← Back to blog

Cohort Analysis for Escape Room Bookings: A Practical Guide

3 November 2026· 14 min read

Decorative watercolor title card frame

Cohort analysis on bookings means grouping your customers by the month they first booked, then following each group to see how many come back, cancel or no-show over time. It shows whether your business is getting better at keeping players, which a single blended repeat-rate figure hides. This guide shows an escape room operator how to define the cohorts, build the table, read it and decide what to change.

Riddlio
Understand Your Booking Performance
Riddlio brings bookings, customer communications and reporting together, helping escape room operators understand booking and revenue performance.

TL;DR

  • A booking cohort is a fixed group of customers who share a starting event, usually their first completed booking in a given month.
  • Build a table with one row per cohort and one column per month since that first booking, then read across for one group’s curve and down to compare groups at the same age.
  • Match bookings to customers by email or phone, otherwise repeat visitors are invisible.
  • Small cohorts and recent cohorts mislead. Wait for later periods to fill in before judging.
  • Treat findings as hypotheses to test, such as a deposit rule or a new room launch, not as proof of cause.

Table of Contents

What cohort analysis means for a venue

Cohort analysis groups records by a shared starting event and follows each fixed group over time. For an escape room, the starting event is nearly always a first completed booking. The group is everyone who played for the first time in, say, March. You then count how many of them booked again in April, May and onwards.

That is different from ordinary segmentation. A segment is a slice defined by an attribute, such as corporate groups or weekend bookers, and its membership can change. A cohort is locked the moment the start period ends. March’s first-time players stay March’s first-time players forever, which is exactly what makes the comparison fair.

You will also see the phrase “cohort modelling”. In practice it means the same thing: grouping records by a shared start and following each group separately. Nothing more exotic is needed to get value from it.

Why a blended repeat rate misleads

A single repeat-rate average mixes recent first-time bookers, who have not had time to return, with older customers who have. Comparing the same age across cohorts, for example month 3 for a January group against month 3 for a June group, tells you whether newer customers are returning more successfully than older ones. The blended number cannot answer that.

This matters in escape rooms because repeat visits are hard to see in totals. Tracking them requires matching bookings to identifiers such as email or phone number, since aggregate reports make repeat customers invisible.

Choose the unit, the start event and the window

Before you open a spreadsheet, make four decisions. Write them down, because they make the result reproducible. PostHog’s guidance is to define retention through a start event, a return event, a counting unit and a time interval. Here is how that looks for a venue.

1. The unit. Count unique customers, not bookings. A group of eight who book once is one customer record (the lead booker), not eight. If you count bookings instead, a single corporate client who books ten times will swamp your picture.

2. The start event. Use the first completed booking, meaning the game was played, not merely reserved. A cancelled or no-show booking should not place someone in a cohort. PostHog’s worked example assigns each user to a monthly cohort by their first-ever booking event and then counts later booking events in subsequent months. You can copy that logic directly.

3. The return event. Another completed booking, in any room. You may later run a stricter version that counts only a different room, which is useful when your catalogue is small.

4. The interval. Monthly cohorts suit most venues. Weekly cohorts only work if you take hundreds of first-time bookings each week, and quarterly ones suit seasonal or small venues. Measure age in periods since the start (month 1, month 2) rather than calendar dates, so cohorts line up. The time series must also be long enough for the behaviour you care about to occur. If your typical player returns after a birthday or a team outing months later, a three-month window will undercount.

Choose the unit, the start event and the window

The booking data you need

Most booking systems already hold what you need. For each booking, you want:

  • Customer identifier (email, ideally phone as a backup)
  • Booking date and the date of play
  • Room and time slot
  • Group size
  • Source (direct site, marketplace, gift voucher, phone)
  • Status: completed, cancelled or no-show
  • Amount paid

This matches the fields a booking and conversion report typically uses: slot, group size, source, status and booking date, alongside room revenue, cancellations, no-shows and repeat-customer frequency.

