How the conversion window works
When someone clicks an ad, Google starts a timer. If the person converts before the timer runs out, the conversion is credited back to that click. If they convert after, it is not. The window is configurable per conversion action, up to 90 days for clicks, and it applies from the moment of the interaction. This matters because bidding learns from the conversions that land inside the window, so the setting quietly defines what the algorithm believes a click is worth.
Matching the window to your sales cycle
The right window is a factual question about your buyers, not a preference. An impulse ecommerce purchase mostly completes within a day or two, so a short window captures nearly everything. A B2B SaaS trial or a legal consultation can take weeks of consideration before the meaningful conversion happens, and a 30-day window would miss a large share of them. The test is to look at the actual distribution of time from click to conversion and set the window to capture the bulk of it, rather than accepting a default that was designed for a different business.
The cost of getting it wrong
A window that is too short understates conversions on considered purchases, which makes profitable campaigns look weak and pushes Smart Bidding to pull back on exactly the audiences that convert slowly but well. A window that is too long can over-credit and delay the feedback bidding needs. There is also a reporting trap: changing the window changes historical-looking numbers, so comparisons across a change have to be handled carefully. The window is not a set-and-forget checkbox, it is a modelling decision.
How iClick sets conversion windows
iClick sets the conversion window from the observed distribution of click-to-conversion time for each business, not from the default. The method is to pull the actual conversion-lag data, find where the bulk of conversions land, and set the window to capture them without stretching so far that bidding feedback goes stale. The rule is that the window should describe how customers really decide, because a window that misrepresents the sales cycle teaches the algorithm the wrong lesson.
