What we mean by funnel drop off
A funnel is a sequence of steps someone has to complete before they become whatever you’re counting as a win: a signup, a paid customer, a booked call. Drop off is just the gap between how many people enter a step and how many make it to the next one. Land 500 people on a pricing page, get 40 to start checkout, and you have an 88% drop off at that step.
Counting it is the easy part. Any analytics tool will hand you the percentages. The hard part, and the part most people skip, is figuring out what the number is actually telling you before you go change something.
I run a few small businesses off numbers like this, and the mistake I made early on was treating a big drop off percentage as a diagnosis by itself. It isn’t. It’s a symptom. It tells you where to look, not what’s wrong.
Draw the funnel before you diagnose anything
Before you can diagnose drop off, you need an honest map of the steps a person actually goes through, not the steps you assume they go through. This sounds obvious and it’s the step people skip most often.
For a small SaaS product, the real funnel usually looks something like: ad or search click, landing page, signup form, email verification, first login, first meaningful action inside the product, upgrade to paid. That’s seven steps, not two. If you only track “visitors” and “customers,” you’re averaging together seven different failure points into one number, and that number won’t tell you whether your problem is the landing page or the onboarding flow.
Write the steps down. Use whatever your customers actually click through, not the flow in your head. If you don’t know the exact steps, that’s the first thing to fix, before you touch a single metric.
Look at drop off per step, not just top to bottom
Once the funnel is mapped, calculate the conversion rate between each consecutive pair of steps, not just the rate from the very top to the very bottom. A single blended number hides where the actual problem lives.
Say a landing page converts 30% of visitors to signup, then only 15% of those verify their email, then 70% of verified users complete onboarding. The overall visitor-to-onboarded rate looks low, maybe 3%, and it’s tempting to conclude the whole funnel is broken. But the data points at one step: email verification. That’s a much smaller, much more specific problem than “our funnel doesn’t work.” Maybe the verification email lands in spam, maybe the link expires too fast, maybe the copy is confusing about what to do next. Each of those has a concrete fix. “Our funnel doesn’t work” doesn’t.
This is the core discipline of diagnosing drop off: resist the urge to explain the whole funnel with one story. Find the step with the worst relative drop, and start there.
Common places people actually leave, and why each one is different
A few patterns show up often enough across small businesses that they’re worth naming, though which one applies to you is something only your own data can tell you.
Drop off right after a form usually points to friction in the form itself: too many fields, a confusing label, a required field that shouldn’t be required. This is often the cheapest thing to test because the fix is small and the step is early, so even a modest improvement compounds through the rest of the funnel.
Drop off between signup and first login often points to a broken or delayed confirmation step, not a lack of interest. Someone who filled out a form already showed intent. If they don’t come back, check whether the email actually arrives, how fast, and whether the next step is obvious once they’re in.
Drop off between first login and first meaningful action usually points to onboarding, not acquisition. The person showed up, they’re just not sure what to do once they’re there. This is where I’ve seen the biggest swings from small changes, because the person is already present and just needs a clearer next step, not more marketing.
Drop off right before payment is different again. By that point someone has invested real time, so the drop is more often about price, trust, or a checkout flow that adds unexpected steps or costs. It’s worth checking whether the price shown at checkout matches what was promised earlier in the funnel. Mismatches there are more common than people expect and easy to miss if you’re not looking at the funnel end to end.
I’m not going to tell you which of these applies to your business, because I don’t know your funnel, and neither does any generic list. The point of naming them is to give you starting hypotheses to test against your own steps, not conclusions to copy.
Segment before you conclude anything
A funnel percentage is an average across everyone who entered it, and averages hide differences that matter. Before deciding a step is broken, split the data by the ways your traffic actually differs: paid versus organic, mobile versus desktop, new visitor versus returning.
It’s common for a step to look fine in aggregate while one segment is dragging it down badly. A checkout step might convert well on desktop and poorly on mobile because a button is hard to tap. Averaged together, that shows up as a mediocre number that doesn’t clearly point anywhere. Split apart, it points exactly at a mobile layout problem.
This is also where I’d caution against over-segmenting. If you slice the data into ten segments with twenty people each, you’re not looking at a pattern anymore, you’re looking at noise. Segment enough to test a specific hypothesis, not so much that every group is too small to mean anything.
Separate real drop off from normal attrition
Not every person who doesn’t convert was ever going to convert. Some fraction of any funnel is made up of people browsing, comparing you to competitors, or just not ready to buy yet. That’s normal attrition, and no amount of onboarding polish removes it.
The way to tell the difference is trend, not snapshot. If a step’s conversion rate has been stable for months and suddenly drops after a change you made, that’s a signal worth acting on. If a step has always converted at roughly the same modest rate, that might just be the natural shape of that part of your business, and chasing it down as a “problem” can burn time for little return.
This is also where I’d push back on treating any single week’s numbers as a verdict. Small businesses often don’t have enough volume for a week of data to be reliable. Look at a longer window before deciding a drop off is a real, new problem rather than ordinary variation.
Deciding what to fix first
Once you’ve found a few steps with real, sustained drop off, you still have to choose where to spend time, because you probably can’t fix everything at once. A reasonable way to prioritize is to weigh two things: how many people are affected at that step, and how close that step is to the final outcome you care about.
A step early in the funnel with a lot of volume and a bad drop off is often worth fixing first, simply because a small improvement there feeds every step after it. A step near the end with fewer people but high intent is also worth attention, because those are people who almost converted and didn’t. Steps in the middle with modest volume and modest drop off are usually the lowest priority, even if the percentage looks bad in isolation.
What funnel data can’t tell you
A drop off number tells you where people left. It doesn’t tell you why. That gap matters, and it’s tempting to fill it with a story that fits whatever change you were already planning to make.
To actually understand why, you generally need something beyond the funnel report itself: watching a handful of session recordings, asking a few people directly why they stopped, or reading support messages from that stage of the journey. The funnel narrows the search. It doesn’t finish the investigation.
It’s also worth saying plainly that fixing a drop off step is not guaranteed to move revenue. Sometimes the people lost at a given step were never going to pay, and improving that step just moves the drop off further down the funnel instead of removing it. Treat every fix as a hypothesis to test, not a certainty, and check the step after the one you changed to see whether the improvement actually held.
Putting it into practice
Diagnosing funnel drop off is less about finding one dramatic number and more about a habit: map the real steps, look at each transition on its own, segment before concluding, separate signal from normal variation, and prioritize by volume and proximity to the outcome you actually care about. None of that requires expensive tooling. It requires writing the funnel down honestly and looking at it regularly enough to notice when something changes.
If you want more on how we think about metrics like this for small, one-person-run businesses, you can find the rest of our writing on the home page.
Get new guides and videos first — join the Telegram channel.