Impression → click → install → signup → purchase. That step-by-step path, where the audience shrinks at every stage, is a funnel — named for the shape it makes, wide at the top and narrowing downward.
Why it matters
If you want to raise conversion but don't know where to start, map the funnel. The pass-through rate between each stage shows exactly where the biggest leak is.
A common trap
The lowest pass-through stage isn't always the top priority. Weigh room for improvement × traffic volume — a stage that's already seen by a lot of users can move more total conversions with a small improvement than a low-traffic stage with a low rate.
Run the numbers
Take a funnel of 100,000 impressions to 2,000 clicks to 800 installs to 200 signups. The step rates are 2%, 40% and 25%. Lifting the weakest of those, install to signup, from 25% to 30% gets you 240 signups — 40 more.
Now lift the very first rate instead, click-through from 2% to 2.5%. That gives 2,500 clicks, 1,000 installs and 250 signups: 50 more, without touching a single downstream rate. This is why the stage carrying the most traffic often moves more absolute conversions than the stage with the worst rate.
There is no fixed number — split it only as far as you can act on each stage separately. For apps, impression, click, install, signup and purchase is a common starting point. Splitting further adds numbers without adding decisions if you have no way to improve that stage on its own.
Which stage should I fix first?
Not the stage with the lowest pass-through, but the one with the largest room for improvement times traffic. In a funnel of 100,000 impressions to 2,000 clicks (2%) to 800 installs (40%) to 200 signups (25%), lifting the weakest stage from 25% to 30% adds 40 signups. Lifting click-through from 2% to 2.5% adds 50, with every later rate untouched.
Is a funnel pass-through rate the same as CVR?
CVR usually names one span, click to conversion, while funnel pass-through looks at each step separately. The same 60 conversions read as 3% against 2,000 clicks and 5% against 1,200 landing visits, so line up the denominator before comparing anything.