Customer retention,explained properly

Most businesses know how many new customers they won last month. Far fewer know how many of last year’s customers are still buying. That second number decides whether growth adds up or leaks away. A business that keeps 70% of its customers grows on top of a base. A business that keeps 20% has to replace most of its customers every year just to stand still.

This guide covers how to calculate retention correctly, what a realistic rate looks like, why customers actually leave, and twelve strategies that work, each with a way to measure it. The formulas are written out and the arithmetic is shown, so you can check every number against your own.

What is customer retention?

Customer retention is whether the customers you already have keep buying from you. It is a measure of behaviour over a fixed period, such as a month, a quarter or a year. New customers are left out on purpose, because retention asks only about the people you started with.

Four terms get mixed up constantly: retention, repeat purchase rate, churn and loyalty. They answer different questions, and using one when you mean another produces bad decisions.

Retention and churn are two sides of one number. If retention is 76%, churn is 24%. Repeat purchase rate is different. It looks across all customers ever, not at one period. It can look healthy for years while current retention falls.

Loyalty is the odd one out, because it describes preference, not behaviour. A customer can keep buying from you only because you are the nearest shop, or because your credit terms are easier. That is retention without loyalty. It lasts until a competitor opens next door or offers better terms.

How do you calculate retention rate?

Take the customers you had at the end of the period, subtract the new customers you won during it, divide by the customers you started with, and multiply by 100.

That mistake is the one to watch for. A business spending heavily on ads can grow its customer count every quarter while losing a quarter of its existing customers each time. The growth hides the leak. The formula removes new customers so the leak shows. If you want to check your own number, the free retention rate calculator does the arithmetic and flags the new-customer mistake.

What counts as a customer “still buying”?

You have to define it before you count. For a subscription, the answer is easy: they are still paying. For a store or a distributor, pick a rule such as “placed at least one order in the period” and use it every time. Change the rule and your numbers stop being comparable.

How period length changes the answer

The same business produces very different retention rates over different periods. Over one week, almost nobody reorders, so retention looks terrible. Over three years, most one-time buyers have gone, so retention looks terrible again for a different reason.

Match the period to how often your customers buy. If a typical customer orders every six weeks, a monthly period will call many active customers lost simply because they had no reason to order that month. A quarter fits them better. A grocery business might use a month. A furniture business might need a year.

Whatever you choose, keep it fixed. The value of a retention rate is in the trend, and a trend only exists if you measure the same way each time.

What is a good retention rate?

It depends almost entirely on what you sell and how often people need it. Ecommerce commonly sits around 25–30%. Subscription software, insurance and media often exceed 70%. The comparison that matters most is your own rate last quarter.

We have deliberately not printed a precise figure for each industry. Many per-industry retention tables circulating online cannot be traced to a named study with a stated method and date. Several are copied from each other, and their figures do not agree. A precise-looking number with no method behind it is worse than a range, because it invites you to compare against something that was never measured the way you measure.

The gap between a 28% ecommerce store and a 75% software company is not about effort. It comes from purchase frequency and switching costs. A software customer who leaves has to move their data and retrain a team. A shopper who leaves opens a different app. So compare yourself with businesses that sell something similar, at a similar price, bought about as often.

Better still, compare yourself with yourself. Measure this quarter. Measure next quarter the same way. If the number moves up after you change something, and a holdout group confirms it, that is the benchmark that matters.

Why does retention beat acquisition on cost?

Because you have already paid to acquire the customer. You already have permission to contact them, and they already know whether your product works. A second order does not have to cover the cost of finding the customer. The first order usually does.

The research is older than most people quoting it realise. In 1990, Frederick Reichheld and W. Earl Sasser published “Zero Defections: Quality Comes to Services” in the Harvard Business Review. They reported that cutting customer defections by 5% raised profits by 85% in one bank’s branch system, by 50% in an insurance brokerage and by 30% in an auto-service chain. The widely repeated line that a 5% rise in retention lifts profits by 25% to 95% comes from Reichheld’s later work at Bain & Company. Amy Gallo’s October 2014 HBR article, “The Value of Keeping the Right Customers”, attributes it to him. The same article states that acquiring a new customer is anywhere from five to 25 times more expensive than retaining an existing one.

Those figures describe the businesses studied, not yours. The multiple you often see quoted with no source — “five times cheaper” — is a rough average of very different cases. Work out your own. Here is how.

