Reduce customer attrition with the seven rules of no-excuses CX

This is part 1 of a five-part series on no-excuses customer experience: seven rules for acting on what you already know about customers, beginning with the two that reduce customer attrition fastest. The other parts are part 2 on profit, part 3 on selling the idea inside your company, part 4 on measurement and part 5 on lifetime value.

Picture the quarterly review. Someone puts up the retention slide. The number is a little worse than last quarter, which was a little worse than the quarter before. There is a pause, and then the room does what rooms do: it explains. Everyone in it wants to reduce customer attrition, and everyone in it has a reason why the company cannot start this quarter.

The data is not clean. IT has a backlog until spring. There is no budget this year. Sales will not share their accounts. We cannot really measure retention anyway. And we tried a pilot once, and it fizzled. Customer attrition is the share of customers who stop buying from you over a period, whether or not they tell you they are leaving, and every sentence on that list is a reason for letting it continue.

Every one of those sentences is also true, or true enough. That is what makes them so effective. Nobody is lying. The company simply has a well-rehearsed set of reasons for not acting on what it already knows, and the reasons get better with practice. Seven rules answer them. This post takes the first two, because if attrition is the problem, they are where the money is.

Key takeaways

• Customer attrition has to be counted against the customers you started with, including the ones who left without saying so.
• The six standard excuses for not acting are all partly true, which is why they need a fixed set of answers rather than a fresh argument every quarter.
• Rule 1, start with the customers you already have, is an instruction about where effort points, and applying it costs nothing.
• Rule 2, act on what you already know, works because the signals of leaving already sit in your service desk, product analytics and point of sale.
• A one-page “what we already know” inventory replaces “we cannot” with things someone could do on Monday, each with a name next to it.

What customer attrition is and why the usual number flatters you

Attrition rate for a period is the number of customers lost during the period divided by the number you had at the start of it. Retention rate is the same fraction seen from the other side. That is the whole formula, and every argument about it starts with the words “lost” and “had”.

“Lost” is easy in a subscription business, where a customer cancels and the system records the date. In most other businesses nobody cancels. They stop coming, and the database keeps calling them a customer for years. A customer who bought once and never came back is a customer who left, whatever the CRM says. If your retention number does not count them, it is flattering you, and measuring retention correctly is the first job. The quiet leavers are also the subject of the post on customers who fire you and never say so.

“Had” matters just as much. Comparing active customers at year end with active customers at the start hides attrition behind acquisition: lose two thousand, gain two thousand, and the slide shows a flat line while the company pays the acquisition cost twice. Count attrition for a defined starting group, or it disappears into growth. Where “churn” and “attrition” are distinguished at all, churn usually means recorded cancellations and attrition the wider set. Count the wider set.

The excuses and the seven rules that answer them

I have come to think of the excuses as a fixed menu. Once you have heard them in three or four companies you can predict the order they arrive in, and each one is a disguised version of a smaller, more tractable problem.

Excuse	What it usually means	Rule that answers it
“The data is not clean”	The data is not joined	Rule 2
“IT has a backlog until spring”	We are waiting for a system to do a person’s job	Rule 2
“There is no budget this year”	The budget went to acquisition	Rule 1
“Sales will not share their accounts”	Nobody has sold the idea to sales	Rule 6
“We cannot really measure retention”	Nobody designed the measurement before launch	Rule 7
“We tried a pilot once and it fizzled”	No owner, no decision date, no comparison group	Rules 6 and 7

So it is worth having a fixed set of answers, too. These are the seven rules I use, and the series walks through them in order.

1. Start with the customers you already have.
2. Act on what you already know.
3. Talk to customers like a person: their channel, their timing, their words.
4. Make coming back easy: remove friction at the moment of return.
5. Give the frontline a reason and a way.
6. Sell it inside before you sell it outside.
7. Measure it, or it did not happen.

None of them requires a new system. Most of them require a decision.

The two rules that reduce customer attrition fastest

Rule 1. Start with the customers you already have

Acquisition has a scoreboard, a budget line and a team. Retention usually has a slide. That imbalance is the first thing to fix, and it costs nothing to fix it.

The customers you already have are the cheapest growth available to you. They have already found you, already trusted you once, already learned how your product or your store or your app works. You do not need to explain who you are or buy their attention from a platform. You need to keep them, and keeping a customer is almost always cheaper than replacing one, because replacement means paying the acquisition cost all over again.

“Start with the customers you already have” is less a slogan about loyalty than an instruction about where to point the effort. Before anyone designs a new campaign for people who have never heard of you, ask what would happen if the same effort went to the people who bought last year and have gone quiet.

Rule 2. Act on what you already know

Here is the part that surprises people who have not looked. Most companies already hold the signals of who is about to leave. They are not hidden, only spread across systems that do not talk to each other and owned by teams that do not meet.

Usage that was weekly and is now monthly. A complaint closed without a resolution note. A support ticket, then silence. An invoice queried and then paid late. A renewal notice opened four times and not acted on. A regular who used to come in on Saturdays and has not been seen since spring.

Every one of these is a customer telling you something without filling in a survey. The excuse “the data is not clean” usually means “the data is not joined”, which is a different and much smaller problem. You do not need a clean customer database to notice that a ticket was followed by nothing. You need someone whose job it is to look.

Acting does not have to be sophisticated either. A call. An email from a person with a name, asking whether the problem got sorted. A note to the account manager. The action matters less than the fact that something happens within days of the signal rather than never. A signal that is three months old is a post-mortem. A signal that is three days old is a conversation.

