Perspective
Find Where Customers Decide to Leave, and Meet Them Before the Cancellation
Retention work tends to begin at the cancellation, where you see the problem most clearly and can do the least about it. We record the earlier signals and give you one view.
4-minute read5 November 2026
In Brief
- A customer who asks for a port code or calls to cancel has usually decided already. A late call, an offer or a plan recommendation can arrive too late, and can prompt the leaving it aimed to prevent.
- We see four actions: record what each customer has done, test where decisions form, define churn once, and target contact by whom it can help, since a risk score does not say who can be helped.
- Thinqmesh shapes your CRM around what customers have done, defines each figure once, and delivers your first report from one source, so that early signals reach the person who can act.
A retention programme at a mobile operator, a broadband provider or a streaming service tends to begin at the end. A customer asks for a port code, calls to cancel or lets a renewal lapse, and the machinery starts: a score, an outbound call, an offer. This is the point where the company can see the problem most clearly and do the least about it. We believe the decision to leave is usually formed much earlier.
The Count of Leavers Shows a Late Step
A regulator's count of porting requests shows scale, and it records only one late, visible step. A prepaid subscriber who stops recharging never files anything, and a household can move its data to another network over several months while keeping the number.
Late contact also carries risk. Research on wireless customers found that a plan recommendation sent to some of them was followed by more leaving, and that targeting by the likely effect of a contact does better than targeting by risk. We would record what customers do in the first weeks, define churn once, and give your managers one report built from one source.
Four Actions That Move Retention Earlier
Treat the following as hypotheses to test against your own records, because the mix differs by market and product.
Record What Each Customer Has Done
Stages that reflect customer actions show who is engaged and who is drifting.
The first weeks, early support contacts, installation visits and the first bill each leave a trace. Record them against one customer record, and a quiet account becomes visible months before it appears in a port request.
Get started
- List the first five things a new customer does in the first month.
- Check that each one is recorded against one customer.
Test Where the Decision Forms
Look at the first weeks: the plan chosen at a retailer, signal at home, an installation that took three visits.
Compare customers who left with those who stayed on these early events. Which events differ most is a question for your records, and the answer tells you where an earlier contact would matter.
Get started
- Pull the first sixty days of records for customers who left last year.
- Compare them with customers who stayed.
Define Churn Once
A port request, a lapsed recharge and a cancelled plan are different events.
Prepaid customers who stop recharging never file anything, and a household can move its data to another network while keeping the number. Agree one definition for each product, write it down, and report each separately.
Get started
- List every way a customer can leave a product.
- Write one definition for each and name an owner.
Target Contact by Who It Can Help
A risk score says who may leave, and does not say who can be persuaded to stay.
Proactive contact can backfire, so test an offer on a small group first. Measure the effect against a similar group that received nothing, and keep the contacts that change the outcome.
Get started
- Pick one outreach and a small test group.
- Compare their leaving rate with a similar group over three months.
How Thinqmesh Solves This
We shape the pipeline around what the customer has done and ask only for what a person knows at that point. Agreed definitions and one source then feed reports that your leadership can use.
- Week 1
Stages Built on Customer Actions
Stages record what the customer has done, and each field is required only when the next stage depends on it.
- Week 2
Entry That Stays Quick
Anything the system can work out is filled in for you, and entry can be spoken on leaving a meeting.
- Week 3
Your Records Moved, Reminders Set
Your data is migrated, each customer has one owner, and reminders keep follow-ups on time.
- Week 4
Two Reports in Use
Managers receive two reports, one on what moved this week and one on where the pipeline loses people, each built from an agreed definition.
You finish with a pipeline your team keeps current, one owner for every customer and two reports built on agreed definitions. Marketing emails, phone links and commission rules are sequenced as their own work.
