How a conversion funnel works
A conversion funnel is simply the ordered list of steps somebody has to complete before they become a customer, written down so you can measure each one. For a shop that might be product page, add to basket, checkout, payment, confirmation. For a consultancy it might be advert, landing page, form start, form submit, qualified call.
It is called a funnel because each step holds fewer people than the one before it. That narrowing is normal and unavoidable — not everybody who reads a product page intended to buy today. What makes the funnel useful is the shape of the narrowing. A gentle taper across every step tells you a different story from a cliff between two particular steps.
Two things make a funnel worth building rather than guessing at. The steps must be in a genuine sequence, so a person cannot reach step three without passing step two, and each step must be measurable as an event rather than inferred. A funnel made of steps you have not actually tracked is a diagram, not a measurement.
Why the conversion funnel matters
An overall conversion rate tells you that something is wrong. A funnel tells you where. If almost everybody who starts the form finishes it, spending money on shorter forms is wasted effort and the real problem sits earlier, in the traffic or the offer. If people arrive in numbers and never start the form, the page has not persuaded them, and no amount of extra traffic will change that.
It also protects budget. Sending more paid traffic into a funnel with a broken step means buying the same loss repeatedly at a larger scale. Fixing the step first makes every subsequent visit worth more, which is usually far cheaper than raising bids.
Common mistakes with conversion funnels
The most common is drawing an idealised funnel that nobody actually walks. Real buyers arrive part-way in from a search result, leave, come back on a phone a week later, and finish on a laptop. If your funnel assumes a single tidy session, its numbers will be pessimistic and its conclusions wrong.
The second is having steps at the wrong grain. A funnel with two steps hides everything interesting; a funnel with a dozen produces losses so small at each stage that nothing looks urgent. Aim for the smallest number of steps where a loss at any one of them would prompt a specific fix.
The third is treating the funnel as a report rather than a diagnosis. Knowing the checkout loses people is not a finding. Watching what those people do before leaving is.
How to act on it
Map the steps on paper first, with the business, before touching any tool. Then instrument each step as its own event so the counts are observed rather than assumed. Compare the same funnel on mobile and desktop separately — on Nepali sites the difference is often stark, because most people arrive on a phone and payment or form steps behave differently there.
Once you can see the largest single drop-off, work on that step alone until it moves, then re-measure. Changing several steps at once leaves you unable to say what helped.