Why do customers leave a business? Find your own reasons

Why do customers leave a business? A five-step check that finds your own reasons in last quarter's leavers, a win-back message to copy, and who not to chase.

Written by OpenRated Team
13 min read

The numbers say you're losing customers. Renewals are down, or the regulars have thinned out, or the agency's retainer list is shorter than it was in spring. What the numbers don't say is why. So you search "why do customers leave a business" and get a list of reasons that could apply to any company on earth: poor service, price, a better offer elsewhere. True, all of it, and useless until you know which one is yours.

This guide is about finding that out: a five-step check you can run on last quarter's leavers in an afternoon, a table of the common reasons with how each shows up in your own records, a way to read reviews as an exit log, a win-back message to copy, and a plain answer on who is worth writing to.

Why do customers leave a business? For reasons already in your records

People leave a company for a handful of reasons, and most guides agree on the headline ones: service that let them down, a product that stopped being worth the money, a competitor's offer, and silence from the business. Most list them and then tell you to monitor feedback; some name where to look (reviews, surveys, what the sales team hears). What they rarely do is sort your own leavers, in order, to find which reason is yours, and few give a table, a message to copy, or a word on which leavers are worth chasing.

We've also left out the figures those guides trade in. One of them dates a 71% customer-service statistic to 2009 and names no study where the figure appears. A number with no traceable source is not evidence, and you don't need it. The reasons are already in three places you have: the support inbox, the reviews people left you, and the billing or sales records that show when each customer slowed down before they stopped.

The five-step check: who left, what they shared, and one question

Run this once a quarter. The first pass takes an afternoon; after that it's an hour.

1. List who left in the last quarter. Define "left" for your kind of business and write it down, because the definition decides the list. For a subscription or software company it's a cancellation or a renewal that didn't happen. For a shop or a restaurant it's a customer whose usual gap between visits has passed twice with no visit. For an agency or a consultancy it's a contract that ended without a renewal conversation. Pull the list with the date each one went quiet and how long they'd been with you.

2. Sort the list by what the leavers had in common. Add a column for each thing you can check without asking anyone: how they found you, which plan or product they bought, whether they had a support ticket in their last month, whether they'd been hit by a price change, whether they'd left a review. Then count.

Here is an illustrative example, with invented numbers, of what the sort looks like for a small software company with 24 leavers in a quarter. Eleven came in through one discount campaign the previous autumn and left when the discounted year ended. Seven had a support ticket in their last month that took more than three days to get a first reply. Four were on the oldest plan when its price went up. Two moved away or closed down. That's four reasons, and one of them is almost half the list.

A horizontal bar chart of 24 invented leavers sorted into four groups: 11 from one discount campaign, 7 with a slow support reply in their last month, 4 on the oldest plan when the price rose, and 2 who moved away or closed down.
The sort is the whole method: whichever pattern outweighs the rest is the one to fix first.

Expect one pattern to come out larger than the others, and don't be surprised if it's one you wouldn't have guessed. The company in the example would have called its problem "support" and spent a quarter on reply times, when the bigger leak was a campaign that bought a year of customers who were never going to pay full price.

3. Ask three of them one question. Pick three people from the biggest group and ask, by email or by phone, "What was the moment you decided to stop?" One question, from a named person, no survey link. You'll get two replies if you're lucky, and a moment is specific in a way a score is not: "the second time an order arrived late", "when the price went up and nobody told me". A proper survey is for the customers still with you; build it from the questions that work in a customer satisfaction survey. For leavers, one question is the right size.

4. Check the places they could have complained. Search each name from the biggest group in your support inbox, your ticket system, your reviews, your social messages and your refund notes. Our working assumption, which this search tests, is that a customer who leaves over service told you first, once, and was missed. If the same names turn up beside an unanswered message, the reason has confirmed itself.

5. Decide the one fix, with a name and a date. Write one sentence: "We lose customers mostly because X, so by [date] we will Y, and [person] owns it." One fix, because two at once means you won't know which worked when next quarter's list is shorter. The other reasons wait for the quarter after.

The common reasons, and how each shows up in your own records

The groups below are the ones this guide sorts by; a leaver who fits none of them is a question for step 3. The middle column is what to look for in the sort; the right column is where to look first.

Reason What it looks like in your records First place to look
Service A ticket or message in the last month before leaving, a slow first reply, a reply that closed the ticket without fixing it The support inbox or ticket queue, searched by the leaver's name
Price Leavers cluster on the plan that changed price, or on a discount that ended; they left within a billing cycle of the change Billing records, filtered by plan and by the date of the last price change
Product Leavers cluster on a release date or a menu change; reviews mention a thing that used to work; usage dropped before the cancellation Reviews and ticket subjects in the weeks after each change
Competition Usage was steady, then stopped dead; the leaver's own reviews or social posts now name another supplier; no complaint on record The leaver's public profile, and your own reviews compared with a competitor's
Life change Moved away, closed down, changed job, outgrew what you sell; no pattern across the group The one-question ask in step 3, because nothing in your records will show it

Two of these are often confused. A leaver with no complaint on record and a sudden stop looks like a life change; our advice is to treat it as possible competition (someone found a better fit and didn't feel the need to tell you) until the one-question ask settles it, which it does in one reply.

