Home/Blog/Conversion Funnels/How to Find Where a Funnel Loses People:…
Conversion Funnels

How to Find Where a Funnel Loses People: Rank by Volume

How to find where a funnel loses people: multiply each step's drop-off by its volume, rank by absolute loss, weigh by effort, and check segments before acting.

Cover of the FIKR Space guide: How to Find Where a Funnel Loses People: Rank by Volume

To find where a funnel loses people, multiply the drop-off at every step by the number of people entering it, rank the steps by the result, and weigh that ranking by how hard each fix would be. The step with the ugliest percentage is often not the one losing the most people.

Localising the problem is the first real gain from a funnel. Slow growth is not a diagnosis, but a named step with a known loss is. This post covers how to read the numbers to get there. What to do once you have found the step is a separate question, and it is handled in how to fix funnel drop off.

How to find where a funnel loses people before you trust the numbers

A ranking is only as good as the counts underneath it, so two conditions come first.

The first is that every step is a state reached in order, as set out in how to define funnel steps. If someone can reach step four without step two, the drop-off from two to four is meaningless and any loss you find there is fiction.

The second is that each step counts only the people who completed every step before it. If your report simply counts how many people did each thing in a period, the group at step one and the group at step three may be different people. The shape will still look like a funnel and every drop-off percentage will be wrong, so you could end up fixing a step that is not losing anyone. The details are in funnel cohort analysis.

It is also worth checking the drop-off numbers against something you know independently. If a step reports a loss that contradicts what you can see with your own eyes, suspect the counting before you suspect the customers.

Why the worst percentage is a trap

The instinct is to find the step with the ugliest conversion rate and fix that. It is frequently the wrong target, because a percentage ignores how many people were there to lose.

The guide gives a clear pair to compare. A step converting at ten percent with two hundred people entering is losing a hundred and eighty. A step converting at sixty percent with five thousand people entering is losing two thousand. The first looks dreadful and the second looks healthy, yet the second is losing more than ten times as many people.

StepConversionPeople enteringPeople lost
Small step, ugly rateTen percentTwo hundredA hundred and eighty
Large step, decent rateSixty percentFive thousandTwo thousand

Ten percent of a small number is a smaller prize than a modest improvement on a large one. So the rule is simple: rank by absolute loss, not by percentage.

Free guide, 30 pages: The Funnel Playbook. Send me the guide.

The method, step by step

  1. List every step with the number of people who entered it. Use cohort counts, so each step includes only people who passed the ones before.
  2. Work out the drop-off at each step. That is the share who entered and did not reach the next step within your time window.
  3. Multiply the drop-off by the volume. The result is the number of people lost at that step.
  4. Rank the steps by that number. The top of the list is where most people are leaving.
  5. Weigh each by effort. Ask how expensive the fix would be, and pick the largest loss that is also cheap to address.
  6. Check the segments. See whether the loss belongs to the step or to one audience passing through it.

Nothing here needs special software. A spreadsheet with a column for entering, a column for lost and one for effort will do the job, provided the counts in the first column are honest.

Weigh the loss by effort

The largest loss overall is not always the right first project. The guide is explicit that the correct first project is usually the largest loss that is also cheap to address. A fix that takes a quarter of work to rescue a modest number of people may lose to a fix that takes an afternoon.

LossCost of the fixSensible order
LargeCheapFirst
LargeExpensivePlan it, after the cheap wins
SmallCheapOnly if nothing larger is waiting
SmallExpensiveLeave it

This is a judgement, not a formula. What matters is that you make it in the open, with the loss written down next to the cost, so that the choice can be examined later.

Check whether the loss is the step or the audience

Before you commit to a step, look at the people who passed through it. A single overall rate is an average across audiences that behave very differently. The loss you have found may belong to one group, such as visitors from a particular source or users of one kind of device, and not to the step at all.

