Customer retention rate: how to measure and raise it

The quickest way to get your number

Here's the usual formula:

Retention rate = (customers at the end minus new customers) ÷ customers at the start × 100

Say you started the year with 50 clients, signed 12 new ones and finished with 54. That's (54 − 12) ÷ 50, or 84%.

There's an easier way. Pull up your customer list from the first day of the period. Cross off everyone who has gone. Divide what's left by what you started with.

Counting names also avoids a trap in the formula. Say one of those 12 new clients signs in March and leaves in October. Your end count falls to 53, so the formula says 82%. But you still kept the same 42 of your original 50. Counting names gets this right. It also leaves you with a list of who left, and you'll need that later.

Before you count, make two rules and write them at the top of the sheet:

• Who counts. Do small one-off jobs count? Free accounts? People who cancelled a free trial? Most small businesses should leave them out.
• When someone is gone. Is it the day they cancel, or the day their renewal passes? Pick one and stick to it. If you change the rule partway through the year, your number moves even though your customers didn't.

If nobody signs a contract with you

A shop, salon or online store has a harder job than a firm with contracts. Nobody emails you to cancel their relationship with your socks. They just stop coming, and you only find out later.

So you have to decide what "still a customer" means. That depends on how often people need you. A loyal laptop buyer won't be back within a year, so a one-year window tells you nothing. A book buyer who hasn't been back in a year is probably gone.

Here's how to do it:

1. Pick a time window that fits how often people buy. Look at customers who did come back and see how long most of them took. A salon might use three months. A clothing store might use 12.
2. Take everyone who first bought in one past stretch, say January to March last year. Choose a stretch old enough that all of them have had the full window to come back.
3. Count how many bought again inside the window.
4. Divide. If 400 people first bought in that stretch and 104 came back within 12 months, your rate is 26%.

Using first-time buyers keeps things fair, because every group gets the same amount of time. The usual formula can make a shop look better than it is. A customer from three years ago who wanders back gets counted as "kept," even though they weren't in your starting number.

On Shopify

Go to Analytics, then Reports. Click the Category filter and choose Customers. Open Customer cohort analysis. A cohort is just a group of people who first bought in the same month. Choose customer retention rate in the Metric menu. Then set the Visualizations menu to Retention curve, because you'll want that picture in a minute.

One thing to watch: each column in the table shows who bought in that one month. Someone who bought in months 4 and 7 appears in both. So adding across a row won't give you a 12-month number.

In a spreadsheet

Any system that exports your orders will do. Make a sheet called Orders. Put each customer's email in column A and the order date in column B. Check that the dates are real dates, not text. Then open a new sheet in Google Sheets or current Excel and enter these:

• A2 lists each customer once: =UNIQUE(Orders!A2:A5000)
• B2 finds their first order: =MINIFS(Orders!B:B,Orders!A:A,A2)
• C2 checks whether they came back within a year: =COUNTIFS(Orders!A:A,A2,Orders!B:B,">"&B2,Orders!B:B,"<="&B2+365)>0

Copy B2 and C2 down the sheet. Change 365 to match your window. Filter column B to the stretch you chose. Then count how many rows say TRUE, and divide by the total number of rows.

Four numbers that all get called retention

Before you compare yourself with anybody, check which one you're looking at.

People also work these out in different ways. So when a tool or a salesperson quotes you a retention number, ask what was divided by what.

Money matters most when your clients vary in size. You can keep eight of ten clients and still lose a third of your income, if the two who left were your biggest.

Is your number any good?

First check the time period. A monthly number looks far better than it is:

• Keep 98% a month and you keep about 78% a year
• Keep 95% a month and you keep about 54% a year
• Keep 90% a month and you keep about 28% a year

Now compare your yearly number with this table. It shows roughly how many customers out of every 100 each kind of business keeps. The sources don't say what time period they cover, so treat these as rough yearly numbers.

Where you see a range, two sources gave different answers. One of them is CustomerGauge, which looks at companies that sell to other businesses. So treat every row as a rough guide. If you run a local service, such as a salon or a garage, there's no good number to compare with. Use your own history instead.

Your own trend beats all of these. For a good target, average your last four quarters and add two or three points.

If you're small, don't read much into the percentage. With 12 clients, losing just one drops you by more than 8 points. Below about 30 customers, skip the math and look at the names.

What a few points are worth in your business

You may have heard that keeping 5% more customers raises profit by 25% to 95%. That range goes back to Fred Reichheld of the consulting firm Bain, and his example was financial services. Your business is different, so work out your own number. It takes two minutes:

Customers at the start × extra share you keep × yearly profit per customer

Say you have 400 customers and each brings in $300 of profit a year. Keep five more out of every 100 and that's 20 more customers. That's $6,000 more profit, every year. Now ask what it would cost to win 20 new customers. The answer tells you where your next dollar should go.

It also adds up over time. Here's what's left of 100 customers after three years:

Some businesses should care less about this. A mattress brand, a roofer and a wedding photographer all fall in this group. If people only need you once a decade, your rate will always look awful. Watch referrals and reviews instead. A little cross selling, like pillows to go with the mattress, may be the only repeat business on offer.

