Attribution windows: how far back should you measure?

An attribution window is the period of time during which a touchpoint is eligible to receive credit for a conversion. Set it to 30 days, and any interaction older than 30 days before the conversion gets nothing. Set it to 90, and the window reaches three times further back into the journey.

The window is a dial, and where you set it changes the answer. The same customer, the same touchpoints, a different window, and a different channel gets the credit. For B2B SaaS, that dial decides whether attribution measures the whole buying journey or only the last few weeks of it.

Two windows, measured in opposite directions

Most confusion about windows comes from mixing up two different things platforms both call a "window."

The lookback window counts backward from the conversion. When a deal closes, how far back does the system look for touchpoints to credit? This is the one that matters most for B2B attribution, and the one this post is about.

The conversion window counts forward from a click or impression. After someone clicks an ad, how long do they have to convert for that ad to get credit? Ad platforms use this to decide when a click has gone cold. A 7-day click window means a click only counts if a conversion follows within 7 days.

Ad platforms split the conversion window further, into click-through and view-through. A click-through window credits an ad someone clicked. A view-through window credits an ad someone only saw and didn't click. View-through windows inflate an ad platform's numbers, because they claim credit for impressions the customer may never have registered.

Platform defaults are built for short cycles

The defaults across ad platforms and analytics tools tend to sit between 7 and 90 days, and they're tuned for e-commerce and direct-response, where someone sees an ad and buys within a week. That assumption breaks for B2B SaaS.

A B2B SaaS sales cycle commonly runs three to six months. A prospect clicks a LinkedIn ad in March, reads three blog posts over April, attends a webinar in May, books a demo in June, and closes in July. Against a 30-day lookback window, the March ad, the April content, and the May webinar are all invisible. Only the June demo and whatever happened in the final 30 days receive credit.

The result is predictable and wrong: last-touch channels look like heroes, and every top-of-funnel channel that started the relationship looks worthless. Budget flows to the closers, the awareness channels get cut, and the pipeline quietly dries up two quarters later.

Match the window to the sales cycle

The rule is direct: the window has to be at least as long as the sales cycle, ideally longer.

Pull the actual number. In the CRM, measure the median time from first touch to closed-won across recent deals. If that's 95 days, a 90-day window is already cutting off half your deals at the first touch. A window of 120 to 180 days is closer to right for a cycle that long.

Two failure modes bracket the decision:

Too short drops early touchpoints and over-credits the end of the journey. This is the common one, and the more damaging one, because it systematically defunds awareness.

Too long starts crediting touchpoints that had nothing to do with the deal. A window of two years will attach a blog post someone read long before they were in-market to a purchase it never influenced. The signal gets diluted with noise.

The window that fits is the one that reaches back far enough to catch the real first touch of a typical deal, and no further.

Different conversions deserve different windows

One window rarely fits every conversion type, because the stages sit at different distances from the first touch.

A demo booking or trial signup happens earlier in the journey and closer to the moment of intent. A 30-to-60-day window can be reasonable for attributing these mid-funnel conversions.

A closed-won deal sits at the far end of the whole cycle. It needs the full 90-to-180-day window to see the touchpoints that started months earlier.

Measuring both, with a window sized for each, is what pipeline attribution versus revenue attribution is really about: an early conversion you can act on now, and a late one that tells you whether the early signal held up.

Set the window before you pick a model

The window isn't a setting inside a model. It's a decision that comes before the model and constrains everything it can do. A w-shaped or data-driven model applied on top of a 30-day window can only distribute credit among the touchpoints that survived the window. If the window already deleted the first two months of the journey, no model can give them back.

That's why window-setting sits early in how to measure marketing attribution, right after capturing touchpoints and connecting them to a person. Get the window wrong and every model downstream inherits the mistake.

Start by measuring your median first-touch-to-close in the CRM. Set the closed-won window a comfortable margin above it, set a shorter window for mid-funnel conversions, and revisit both when the sales cycle shifts. The window is one number, and it decides how much of the journey your attribution is allowed to see.

Sizing windows to the sales cycle, and keeping the whole journey intact behind them, is the kind of plumbing Cascayd handles for you. Try Cascayd for free.