▤The Test Group experiment running Open the partner account
paid
Affiliate disclosure. The partner link in the masthead and in the band beside the copy on this page is a sponsored link to a partner operator, and this site may be paid if you open an account through it, at no extra cost to you. It carries rel="sponsored noopener" and opens in a new tab. A desk about how a screen is measured and tested should not leave its own funding unsaid: one link funds the site, no operator and no product is named, rated or recommended anywhere on it, and this site runs no analytics of its own on its readers.
The Test Group / The funnel
Six steps from a visit to a first bet, and where they are lost

The funnel to a first deposit

A rate on one screen is only meaningful inside an order: land, register, open the deposit page, submit, confirm, bet. The funnel is that order, and its purpose is to tell the operator which step to work on. Of 100,000 visits the samples end in 2,074 first bets, and more than three quarters of the loss happens before the deposit step a test usually touches.

Desk spec
steps
6
visits
100,000
first bets
2,074
end to end
2.1%
the eventOne interaction, written down. A click carries a name, a time, an account, a session and what was on the screen, and it is kept whether or not the reader chose to be measured.
the funnelThe order the steps happen in. A landing visit becomes an account, an account becomes a deposit page, and 100,000 visits end in 2,074 first bets - a 2.1% path the whole loop is aimed at.
the armsTwo versions of one screen shown at the same time, and a rate for each. 4.30% against 5.16% is a 0.86-point lift, and the interval decides whether it is a result or a coincidence.
Direct answer

A conversion funnel is the fixed order of steps a visitor must complete, with the count at each one. On the samples 100,000 visits become 24,000 accounts, 9,600 deposit pages, 4,320 submissions, 3,456 confirmed deposits and 2,074 first bets - 2.1% end to end, with 76.0% of the traffic lost before the deposit step.

The six steps and the rate between them

Each step is a rate on the step before it, and each rate is a different kind of problem. A 24.0% registration rate is a proposition problem, a 40.0% deposit-page rate is a motivation problem, and the 80.0% confirmation rate is mostly a payment problem.

Sample C - the six steps, with the rate at each and the drop it represents
StepCountOf the step beforeLost at this step
land on the site100,000--
create an account24,00024.0%76,000
open the deposit page9,60040.0%14,400
submit a deposit4,32045.0%5,280
deposit confirmed3,45680.0%864
place a first bet2,07460.0%1,382
end to end2,0742.1%97,926
sample C - the two ways to read a funnel by count, the largest single loss is registration: 100,000 - 24,000 = 76,000 visitors, or 76.0% of all traffic by proportion, the largest loss is the account-to-deposit step: 9,600 / 24,000 = 40.0%, so 60.0% of the accounts stop here the deposit step a test usually edits is the fourth: 4,320 / 9,600 = 45.0%, and the loss available to it is 5,280 so a 20.0% lift on the deposit step is worth 5,280 x 0.20 = 1,056 extra submissions, against 14,400 accounts that could have been reached instead by working on the step before it.

Why a funnel can only measure what it can see

A funnel's steps are the ones the site owns. Anything a reader does off the site - reading a review, comparing a price, waiting for a bank transfer, deciding not to deposit at all - is invisible, and it is recorded as a drop at whichever step came last. That is why a funnel gap is a question rather than a diagnosis.

Sample C - what a funnel can and cannot attribute
The gapCountWhat the funnel can say
never created an account76,000they left before the first step the site controls
account, no deposit page14,400something between the two screens, inside the site
deposit page, no submission5,280something on the page, or the amount, or the method
submitted, not confirmed864mostly a payment decision the site does not make
confirmed, no first bet1,382the money arrived and nothing happened next
the funnel's blind spot-everything that happened outside the six steps, which is recorded as a loss at the step it left
sample C - the size of the blind spot of the 76,000 who never created an account, the samples attribute 42,000 to a deliberate comparison with another site, 19,000 to a decision not to gamble at all, 9,000 to a page that did not load and 6,000 to an age or territory check so the funnel records 76,000 "not interested" and the site can act on one of the four causes - the 9,000 - which is 9,000 / 76,000 = 11.8% the other 88.2% is a cause the step cannot reach, and every change made to the landing page is spent against all of it.
The counts are invented and the proportions are illustrative. A real funnel's numbers come from the operator's own event log, which means they share its blind spots: unmeasured visitors, a declined consent, a blocked script and a different device all remove people from the denominator without leaving a trace.
Reading a funnel — or a rate quoted from one — without being misled
  • Check the denominator, because "of visitors" and "of accounts" are different claims.
  • Ask which visitors were measured, since a consent rate removes part of the population.
  • Compare a step's rate with the step before it, never with the traffic at the top.
  • Remember that an off-site cause is recorded as a drop at the last on-site step.
  • Treat a step's improvement as a claim about that step, not about the reader's intentions.

Read next

The two armsHow a step is tested rather than changedThe eventsThe raw material the steps are built fromThe limitsWhat a change to a step cannot fixThe metricThe number a funnel step is judged on