Silent Quitters: The customers who disappear without cancelling

Short answer

How do you detect customers who have left without cancelling?

By monitoring actual activity, not just contract status. Define what active use means, set inactivity thresholds per customer segment, and let a report flag the customers who fall below them. Every flagged customer must be investigated before being counted as lost – inactivity has many explanations, and only some of them are churn.

A customer has an ongoing agreement. There is no renewal date, no minimum term and no invoice that goes unpaid, because the customer pays per transaction. One day the transactions stop. Nobody cancels. In the CRM the customer is still active, and it is still included in the customer count.

Six months later, someone notices. By then it is too late to ask what went wrong, and too late to do anything about it.

This article is about those customers: how they differ from customers who cancel, why ordinary churn reporting does not catch them, and how Customer Success can build a fixed process for finding them. The approach is based on experience of working with customers on ongoing, transaction-based agreements. The experience has been generalised, and the examples are anonymised or hypothetical.

What is a Silent Quitter?#

A Silent Quitter is a customer who has, in practice, stopped using the service but still has an active agreement. The term is not an established professional term. We use it here as a practical label for what is often called silent churn: churn without cancellation.

The problem arises mainly where there is no natural decision date – SaaS without a minimum term, payment solutions, platforms with usage-based pricing and other ongoing customer relationships. With an annual agreement, the renewal forces a decision. Without it, an inactive customer can stay in the customer base indefinitely.

Five concepts are often mixed up. They should be kept apart, because they require different follow-up:

The breakdown is a professional judgement, not an established standard. A Silent Quitter is a customer with commercial churn but no formal churn.

The most important distinction is between the first two. An active contract says that the customer has the right to use the service. It says nothing about whether they do.

Why traditional churn reporting is not enough#

Churn reporting is usually based on events: a cancellation, an ended agreement, a subscription that is not renewed. Silent Quitters create no event. They disappear from the usage data, but not from the customer list.

The report shows 2 percent churn. Reality is closer to 8. That is not a rounding error but a different order of magnitude – and it affects everything built on the customer count: revenue per customer, share of active customers, capacity planning and LTV (lifetime value, also called CLV).

In transaction-based models there is a third measure: revenue churn, the revenue that has been lost. Here it is especially easy to draw the wrong conclusion. The revenue from the 60 customers has already gone from the accounts, but it has not been classified as churn. It looks like lower volume among existing customers. So the revenue report may show that something has happened, while the churn report says that all is well.

None of the measures is correct on its own. Together they show whether the customer has cancelled, has left, or is just using the service less. See How to measure churn in a fast-growing company for the distinction between customer churn and revenue churn.

How to identify Silent Quitters#

The method below has seven steps. The first four build the report. The last three are the manual work the report makes possible – and that is where the value lies.

Step 1Define what active use means#

Choose one event that shows the customer is getting value, not just that someone has logged in. In a payment solution, it is a completed transaction. In a SaaS product, it may be a completed task. The event should generate revenue or be closely tied to what the customer pays for. Logins and page views are too weak as signals.

Step 2Segment the customer base#

Group customers by size and normal activity level. A customer with hundreds of transactions a week and a customer with a handful a month cannot be assessed with the same deadline. See segmentation in the glossary.

Step 3Set thresholds per segment#

The threshold is how long a customer can go without activity before being flagged. It should be long enough to tolerate normal variation, and short enough that it is still possible to do something.

The thresholds are illustrative and not a recommendation. They must be calibrated against the company's own usage patterns – for example by looking at how long customers who later became active again were typically inactive.

The logic behind the table is that the threshold gets shorter the higher the customer's normal activity and value. For large customers, a drop in activity is often an earlier and better signal than a full stop.

Step 4Build an automated report#

The report should run on a fixed schedule, for example weekly or monthly, and show the customer, segment, date of last activity, number of days without activity and who owns the relationship. The most important thing is that it runs without anyone having to remember to ask for it.

Step 5Investigate every flagged customer#

A flagged customer is a hypothesis, not a conclusion. Before anyone contacts the customer, check whether there is a known explanation: open cases with Customer Support, technical errors, an ongoing implementation, agreed pauses or a seasonal pattern from previous years.

Step 6Contact the customer or the account owner#

The account owner often knows more than the data shows. Clarify internally first, and then contact the customer if the status is still unclear. The question should be open – what has changed? – not an attempt to save the agreement in the first sentence.

Step 7Record the outcome and set the next follow-up#

Every investigation ends in a status: active again, temporarily inactive with an expected date, legitimate low use, or confirmed churn. Confirmed churn is recorded with reason, date and revenue information. Every other status gets a new follow-up date, so that the customer is not flagged and forgotten again.

Inactivity does not necessarily mean churn#

It is tempting to let the report draw the conclusion: no activity for four months, therefore lost. That is the mistake that does the most damage. Four anonymised typical cases show why.

