Bookings are down this year. Not dramatically, but noticeably. You check your analytics: traffic is comparable, conversion rate is slightly lower, and campaigns are doing what they always have. You decide to increase the budget for new customer acquisition, because there clearly aren't enough of them.
What you don't see: the cause is from a year ago. And your most loyal customers had already left before you realised it.
That retaining customers is cheaper than acquiring new ones is basic knowledge. What most businesses don't know: which customers they're losing, when, and why. Without that knowledge you invest blindly in acquisition while leaking in retention.
What your report tells you, and what it doesn't
Standard analytics, whether GA4, your booking system, or a dashboard tool, shows you a snapshot. Sessions today. Conversions this month. Revenue this quarter. Useful for short-term adjustments, but it won't tell you why customers who came to you last year aren't coming back this year.
GA4 has a cohort report, but it's limited and rarely used correctly. And even the standard metrics you do pull from it deserve a critical look. What happens in practice: businesses look at the overall picture, see a decline, and look for the cause in the current year. They optimise campaigns, improve landing pages, test new channels. While the real cause emerged twelve months earlier.
Repeat share versus repeat rate: a difference with enormous consequences
I want to pause here, because this is the mistake I encounter most often. Even at businesses that think they're tracking retention well.
Repeat share is the proportion of your bookings or revenue that comes from returning customers today. If 35% of your bookings this year come from repeaters, that sounds like a healthy sign of loyalty.
But repeat share only tells you about the composition of your current bookings. Not about how loyal your customers actually are.
Suppose you acquired fewer new customers last year. The total drops, but the number of repeaters stays the same. Your repeat share rises, while your retention has actually stayed flat or worsened. The conclusion you draw: customers are becoming more loyal. The reality: you've simply attracted fewer new customers.
Repeat rate is the measure you actually want. Of all customers acquired in year X, how many ever came back? That's about the quality of an acquisition year. Not about the composition of your current revenue.
At a seasonal hospitality business I worked with, a cohort analysis revealed that the repeat rate of the most recent two acquisition years was markedly lower than that of earlier cohorts. Customers weren't less satisfied. The cause lay elsewhere, and you'll read why shortly. That pattern had remained invisible for years. Everyone was looking at repeat share. The conclusion was always: "things are going well, more than a third of our bookings come from returning customers." Meanwhile, newer cohorts were consistently less likely to come back.
"Repeat share tells you what today looks like. Repeat rate tells you whether your business will still exist in three years."
What a cohort analysis reveals
A cohort is a group of customers acquired in the same period, typically the same year or season. Track those groups separately in the years that follow, and you see something that disappears completely in aggregate numbers.
At the business I'm describing, the retention heatmap looked like this: customers acquired in 2018 and 2019 came back at a rate of around 25%. Customers acquired in 2022 and 2023 were at 18 to 19% after one year. Not dramatically lower, but consistent. And it wasn't coincidental.
| Cohort | Year +1 | Year +2 |
|---|---|---|
| 2018 | 25% | 15% |
| 2019 | 24% | 14% |
| 2020 | 22% | 12% |
| 2021 | 21% | — |
| 2022 | 19% | — |
| 2023 | 18% | — |
What's different about those more recent cohorts? Same marketing, comparable volumes, no major pricing changes. The cause wasn't in the campaign or the price. It was in something no one had looked at: timing.
The mechanism: your most loyal customers book earliest
This is what made everything fall into place.
Returning customers book much earlier than new customers. At this business, the median lead time for a first booking was around 70 days. For a second booking, already 110 days. For a fifth booking, more than 125 days.
Makes sense. Someone who has already stayed with you knows what they want. They don't need to compare or hesitate. When they're ready to book again, they do it. And that moment comes earlier in the year than you'd expect.
What had happened with the more recent cohorts: due to a strong focus on yield management, pricing for the following season had been made available too late. At the moment returning customers were ready to book, the prices weren't there yet. They showed up ready to book, but there was nothing to book. They didn't wait. They booked something else, or nothing at all.
That loss only shows up in your retention figures a year later. By then it's too late.
"Your most loyal customers rarely leave after a bad experience. They leave because they were ready to come back, and you weren't there."
The broader lesson: every business has a purchase cycle
Hospitality is a clear example here. The annual cycle is unmistakable in the data. But the same applies to any business with returning customers and a predictable purchase rhythm.
A B2B service provider whose clients renew annually. An accounting firm in spring. A garden centre that knows customers return every March. The core question is always: do you know when your customer is ready to buy again? And are you there at that moment?
In hospitality, the purchase cycle is exceptionally visible in booking data. In other sectors you have to reconstruct it yourself from purchase history. But once you know it, you know when you need to be available and visible.
Three questions every business should be able to answer
You don't need a complex dashboard system for this. Three questions are enough.
What is the repeat rate per acquisition year? Not today's repeat share, but the percentage of customers acquired in year X who ever came back. If that percentage has been declining in recent years, something deeper is going on.
What is the median lead time of returning customers? How early do they book? Does that lead time increase as the booking number gets higher? If your fifth-time booker decides on average four months in advance, you need to be available four months before the season.
When do returning customers reach your booking page? If you know that peak, you know when you cannot afford any obstacles. No missing pricing, no technical issues, no confusing availability calendar.
What else you can see when the data is there
The core of this story is the difference between repeat share and repeat rate, and the timing mechanism behind customer loss. But a full cohort analysis delivers a range of other insights.
At the business I'm describing, I was able to look at which type of first stay creates the most loyalty. Not every product attracts equally loyal customers, and that has consequences for how you structure marketing and pricing. I could see how the time between bookings evolves as a customer returns more often, and whether more recent cohorts show a different return rhythm. I could analyse churn by season: recurring, or tied to specific years? And for businesses with multiple locations or product lines, I could compare repeat rate by location to see where the strongest loyalty is created.
Each of those insights starts from the same question: of the customers you've acquired, how many came back, when, and why or why not.
In a follow-up article I'll go deeper into how you measure that purchase cycle precisely, which signals point to imminent churn, and how you turn that into action at the right moment.
If bookings are down this year, don't jump straight to acquisition or campaign performance. Start with the basic question: are bookings falling because there are fewer new customers, fewer returning customers, or both? Look not just at the proportion but at the absolute number of repeaters. If that number is falling while your total stays flat or grows, something fundamental is happening on the retention side.
Only then does it make sense to ask which acquisition year is underperforming in repeat rate, and what happened twelve months ago at the moment those customers were ready to come back. Sometimes it's pricing that wasn't available in time. Sometimes a season with unusually low satisfaction. Sometimes there's no clear cause, and it's simply the reality of an experiential product where most customers only come once.
Standard analytics won't give you those questions. A cohort analysis will.
The figures and patterns in this article are based on a real case. The company name and exact figures have been anonymised.
Joey Vangaeveren founded Intzicht and works as an embedded marketing and data analytics partner for B2B and B2C businesses across hospitality, business solutions, e-commerce and SaaS. His work spans strategy, custom analytics dashboards, and applied AI. He writes about what he sees in practice.
Curious what cohort analysis could mean for your business? Get in touch.