Clean the identifiers first. Lower-case emails, trim spaces, and merge records where the same person used two addresses if you can tell. Gift voucher redemptions deserve a decision: a person who receives a voucher and plays is a new customer, and the purchaser is a separate record with their own cohort.

If your bookings sit in several tools, one place for them makes this far easier. That is a large part of why operators move to an escape room booking system built around venues, where customers, slots and payments share one record.

Build the cohort table step by step

  1. Export completed bookings with customer identifier and play date.
  2. For each customer, find the earliest play date. Its month is their cohort.
  3. For every later completed booking, calculate months since the cohort month.
  4. Count unique customers per cohort per month-age. Month 0 is the cohort’s starting month.
  5. Divide each count by the cohort’s month 0 size to get a percentage.

The layout is standard: cohorts in rows and periods since the start in columns, with period 0 as the starting period and later cells showing the share who came back.

A worked example

The figures below are invented for illustration, to show how to read the table. They are not benchmarks.

Cohort (first visit) New customers Month 1 Month 2 Month 3 Month 6
January 240 4% 5% 6% 9%
March 310 5% 6% 8% 11%
June 280 7% 9% 11% n/a
September 350 8% 10% n/a n/a

The empty cells are not zeros. Recent cohorts are incomplete because they have not had enough time to mature. Mark them as not yet available, never as a drop.

Colour the cells as a heat map if you like: darker for a higher share. In a spreadsheet, a simple colour scale on the percentage block does the job.

Reading the table

There are two directions to read, and they answer different questions.

Across a row gives one cohort’s retention curve. In the example, the January cohort returns slowly and steadily, reaching 9% by month 6. Escape room curves tend to rise as a cumulative share of people who have come back at least once. You can also build the version that shows only those who return in each specific month, which is flatter and shows the timing of return visits.

Down a column compares cohorts at the same age. Read down a column to compare the same age across cohorts. Month 3 climbs from 6% to 8% to 11% in the example. Newer customers are returning more, and the useful next question is what changed between January and June.

Look for cliffs and outliers

A row that suddenly looks different from its neighbours is worth investigating. Amplitude’s advice is to compare cohorts around a product change and investigate outliers or retention cliffs. In a venue, the “product change” might be a new room, a price rise, a new reminder message, a change of booking page or a cancellation policy update.

Which cohorts to run

The first-booking-month cohort is your foundation, but it is not the only useful one. Amplitude describes acquisition cohorts for retention trends, behavioural cohorts for actions linked to retention, segment-based cohorts for business-dimension comparisons and time-based cohorts for evaluating changes. Translated to a venue:

  • Acquisition cohorts: all first-time players by month. Answers: are we keeping more of them than last year?
  • Source cohorts: first-time players by how they found you (direct, marketplace, voucher). Answers: which channel brings people who come back?
  • Room cohorts: first-time players by the first room they played. Answers: which entry room leads to a second visit?
  • Group-type cohorts: families, friends, corporate teams. Answers: who rebooks, and who needs a different offer?
  • Policy cohorts: customers who booked before and after a deposit or cancellation change.

Run one cut at a time. Cutting a monthly cohort by source and by room and by group type quickly leaves rows of ten people, which tells you nothing.

Beyond repeat visits: cancellations, no-shows and revenue

The same table structure works for other measures. Swap the return event for another outcome:

If you also run other kinds of venues or compare against them, putting occupancy and yield across venues side by side is a good habit, so long as you define slots and revenue the same way for each.

For monthly tracking, the escape room owner community suggests watching booking conversion speed, rebooking rates, game master utilisation and cost per booked player. Rebooking rate is what the cohort table refines.

From pattern to test

A cohort table points at questions. It does not prove causes. Use findings to form a hypothesis and test it, because differences between cohorts alone do not establish causation. Cohort analysis can reveal differences in long-term behaviour and generate hypotheses for more rigorous testing later.

Here are examples of how a pattern becomes a test for a venue.