The seasonal business. A customer runs a business that is only active part of the year. In winter there are no transactions. Every year the customer is flagged, and every year the activity returns. The mistake here is to treat a known pattern as a new event. The solution is to compare with the same period last year, not with last month.

The customer who uses the solution as a backup. A customer uses another provider as their main solution but keeps the agreement as a fallback in case something fails. Activity is almost zero, but the agreement has real value to the customer. Recording it as churn is wrong. Ignoring it is also wrong: the customer is not very loyal and generates little revenue, and that should be recorded in the CRM.

The customer who has left but never cancelled. A customer has switched provider, closed the business or stopped the activity the solution was used for. Nobody has thought of cancelling, because the agreement costs nothing when it is not used. This is the true Silent Quitter, and this is where the recording has to be put in place.

The customer with technical problems. A customer has been inactive for several weeks because an integration has stopped working. The customer may not have noticed, or thinks the problem is on their own side. Here, inactivity is a symptom of a problem that can be solved – and this is the customer who is easiest to win back, if someone notices in time.

The same line in a report can hide a seasonal pause, a backup solution, a lost customer and a broken integration. Only one of them is churn.

Automatic classification produces errors in both directions. If seasonal customers and backup customers are recorded as lost, churn will be too high, and customers who come back are counted as new sales. If the technical problems are not discovered, you lose the customers who were easiest to keep. So the rule should be that the report flags, and a human concludes.

Separate last activity from the actual termination date#

Once churn has been confirmed, a new question arises: when did the customer leave?

There are at least three dates, and they mean different things:

Last activity is when the customer left in practice. The termination date is when the agreement ended. There can be many months between them.

If the dates are mixed up, the analysis goes wrong in predictable ways:

• Churn lands in the wrong period. A customer who stopped using the service in September but was recorded as terminated in March becomes part of the spring churn figures. The quarter looks worse than it was, and the autumn better.
• Customer lifetime becomes too long. Using the termination date gives the customer six extra months of lifetime without revenue. That inflates LTV calculations.
• Cohort analyses are skewed. Churn is dated too late, and the curves look better than they are.
• Previous reports change. If you instead backdate churn to the last activity, you change figures that have already been reported. That may be right, but it must be done deliberately and explained.

A tidy solution is to store all three dates and be explicit about which one is used in which report. Last activity is suited to analysing churn. The termination date is suited to contract and financial reporting. The gap between last activity and detection is a measure of how well the process works.

Build an operational workflow#

A report that nobody owns gets read a couple of times and then forgotten. The process needs the fields that make status visible, and clear roles for who does what.

What matters is the distinction between the fields, not their names.

From flag to recorded status

1. 01

   Report

   Flags customers above the threshold for their segment.

2. 02

   Customer Success

   Checks known explanations and clarifies with the account owner.

3. 03

   Sales or account owner

   Contacts the customer and confirms status.

4. 04

   Operations

   Records termination, date and reason in the CRM.

Customers who are not lost get a new follow-up date and come back into the report.

How the roles are divided depends on the organisation. In many companies it is natural for Customer Success to own the process as a whole: that the report runs, that the flags are investigated, and that every case gets a status. The account owner, whether in Sales or in Customer Success, confirms the status with the customer. Operations, or CS Operations where that function exists, makes sure that ended customer relationships are actually closed in the CRM.

The last step is easy to underestimate. When Customer Success has confirmed that a customer is lost but nobody closes the agreement in the system, the customer still appears as active in every other report. A fixed routine is needed for the handover – who is notified, in what form, and within what time.

Measure the effect#

The process should be measured at two levels: whether it is carried out, and whether it produces results. The first is easier to measure. The second is what matters.

The split between process and outcome is a professional judgement. Time to detection is both: it measures the process, but it also determines how much can be saved.

The share of verified Silent Quitters says something about the thresholds. If it is very low, too many customers with a good explanation are being flagged, and the thresholds are probably too short. If it is very high, the thresholds may be too long – the customers are already gone when they are flagged.

Revenue saved through reactivation must be interpreted with care. Some of the customers would have come back anyway, especially seasonal customers and customers with temporary problems. See The churn you never see for why such figures are always an estimate, and how the estimate can be improved.

The most important effect may not be in any of the figures above. Once the customer base has been cleaned up, every other measure becomes more reliable: churn, retention, NRR and revenue per customer.

The most important lesson#

Churn is not necessarily something that happens when the customer cancels. It may have happened long before anyone notices. The cancellation is just the point at which the company finds out.

Customer Success must therefore monitor actual customer behaviour, not just contract status. That does not require advanced tools. It requires a definition of active use, thresholds that suit the customers, a report that runs by itself – and someone responsible for looking at every single customer it flags.