How to read your reviews as an exit log

Reviews are the one complaint channel where unhappy customers wrote it down in public, so they're the richest part of step 4, and the part it's easiest to read for the star rating rather than the reasons. Read them this way instead:

  1. Open the one- to three-star reviews from the last two quarters, in date order, on every site where your business is listed. Skip the five-star ones for now; they say what to keep.
  2. For each one, note the stage it describes: joining, the first order, billing, a support exchange, or a change you made. Ignore the adjectives.
  3. Count the stage that repeats. Three reviews about billing in a quarter is a pattern; one review about everything is a bad day.
  4. Match the reviewers against your leaver list. A reviewer who is still a customer gave you a reason you can still fix for them; one who has left gave you the reason they left, in their own words.
  5. Reply to each, briefly and specifically, because the next customer deciding whether to join reads your replies too. How to respond to a customer complaint covers the shape of a reply that doesn't make it worse.

A quarter with no new reviews tells you nothing about why people leave, and a profile that is all five stars and two years old tells the next customer you've stopped asking. The reasons reviews matter to a business start with being found and trusted, but a steady flow of them is also the only exit log that writes itself.

For a business with more than a few hundred customers, the sort in step 2 is easier when the support history and the account history sit together. That is what a help desk does for tickets and what customer success software does for usage and renewal signals; neither is needed for the first pass, which a spreadsheet handles.

A win-back message you can copy, and the rules for it

Write to lapsed or lost customers only after step 5, because the message has to name what changed. A win-back note sent before the fix is a reminder of why they left. Here is the message, written for the software company in the example; swap the specifics for yours.

Subject: The slow replies you told us about: fixed

Hi [first name],

You stopped using [business] in [month], and the last message you sent us before that was about an order question that took four days to get an answer. You were right to be annoyed.

Since then we've changed how support works: every question now gets a reply from a person within one working day, and I check the queue myself each morning.

If you'd like to give us another go, reply to this email and I'll switch your account back on with your old settings in place. That's the only thing I'm asking.

No worries if not. Reply "stop" and you won't hear from us again.

[Name], [role], [business] [Postal address]

The rules it follows:

  • One ask. Reply, or click one link, or show the email in the shop. A message with three offers reads as a campaign; a message with one reads as a person.
  • No guilt. "We miss you" and "it's been a while" put the leaver on the back foot. Name what they told you, say they were right, say what changed.
  • A reason to return, which is the fix. If the thing that drove them out is still there, there's nothing to write yet, and a discount doesn't change that.
  • Honour the opt-out, and make it easy. One word in a reply is enough. A leaver who says stop and hears from you again has been lost twice.
  • Send it once, from a named person. Our advice, not a measured rule: one message, within 60 to 90 days of the exit, then leave it. A second message from the same address is the start of a drip campaign, which is the thing they walked away from.

For old customers who simply drifted, with no complaint on record, the message is shorter: what's new since they last bought, one way back in, and the same opt-out line. For a customer who left for a competitor, say what you now do that you didn't then, and nothing about the competitor.

The email rules, if you're in the United States

A win-back email is a commercial message, so in the US it falls under the CAN-SPAM Act. The Federal Trade Commission's compliance guide, in its own words, requires that your "From," "To," "Reply-To," and routing information be accurate; that "the subject line must accurately reflect the content of the message"; that you "disclose clearly and conspicuously that your message is an advertisement"; that the message "include your valid physical postal address"; that it carry "a clear and conspicuous explanation of how the recipient can opt out"; and that you "honor a recipient's opt-out request within 10 business days". It adds that "even if you hire another company to handle your email marketing, you can't contract away your legal responsibility". The same page, edited in January 2024 for inflation-adjusted penalties, puts the maximum at $53,088 per separate email in violation.

The template above covers three of those by design: a plain subject, a postal address and a one-word opt-out. How you make the advertisement disclosure in a one-to-one note is yours to settle against the FTC's guide. This is a summary of that published guidance, read on 2026-10-05, and not legal advice. If your business or your customers are outside the US, different rules apply, so check the ones that cover you before you send anything to someone who has stopped buying.

What not to do, and who you can't win back

Don't send the blanket discount. It's the obvious thing to reach for and it fixes nothing on the table above. The customer who left over a four-day support reply is not coming back for 20% off a service that still replies in four days. The one who left for a better fit uses the discount once and leaves again. And every customer who stayed learns that leaving earns a coupon. A discount is a reason to return only when price was the reason they left, which the sort tells you, and even then a fixed, honest price beats a one-off.

Don't write to everyone. The triage below is the last step before any message goes out.

Three cards sorting lapsed customers into groups: those who drifted away with no complaint, to be written to once; those who left over a fault you have since fixed, to be written to with the fix; and those who moved, closed or outgrew you, who are not chased.
Write to the first two groups, differently, and leave the third alone.

Some leavers are gone, and chasing them costs the time you should spend on the fix. A customer who moved away, closed their business or outgrew what you sell has no reason to return that you can supply. A customer who was never the right fit, like the eleven in the example who came in on a discount for a product they wouldn't have paid for, leaves again the moment the price is real; the fix there is the campaign. And a customer who left over a breach of trust, a rude exchange, a mishandled refund or a data slip, is sometimes recoverable, but only by a direct apology from the person responsible and proof that it was fixed, never by a marketing message. To win back customer trust you have to name the thing, and a win-back email that avoids naming it will not do it.

What a win-back message will never do is rescue a product people have stopped wanting. If the sort puts most of last quarter's leavers under "product", the next step is the product, and the message waits.

The one thing to do this week

Pull the list. Last quarter's leavers, with the date each went quiet, in a spreadsheet, and the columns from step 2 beside them. The sort takes an hour and tells you more about why customers leave your business than any list of reasons written for everyone. When it's done, write the one-sentence fix and give it a date.

Then make sure next quarter's list comes with a fuller exit log: ask the customers still with you for reviews, so that when someone is about to leave, there's a record of why. How to get more reviews is where to start.

Related Articles

Back to all articles