The numbers are illustrative, but the shape holds for most businesses. The message rate above is Meta’s India rate for marketing messages. The India WhatsApp rate card lists every category. Put in your own order value, margin and acquisition cost. If your repeat order costs less than your acquisition cost, and for almost every business it does, retention is where the next rupee of marketing should go.

One caution. Count only the orders your retention work caused. Customers who would have bought anyway do not belong in the “40”. That is why the example says “measured against a holdout”, and why the measurement section below matters.

What actually causes customers to leave?

Five things, and each needs a different fix. They had a bad experience. They had no reason to come back. They forgot you. They found it cheaper elsewhere. Or they no longer need what you sell. Diagnose before you prescribe.

Most retention advice skips this step and jumps to a discount. A discount fixes, at best, one of the five: the customer who found it cheaper. It does nothing for a customer who had a bad experience. That customer reads a “we miss you, here’s 15% off” message as proof that nobody read their complaint. It wastes margin on the customer who simply forgot, because a reminder alone would have brought them back. And it cannot help a customer whose need has gone.

“Forgot you exist” is the easiest cause to miss and the cheapest to fix. Customers do not decide to leave. Life gets busy, a different brand is in front of them when they need to reorder, and the habit breaks. A well-timed reminder solves more churn than any loyalty scheme.

To find out which cause applies to you, ask. Send lapsed customers one question: “What made you stop ordering?” Offer four or five options as reply buttons, plus a free-text choice. Replying takes one tap, so more people answer. Even a few dozen answers will tell you whether your problem is service, price or memory.

Then look at your own data for the signals in the middle column. Complaints and returns before a customer goes quiet point to experience. A single order followed by silence points to no reason to return. A steady rhythm that stops abruptly with no complaint usually means they forgot.

Twelve retention strategies that work

Each strategy below lists what it is, who it suits, how to put it in place and how to tell whether it is working. None of them suits every business. Pick the two or three that match your causes of churn from the section above, and measure them properly before you add more.

Several of these run best on a messaging channel your customers already check. In India that usually means WhatsApp. If you plan to run them there, read the WhatsApp Business API guide first. It explains the templates, the 24-hour window and the opt-in rules these strategies depend on.

How do you measure whether retention work is working?

Compare customers who received the retention work with a random group who did not, and count the difference in revenue. That difference is incremental revenue. Everything else is supporting detail.

Incremental revenue, not attributed revenue

Most tools report attributed revenue: any order placed within a few days of a message is credited to the message. It always looks good, because many of those customers would have ordered anyway. Incremental revenue is smaller and honest. It needs a holdout: a random share of eligible customers, often 10%, who receive nothing. The guide to agentic retention marketing works through a full holdout example with the arithmetic.

Cohort analysis

A cohort is a group of customers who made their first purchase in the same month. Track what share of each cohort buys again over time. If newer cohorts come back faster than older ones, your retention work is improving things. If they come back more slowly, something has got worse, even if total revenue is rising.

The metrics that matter

• Retention rate, measured the same way each period.
• Repeat purchase rate within a fixed window of the first order, by cohort.
• Time between orders. A shrinking gap is often the earliest sign that something is working.
• Incremental revenue and cost per incremental order, against a holdout.
• Customer lifetime value, by segment rather than as one average.
• Opt-out and block rate. This is the cost of your messages in goodwill.

Open rates, click rates and messages sent are useful for diagnosing a single campaign. They are not measures of retention. A campaign can be opened by everyone and change nothing.

What are the most common retention mistakes?

Five mistakes account for most wasted retention budgets. Each one feels sensible at the time.

• Discounting too early. A discount in the first message teaches customers to wait for one. Start with a reminder or something useful. Hold the discount back for customers who did not respond to anything else.
• Treating all customers the same. One message to your whole list spends the same effort on a customer worth ₹50,000 a year as on one worth ₹500. Segment first. Your top 20% of customers deserve attention, not coupons.
• Measuring engagement instead of revenue. A campaign with a 90% read rate that produces no extra orders has failed. Report incremental revenue first and engagement second.
• Fixed-day win-back rules. “Message everyone after 60 days” is late for customers who buy every fortnight and early for those who buy twice a year. Time win-back to each customer’s own gap between orders.
• Ignoring why they left. A customer who left after a damaged delivery needs an apology and a fix, not an offer. Find the cause before you choose the message.

All five come from the same habit: acting before diagnosing. Measure retention correctly, find out why customers leave, match the fix to the cause and check the result against a holdout. That order of work beats any single tactic in this guide.