Registration funnel

Open Journal -> Advanced analytics -> Funnel. Compare registered accounts with those who successfully signed in and reached a first payment. The examples below use fictional training data.

Registration and observation periods

Choose the registration period first. The cohort consists of accounts first registered within these dates. Then choose how long to observe their logins and payments. You can compare the same cohort on its registration day and several weeks later.

Field	Meaning
Period	Last 30 days by default, another preset, current or previous calendar year, or custom dates.
Registration from / Through (UTC)	Both registration dates are inclusive.
Project	All available projects or one project. Accounts in different projects are counted separately.
Observe through	Last included day of logins and payments. Cannot precede the end of the registration period or exceed today. Select today if the registration period has not ended yet.

All boundaries use UTC. Only elapsed time is available for today. Refresh applies the filters; reset restores their defaults. The page URL retains the selected conditions.

The report concerns project master accounts, not characters or individual game servers, so it has no server filter. One person with multiple accounts contributes multiple accounts.

Of 36 registrations, 27 accounts logged in and 24 made their first payment after a login. Two more accounts paid outside that sequence.

Stages and conversion

1. Registered: accounts with their earliest known registration inside the chosen period. Repeated registration records do not add another account.
2. Logged in: accounts with a successful login after registration and by the observation cutoff. Page visits and failed attempts do not qualify.
3. First payment after login: the first qualifying payment occurred no earlier than the first successful login after registration.

Registration, login and payment may have equal timestamps. Multiple logins or payments do not increase the account count.

Metric	Calculation
Registration -> login	Logged-in accounts / registered accounts x 100%.
Login -> first payment	Accounts completing the full sequence / logged-in accounts x 100%.
Full funnel conversion	Accounts completing the full sequence / registered accounts x 100%.
Percentage on a stage card	Accounts at that stage / registered accounts x 100%.

In this example, full conversion is 24 / 36 = 66.67%, and login-to-payment conversion is 24 / 27 = 88.89%. A missing denominator displays Not available, not a percentage.

All paying accounts includes every cohort account with a qualifying payment after registration. No payment yet is the rest of the cohort at the observation date.

Paid outside the sequence means there is no recorded successful login before the first payment. These accounts are part of all payers, not an additional population. Payment may have preceded login, or login history may be incomplete. A later login and another payment do not change the order of the first payment.

Included payments

Positive settled balance top-ups and purchases associated with a registered account qualify. Test operations, administrative credits and bonus grants alone are not customer payments. Guest payments without a registered account do not belong to this funnel.

Currencies are not added together: the report counts accounts, not money. An account paying in several currencies is still one payer. A refund does not erase a previously completed payment; use payment statistics to inspect amounts and refunds.

Observing through August 1 retains 36 registrations but only 11 payers, of whom 9 completed the sequence. Later payments do not contribute yet.

Time to first payment

Timing starts at registration and ends at the first qualifying payment after it. It includes all cohort payers, including those outside the sequence. Nonpayers are not treated as zero-duration samples and do not reduce the average.

• Average is sensitive to accounts paying much later than others.
• Median describes the middle of the distribution and helps estimate typical waiting time.
• 25th, 75th and 90th percentiles describe early and late payments. Values between adjacent observations are interpolated.

Labels use seconds, minutes, hours or days. The calculation retains fractions of a second. The six bands cover all durations: under an hour; one hour to a day; one to three days; three to seven days; seven to thirty days; and thirty days or more. Lower bounds are inclusive and upper bounds exclusive: exactly one day belongs to 1 to under 3 days.

Each band's share divides by all paying cohort accounts, not registrations or payment count. Rounded percentages may not sum to exactly 100%.

Zero duration is a real payment at the same recorded time as registration. The payer belongs to the first band.

Calendar years and available history

Select a calendar year or a custom interval of available history without a ten-year limit. Registration dates are not payment dates: an August account first paying in September remains in the August cohort when observation includes September.

An unfinished year's observation stops at today. Cohorts observed for different lengths of time are not equally mature.

Empty results and errors

This account has registered and logged in but has not paid by the observation date. Timing is unknown and shown as a dash, not zero.

No accounts registered within these dates. This is a valid empty report, not a loading failure.

New facts may arrive with a short delay. The report uses available history: an absent event does not prove an action never happened. Late-arriving records may refine historical results.

If calculation capacity is exceeded, select a project or a shorter period. During an outage, previous figures are hidden and a warning is shown. Refresh after connectivity returns. Individual login details are available in the login journal.