What the table shows Hypothesis A simple test
Cohorts from one channel return less That channel attracts one-off visitors Offer a second-visit incentive to that channel only, and compare the next two cohorts
Higher no-show share in a recent cohort Reminders or deposit rules changed Switch on SMS reminders for half of the week’s slots, then compare
Month 1 return jumps after a new room New content drives rebooking Announce new rooms to past players by email and measure return in the next 30 days
Prime-time cohorts cancel more than weekday-afternoon ones One policy fits poorly Test stricter deposits on peak slots only

The last row follows a sensible principle: friction should match demand, with firmer deposit terms on high-demand slots and a lighter touch on quiet weekday afternoons. Our post on escape room cancellation policy covers how to word those rules for customers, and the booking conversion audit helps check that the booking page itself is not the problem.

Change one thing at a time. If you alter the price, the reminders and the policy in the same month, the next cohort tells you nothing about which one worked.

Pitfalls that make cohort results misleading

Tiny cohorts. Tiny cohorts produce noisy percentages. If a cohort has 25 people, one extra returner moves the rate by four points. Combine months into quarters, or show counts beside percentages, before drawing conclusions.

Immature cohorts. Recent cohorts have not had time to fill later periods. Compare them to older cohorts only at the ages they have both reached.

Survivorship bias. If you include only customers who have stayed, newer cohorts can look inaccurately better than earlier groups that have already experienced natural churn. Always start the denominator from everyone in the cohort at month 0, not from those who came back.

A capped catalogue. Repeat rate has a natural ceiling in escape rooms, because a customer can finish everything you offer. A flat curve after a certain point may mean they have played every room, not that they are unhappy. Check by looking at how many rooms each returning customer has left to play.

Identity gaps. A customer who books with a work email once and a personal one next time looks like two new people. This understates repeat visits across every cohort, so improve identifier quality before you compare periods.

Pitfalls that make cohort results misleading

Seasonality. December cohorts, full of gift-driven and holiday visitors, will behave differently from February ones. Compare like with like by year when you have the history.

Our perspective

We built Riddlio around the idea that an escape room owner should not need a data team to answer a simple question: are the people who played with us last quarter coming back, and what should we do about it?

Cohort analysis only works when bookings, payments, customer messages and attendance live in one record per customer. When they are scattered across a calendar, a payment tool and a spreadsheet, owners spend their time reconciling data instead of reading it. That is why we keep bookings, availability, payments, communications and reporting together, and why our workflows are designed around player numbers, room availability and group bookings.

Our advice is to start small. Pick monthly first-booking cohorts, track one return event, and write down one change you will make when the numbers move. Reminders, deposits, voucher offers and pricing by time of day are all things you can adjust and then measure in the next cohort. We designed Riddlio so those changes and their effects sit close together.

Iosif Beraru

Put your booking data to work

Cohort analysis is only as good as the booking records under it. Riddlio is an online booking and management platform for escape room venues that brings bookings, payments, customer reminders and reporting into one place. Plans start at $29 per month for Starter, with Pro at $79 per month and Unlimited at $199 per month. Enterprise has no published price. See the plans and pricing to find the tier that fits your venue.

Riddlio

If you are comparing tools, our overview of escape room management software explains how bookings, scheduling and reporting fit together, and the online booking calendar shows how availability is handled for customers.

FAQ

What is the difference between cohort analysis and segmentation?

A cohort is a fixed group defined by a shared starting event, such as first booking in March, followed over time. A segment is a slice defined by an attribute, such as corporate groups, and its members can change. You can combine them by building cohorts within one segment.

How many customers do I need for cohort analysis to be useful?

There is no magic number, but small cohorts give noisy percentages. If a monthly cohort has only a few dozen people, group months into quarters or show the raw counts next to the percentages. Treat any difference of a point or two as noise until it shows up across several cohorts.

Which start event should an escape room use?

Use the first completed booking, meaning the group actually played. Reservations that were cancelled or no-showed should not place a customer in a cohort. Count unique customers by email or phone so that repeat visits are matched correctly.

What does cohort modelling mean?

It means grouping records by a shared starting event and following each fixed group separately over time. For a venue, that is the same as the monthly first-booking cohort table described above.

Sources

← Back to all posts