How to reduce customer attrition with a “what we already know” inventory

You can build this in a week, with a spreadsheet and a few conversations. Its purpose is to make the excuses harder to say out loud. Make five columns.

Signal	Where it lives	Who sees it today	What happens now	What could happen
Usage dropped by half over two months	Product analytics	Nobody, unless they run the report	Nothing	Account owner gets a weekly list
Complaint closed without resolution	Service desk	The agent who closed it	Nothing	Callback within three days
Support ticket followed by silence	Service desk	Nobody	Nothing	One check-in email from a person
Renewal notice opened, no action	Email platform	Marketing ops	Automated reminder	A call from someone who knows the account
Regular customer absent for a season	Point of sale or CRM	Store manager, informally	Nothing	A note, a small welcome back

Then work through it in order.

1. List the signals you already hold. Ask the service desk, the product team, the store managers and whoever runs billing what they notice before a customer leaves. One row each, in their words. Stop at ten.
2. Fill in “who sees it today” truthfully. This column is where the inventory earns its keep. The signal exists, but the person who could act on it has never been shown it. That is not a data problem. It is a plumbing problem, and plumbing is cheaper than you think.
3. Pick two rows, not ten. Choose the signals with the clearest link to leaving and the simplest possible action. A ticket followed by silence is a good first row, because the action (someone asks “did that get fixed?”) needs no system at all.
4. Write down what “acted on” means. Who does it, within how many days, and where they record that they did it. Without that, next year’s review has the same slide and one more excuse: we tried that, and we are not sure it worked.
5. Hold out a comparison group from day one. It is the difference between “attrition fell” and “attrition fell among the customers we called and not among the ones we did not”, and only the second sentence survives a budget meeting.

The inventory will not be complete and it does not have to be. Its job is to replace “we cannot” with a list of things you could do on Monday, each with a name next to it.

A worked example (illustrative)

The numbers here are round and invented, to show the arithmetic rather than a benchmark.

A company has 10,000 active customers and loses 2,000 a year. Its service desk closes about 1,500 tickets a year without a resolution note, and nobody has checked what happens to those customers afterwards. So check. Suppose half of them left within six months, against one in five for customers overall. That one comparison shows the signal is real, and it says that around 750 of the 2,000 annual losses walked out through a door you can stand next to.

Now the pilot: two people, three hours a week each, calling within three days of each closure, with one closed ticket in ten held out at random. After a quarter the question is one sentence long: are the called customers still buying at a higher rate than the held-out ones? Either answer is worth a few hundred hours.

What quietly breaks attrition work

The inventory is easy to build and easy to let rot.

The list gets longer instead of the actions getting done. Ten rows becomes thirty, each one interesting, none with an owner. Keep two rows live until each has a name, a deadline and a record of what happened.

“Acted on” turns into a script. The callback that started as a person asking “did that get fixed?” becomes a template with a survey link at the bottom. Customers can tell the difference, and the moment they can, the signal stops being answered and starts being processed.

Attrition is counted on cancellations only. The silent leavers never make the list, so the program congratulates itself on a number that excludes most of the problem.

The work is measured on activity. Calls made, emails sent. Those confirm the work is happening and say nothing about whether it worked, and that difference is the whole of rule 7.

When rules 1 and 2 are not enough

Two rules do not cover every cause of attrition.

When the product does not invite a relationship. Some purchases happen once a decade by nature. Signals of drifting mean little where there was never a rhythm to drift from, and whether your products suit relationship marketing at all is a prior question.

When the customers leaving are ones you would not keep at any price. A segment that costs more to serve than it brings in leaves, and the attrition number rises while profit rises with it. Before you chase the number, decide whether some customers should be let go on purpose rather than by neglect.

When the leak is at the moment of return. If customers try to come back and the renewal flow or the reorder page stops them, no callback will help. That is rule 4, and it is the subject of part 2.

Where to start

1. Recount attrition the honest way. Take the customers who were active at the start of last year and count how many bought anything in the twelve months since. Put that number next to the one on the slide.
2. Ask four people what they notice before a customer leaves. The service desk lead, a product analyst, a store or account manager, and whoever runs billing. Their answers are the first rows of the inventory.
3. Fill in “who sees it today” for every row. Be literal. A report nobody opens counts as nobody.
4. Pick two rows and name an owner for each. The owner makes sure something happens within days of the signal and records that it did.
5. Hold out one in ten and set a date. A quarter from now, someone compares the called customers with the held-out ones and says which group stayed, in one sentence.

FAQ

What is customer attrition?

Customer attrition is the share of customers who stop buying from a company over a given period. It includes customers who cancel and customers who simply stop coming back without saying anything. In most businesses the second group is larger than the first.

How do you calculate customer attrition rate?

Take the number of customers you had at the start of a period, count how many of them made no purchase or renewal by the end of it, and divide the second number by the first. Count against the starting group only, so that customers acquired during the period do not hide the losses. For businesses without contracts, define “lost” as no purchase within a window that fits the normal buying rhythm.

What are the early warning signs of customer attrition?

The common ones are a fall in usage or visit frequency, a complaint closed without a resolution, silence after a support ticket, a late or queried invoice, and a renewal notice opened but not acted on. Most companies already record all of these. The problem is rarely that the signal is missing and usually that nobody is assigned to look at it.

How do you reduce customer attrition without a new system?

List the signals you already hold, find out who sees each one today, and pick the two with the clearest link to leaving. Assign a named person to act within days of each signal, with a simple action such as a call or a personal email, and record what happened. Hold out a small random group so you can compare the customers you acted on with the ones you did not.