in the account. Switch view through off and the same campaign can go to nearly zero. Neither view is the truth on its own, which is exactly why they belong in separate columns.
Which attribution window should you choose?
Start from how long your buyers actually take, then round down rather than up. If most enquiries arrive within a day of the click, a short window is honest. If your buyers deliberate for a month, a seven day window will permanently understate the channel that starts the conversation.
- Measure your real lag first. Pull the gap between first touch and enquiry from your CRM for the last ninety days and use the point most enquiries fall inside, not the longest one you can find.
- Keep the click window longer than the view window, always. A view credited for as long as a click will quietly reallocate your budget towards reach.
- Use the same window on every channel you intend to compare. Platform defaults differ, so leaving them alone guarantees an unfair comparison.
- Change it at most once a year, at a period boundary, and write down the date.
- Hold the window steady while you test anything else. A window change during a creative test makes the test unreadable.
How does a long sales cycle break a short window?
Silently, which is the problem. Every conversion that lands outside the window simply does not appear, so the channel does not look worse, it looks smaller. Spend gets moved to whatever converts fastest, the slower channel shrinks, total enquiries fall a quarter later and nothing in the report explains why.
This is the standard failure in business to business accounts, where a buying decision passes through several people over weeks. The channel that creates the first conversation gets credited with almost nothing, the branded search that closes it gets credited with almost everything, and the logical next step looks like cutting the thing that generates the demand. We cover the cycle length side of this in how long a B2B sales cycle really runs.
Does changing the window rewrite your historical data?
On most platforms the reporting recalculates, so yesterday looks different today and your year on year comparison quietly breaks. That is reason enough to change a window rarely and deliberately, and to write down the date you changed it so the step in the chart has an explanation attached.
Keep a one line change log next to your reporting: what changed, when, and why. It costs nothing and it is the only thing that will stop somebody six months from now reading a settings change as a performance collapse.
How do you compare two channels fairly?
Only by giving both the same window, which often means neither platform default. A channel on a twenty eight day click window compared with one on seven days is not a comparison, it is an accounting artefact. Normalise first, then look at cost per qualified enquiry rather than cost per conversion.
Two further things need to match before a cross channel comparison means anything: the conversion action has to be defined the same way on both sides, and you have to know which system is your decision maker, because analytics and the ad platforms will disagree even on identical settings. That second problem has its own article, why your analytics and your ad platform disagree, and the enquiry value arithmetic sits in what a lead is actually worth.
What should you actually judge?
Trends on a fixed window, not absolutes on a moving one. Fix the window, keep it fixed, and ask whether cost per qualified enquiry is improving month on month inside your own account. That question survives every platform change, every consent shift and every attribution update they ship.
Absolute figures borrowed from somewhere else will not survive any of those things, because you have no idea what window produced them. Your own trend line, on a setting you chose on purpose and have not touched since, is the only performance claim you can defend. If nobody can tell you what window your reporting currently runs on, book a call and we will fix the measurement before we touch the spend.