Find where people leave before you fix anything

Look at a chart of one group of customers over time. Use Shopify's retention curve, or make a line chart in your spreadsheet of how many in the group are still buying each month. The shape tells you where to start.

Online stores should expect a big drop after the first order. So don't panic at the shape. Compare one group with another.

Next, sort customers by where they came from. Then move your marketing money toward the campaigns that bring people who stay. Shopify's first-order filters let you sort by marketing channel, sales channel and product. Your cheapest ad may turn out to bring people who never come back.

Now ask them. Take the last ten customers who left and email them yourself, from your own address:

Hi [name], you were a customer of ours until [month] and I noticed you haven't been back. I'm not trying to sell you anything. Would you tell me the main reason? One line is plenty. Thanks, [your name]

Sort the replies into piles. "I don't need it anymore" is usually a customer you can't save, so spend nothing there. "I forgot about you" is the cheapest problem you'll ever have. "Something went wrong" and "I found something better" are the ones to fix this month.

How to raise it, easiest fixes first

Stop losing people by accident

If you bill on a schedule, some of the people you lost never chose to leave. Their card simply expired. If you use Stripe, it can try failed payments again by itself. Look under Billing, then Revenue recovery, then Retries. New Stripe accounts get up to 8 tries over 2 weeks by default. Some cards can't be saved this way and need replacing, so also send a plain email asking people to update theirs. In another billing tool, look for a setting called failed payments or dunning, which is the billing word for chasing them.

Fix the first 30 days

Tell people what to expect, because surprises drive them away. Send one message right after the sale. Say what happens next, when it happens and who to contact. For a product, it might read:

Your order arrives by Thursday. Most people notice a difference after about two weeks of daily use, so don't judge it on day three. If anything's off, reply to this email and I'll answer it myself.

Put your own true details in there. For a service, use the same idea. Say what you'll do in the first month, what you need from them and the date you'll review it together.

Reach people before they drift

Work out when your good customers normally come back, then show up just before that. For a shop, the time between the first order and the second tells you more than the percentage does. Online stores win or lose on repeat customers at exactly this moment.

Shopify's One-time customers report gives you the list. It shows each person's first order date and whether they agreed to marketing. Email only the ones who did. Most of this work is retention marketing, and you can start with one plain reminder.

Make the message useful. A how-to for what they bought, a refill reminder or a new way to use it gives people a reason to read.

Bring evidence to the renewal

Clients leave when they can't remember what they paid for. Sixty days before renewal, send a short summary with three parts: what you delivered, how the spending compares with the plan, and one idea for next quarter. If a client's budget year starts in July, get next year's proposal to them in May. If you're raising prices, say so here, well before the invoice.

Handle the bad moment fast

Things go wrong, and what you do next matters a lot. In one consumer survey by Klaviyo, half of people said they'd give a brand another chance if it offered a refund, replacement or discount. One in five stop buying after a bad experience. So give one person the job of answering complaints the same day.

Sell to people who'll stay

The steep drop after a first purchase often starts before the sale. Someone bought the wrong thing, or believed the wrong promise. The investment firm Vanguard once turned away a $40 million investor because it expected them to leave soon. You probably don't need to go that far, but you can ask better questions up front.

That's what we built Possible to do, so bear in mind this is our own product. Possible is an AI built on real sales psychology. It asks each visitor the right question at the right moment, remembers the answers and guides them to a sale, sign-up or booking. It's free to start, and here's more on our AI sales assistant for online stores. It only belongs on this list if your chart drops early and customers tell you it wasn't what they expected. If they're leaving over late deliveries, fix the deliveries.

Leave loyalty programs until last

Points schemes are popular, but they tend to reward people who were coming back anyway. So one won't help much until you know why the others left. When you get there, keep it small. For clients, you could lock their rate for 12 months if they renew for a full year. Once two or three of these moves are working, you can turn them into a proper customer retention program.

Check the fix really worked

Some customers would have stayed without your email. If you only count who came back, you'll give yourself credit for them. You'll also pay for discounts you never needed to offer.

So hold some people back. Add a column to your list with =RAND(), which gives each person a random number. Sort by that column and contact only the top half. Then compare the two groups 90 days later.

Say 500 customers have stopped buying, and you email 250 of them 15% off. Forty of those come back. Of the other 250, who got nothing, 34 come back anyway. So your email really brought back about six people, and you gave a discount to 40. With numbers this small the gap could be luck, so run it a couple of times before you believe it. Even so, it tells you more than counting who came back.

One hour this week

1. Pull your customer list from a year ago and count who's left. For a shop, use the first-time buyer method above.
2. If you've been reading a monthly number, turn it into a yearly one.
3. Look at the chart, and sort it by where customers came from.
4. Email the last ten customers who left.
5. If you bill on a schedule, check your failed payment settings.

Do that once a quarter with the same rules. Then you'll have a number that means something.

Questions about customer retention rate