If so, the step itself may be fine. Fixing it would be wasted effort, and the right response might be to change who you send there, or to stop paying for the traffic that is dragging the number down. This is covered in full in how to segment a funnel. For now, the habit to build is simple: never accept a loss at face value until you have asked who it is made of.

Watch for losses that are really measurement problems

Some of the sharpest drops are not customers leaving. They are instrumentation losing the thread.

  • A cliff at signup. If the anonymous visitor and the known user are not recognised as the same person, the funnel breaks at exactly the point it becomes interesting.
  • A step that never fires on one device. It will look like everyone on that device left.
  • An event that fires twice. It will make a step look healthier than it is.
  • Tracking that never ran. Declined consent can mean part of the audience is simply absent from the counts.

Walk the funnel yourself, from a device that is not yours, and confirm your own journey appears in the right order. Twenty minutes spent that way can save a month of confident decisions. Putting the events in place properly is a separate job.

FIKR Content, multi-channel publishing: reads, listens, opens, clicks in one dashboard. See FIKR Content.

Write the finding as one sentence

Once you have the top of the ranking, state it in a single sentence that has a location, an owner and a plausible cause. A statement that many people who start the signup do not finish it is the right shape: it names the step, it points to someone who can look into it, and it suggests where to look.

This small act matters more than it seems. A sentence like that ends the unproductive argument in which every department proposes its own solution to slow growth. Everyone is now looking at the same place, and that shared view is a large part of the value of measuring anything in a small company. Write the sentence down, put the date on it and keep it with the ranking, so that next month you can see whether it changed.

What the ranking can and cannot tell you

The ranking gives you a location and a size. It does not give you a cause. The funnel tells you where, and it very rarely tells you why. That is why the next stage is to look at the failing step with your own eyes, by watching a handful of people go through it or asking a few who abandoned.

It also does not tell you whether the funnel is the right place to be working at all. If every step converts reasonably and the totals are still too small, you have a volume problem. If every step is mediocre and none is catastrophic, the offer is the likelier suspect. A ranking of losses is useful in those cases too, because a flat list with no clear leader is itself the finding.

Free guide, 30 pages: The Funnel Playbook. Send me the guide.

A short routine for the month

  1. Confirm the counts are cohort counts and that your own walkthrough still passes.
  2. Rebuild the ranking from the latest numbers, not from memory.
  3. Compare it with last month's ranking and note which step moved.
  4. Pick the largest cheap loss and write down what you think is causing it.

The last item hands you on to the loop described in how to run a funnel improvement loop, where the constraint moves each time you fix a link.

FIKR Content, multi-channel publishing: reads, listens, opens, clicks in one dashboard. See FIKR Content.

Frequently asked questions

Should I always fix the step with the worst conversion rate?

No. A low rate on a small number of people can lose fewer customers than a decent rate on a large number. Rank by absolute loss, then weigh by effort.

What if two steps lose about the same number of people?

Choose the one that is cheaper to address, and check the segments for each. A loss that belongs to one audience may need a different response from a loss spread evenly.

How do I know the drop-off is real?

Check that the steps are sequential, that counting is by cohort, and that your own walkthrough appears correctly in the data. Then compare the numbers with something you know independently.

Does finding the step tell me what to change?

Not by itself. It tells you where and how large. The cause needs qualitative evidence beside the numbers.

The full method, with the checklists, is in The Funnel Playbook, a free 30-page playbook.

LG

Luis Goncalves

// Founder & CEO at FIKR Space

Three-time founder. Built and exited Evolution4All before this. Now building FIKR Space — the operating infrastructure underneath every innovation ecosystem (startups, accelerators, governments, investors). Lisbon-based, works global.

Claim your founding member spot.

Every founding member is hand-reviewed. Tell us who you are and how you want to co-create — we get back inside 48 hours.

Role
Job Title
Country
Cookies

We use cookies.

Essential ones keep the site working. The rest only run if you say so. No dark patterns, no pre-ticked boxes.Privacy policy

Co-create with us. It's about building it together.Co-Create With Us