Conversion Funnel Calculator

This calculator answers a different question from its siblings. The conversion rate calculator measures a single number, the site wide rate, and compares it to your sector. The conversion rate impact calculator turns a rate gain into money. Here the subject is the path: where people vanish between the first visit and the conversion, and which of those leaks is worth plugging first.

How to use it

1. Type the name of each stage of your funnel, from top to bottom, in the order people go through them.
2. Enter how many people reached each stage in the same period (same base: either unique users or sessions, never mixed).
3. If your funnel has fewer than five stages, leave the extra rows blank: they are ignored.
4. Set how much of the drop-off you think you can recover. Ten percent is a conservative starting point for a solid improvement.
5. Read the table: the highlighted row is the bottleneck that returns the most final conversions, and the line above it also names the step that loses the most people in raw numbers.

How it works: the formula

The easy part is the drop of each step. The part almost no calculator does is telling you which drop is worth the most money, because a gain at the top arrives diluted at the bottom:

Dilution is the heart of the tool. Recovering a thousand people at the last step turns into a thousand conversions, because there is no other step left to survive. Recovering the same thousand people at the top of a funnel with three steps ahead turns into whatever survives those three pass rates. That is why the biggest leak in headcount is rarely the right priority.

Worked example (reproduces the default result)

With the e-commerce funnel already filled in: 100,000 visitors, 40,000 viewed product, 12,000 added to cart, 6,000 started checkout and 3,000 purchased, with a 10% recovery.

• Pass rates: 40% (visit to product), 30% (product to cart), 50% (cart to checkout) and 50% (checkout to purchase).
• Funnel rate: 3,000 ÷ 100,000 = 3.00%. People lost along the way: 97,000.
• Step 1: 60,000 lost, dilution 0.30 × 0.50 × 0.50 = 0.075, gain = 60,000 × 0.10 × 0.075 = 450 purchases.
• Step 2: 28,000 lost, dilution 0.50 × 0.50 = 0.25, gain = 28,000 × 0.10 × 0.25 = 700 purchases.
• Step 3: 6,000 lost, dilution 0.50, gain = 300 purchases. Step 4: 3,000 lost, dilution 1, gain = 300 purchases.

That is exactly what the tool shows when the page loads: a 3.00% rate, 97,000 people lost and the bottleneck at "viewed product to added to cart", worth 700 extra purchases. Note the useful paradox: the first step loses more than twice as many people, yet returns less than half the conversions.

How to read it, and where it misleads

A funnel is a model, not reality. It assumes a single sequential path, when in practice people skip steps, come back days later through another channel and convert over the phone. If your analytics counts every visit as a new person, the pass rates come out lower than the real experience of a buyer. Treat the numbers as a comparison between steps, which is where they are reliable, not as absolute truth about individual behavior.

The estimated gain is optimistic by construction: people who drop off have lower intent on average, and the model assumes the recovered ones behave like the average of those who already advance. Add seasonality and channel mix shifts and the number becomes an upper bound. It exists to rank priorities, not to promise revenue. To turn the gain into money, take the result to the conversion rate impact calculator.

One measurement detail can wreck the whole analysis: stages measured over different periods, or one in users and another in sessions. If any stage shows more people than the one before it, the tool warns you. Fix the tracking before concluding anything.

From bottleneck to test

Finding the bottleneck is half the work; the other half is proving your idea improves that specific step. Write the hypothesis, run an A/B test on that step alone and use its pass rate as the primary metric, with final conversion as a guardrail.

• Size the test before launching with the sample size calculator: the bottleneck step usually gets far less traffic than the homepage.
• Confirm the result with the statistical significance calculator before celebrating any lift.
• Re-run the funnel analysis after the test. Fixing one bottleneck pushes the problem to the next step, and the priority changes.
• If volume is small, attack the lower steps first: they convert nearly one to one and need less sample to prove the gain.

The full context on how to prioritize improvements is in the conversion rate optimization guide. If your overall rate looks low but you have nothing to compare it to, see the conversion rate benchmarks.

FAQ

How do I calculate the conversion rate of each funnel step?

Divide the people at the next stage by the people at the current stage. If 40,000 viewed a product and 12,000 added to cart, the pass rate of that step is 12,000 / 40,000 = 30%, and the drop is 70%. The rate of the whole funnel is the last stage divided by the first, which is also the product of every pass rate.

Is the step that loses the most people always the one to fix first?

No, and that is where most funnel analyses go wrong. Anyone you recover at the top still has to survive every later step and arrives diluted at the end. In the default example on this page the first step loses 60,000 people and the second loses 28,000, yet recovering 10% of the second yields 700 extra purchases against 450 from the first. Prioritize by final conversions gained, not by raw loss.

How many stages should my funnel have?

Use three to five stages that map to real, sequential steps with a stable definition in your analytics. Too many stages fragment the volume and leave each step with too few people to analyze; too few hide where the leak actually happens. Leave unused rows blank.

Can I use sessions instead of unique users?

Yes, as long as you use the same base for every stage. Mixing unique users at the top with sessions in the middle inflates the lower stages and distorts the pass rates. Pick one base and keep it from start to finish.

Does the calculator guarantee those extra conversions?

No. It sizes the prize of each step assuming that whoever you recover behaves like the average of those who already pass, which is usually optimistic, because people who drop off have lower intent. Use the number to choose where to test first, then confirm the real gain with an A/B test.

Why does one stage show more people than the stage before it?

In a sequential funnel that should not happen, so it is almost always a wrong event definition, a different date range between stages, or counting in different bases. The tool flags this case. Fix the measurement before drawing any conclusion.

Keep going

With the bottleneck in hand, the next step is testing. Start with the CRO guide, size the experiment in the sample size calculator and validate the outcome in the significance calculator. If you do not run tests yet, start with what A/B testing is.