9 Ways a Telecoms Transformation Quietly Increases Churn
The programme hits every technical milestone. Cost to serve comes down. Then, months later, churn rises in cohorts nobody predicted — and the decisions that caused it were never called customer decisions.
- Transformation programmes routinely push churn up through frictions that never appear in a technical report. The cost only shows in cohort data, months after the programme has been declared a success.
- The nine below describe where the customer experience quietly degrades during BSS, OSS and platform work. Each is recoverable, but not by spending more on retention.
- The position worth holding is not “the programme is on track”. It is “we are measuring what each milestone does to churn, in the cohorts it touches”.
Telecoms measures the customer relationship in churn, and every point of it matters. Keeping a customer is worth far more than replacing one. Boards review it quarterly. Operators spend heavily on retention.
What boards see far less clearly is what their own transformation programme is doing to that number. A BSS modernisation delivers. The technical measures improve. Cost to serve comes down. Then, six to twelve months after the big milestones, churn rises in cohorts nobody expected — and by then it is genuinely hard to attribute to a programme that did everything it said it would.
The pattern is familiar across UK mid-market telecoms, in business connectivity, hosted voice and consumer providers alike. The programme is competent. The technology works. The customer experience changes in ways that push people towards the door, and the decisions that caused it sit in technical specifications that nobody framed as customer decisions.
The nine below describe how it happens. None of them are fixed by spending more on retention.
The new platform standardises away everything you had negotiated
A legacy billing stack accumulates years of arrangements. Custom configurations for the accounts that matter. Grandfathered tariffs. Billing dispensations. Bespoke service commitments. What looks like technical debt is, in part, years of negotiated trust written into a database.
When the new platform is configured, the default is standard. Every exception either gets rebuilt deliberately, which is expensive, or absorbed into a standard tier, which is not. Nobody funds the first across the whole base. So customers who had something special find they now have something ordinary.
None of this appears in a technical report and all of it appears in churn. The transformation team can defend the build. Nobody can defend the consequence, because nobody owned it. The fix is to make personalisation an explicit migration question — who had bespoke arrangements, what are they worth, what has to be preserved. Commercial questions, asked before the configuration, not after.
Fault resolution drifts back towards average
Customers know how long you normally take to fix things. Business customers in particular have built their own arrangements around your responsiveness, and that expectation is part of why they stay.
OSS work moves those times. New ticketing. New escalation. New suppliers. New routing. Each change is justifiable on its own. Together, resolution times drift — and an operator who used to be better than average slides towards it.
The drift rarely gets flagged, because the new system reports its own times and they look fine. What nobody reports is the comparison with what the same customers used to get. The fix is to make resolution time for affected segments a condition of sign-off rather than something to monitor afterwards.
Number porting goes wrong during the migration window
For consumer and small business customers, keeping the number is part of the relationship. The technical requirement is simple. The reality during a migration window often is not.
Migration produces edge cases. Numbers mid-port at the moment of cutover. Accounts split across old and new. Business customers with complicated multi-line arrangements. Each case is rare. There are a lot of rare cases, and every one of them is a high-emotion moment for the person it happens to.
When someone cannot move their number cleanly because of your internal migration, the technical explanation is usually real and completely irrelevant to them. This is the friction most likely to produce a churn decision quickly rather than slowly. The fix is to treat porting as the stress test of the migration rather than as an operational detail downstream of it.
Churn rarely starts with a decision the customer noticed. It starts with one they never saw being made.
Self-service replaces human contact for the wrong customers
The cost case is clean. Automation scales, people do not, and moving contact to digital takes real money out. Every operator pursues it, correctly.
What the cost case rarely measures is which customers valued the human relationship and how much loyalty was attached to it. Business customers with a named contact. Long-tenure subscribers who built trust with specific people. These tend to be the high-value ones, and they are the ones most affected when the channel closes.
The saving is real. So is the churn, and the second number is almost never set against the first. The board sees the cost reduction and does not see what it bought. The fix is to segment by customer value rather than by contact type: high-value customers keep the human option, and the savings come from the customers who genuinely prefer digital anyway.
Simplifying tariffs creates winners and losers, and only one group reacts
Old tariff structures get complicated. Hundreds of plans, grandfathered pricing, discounts, bundles. Finance sees complexity as cost, in billing, in support, in not being able to see the margin.
So it gets rationalised. Fewer plans, clearer logic, better margins, and everyone migrated to something comparable. Except comparable is doing a lot of work in that sentence.
Some customers come out ahead. Others lose a discount or a feature they were used to. The two groups behave very differently, and only one of them does anything about it. Rationalisation often looks worse once churn is netted against the margin gain. The fix is to model it cohort by cohort before migrating, and to accept that the group who lose out may need holding on legacy pricing rather than moving on principle.
Recognising the pattern is the first step. Knowing which cohorts it has already touched is the next.
A free Executive AI Opportunity Assessment. Forty-five minutes, online, with your leadership team. We arrive with a model of how firms in your sector make money and where AI has moved those numbers — and you tell us where it is wrong.
Book a free assessmentProvisioning takes longer during the transition, and customers remember
Legacy provisioning has years of refinement in it. Integrated with the field, with stock, with the messages that go to customers. Same-day or next-day on standard services is expected, and customers plan around it.
New systems, during the migration window, routinely take longer. The integrations are immature, the handoffs to the field are new, and the capacity to handle exceptions has not been built up yet. Customers wait longer, notice, and tell people.
The programme sees this as temporary. The customer files it as the service getting worse, and that impression does not expire when the migration does. The fix is to make provisioning time a gate rather than something to remediate: the new route has to match the old one on representative services before cutover, not after.
Happy enough. Rarely in touch. Trust building quietly, year after year.
Small things going wrong repeatedly. More contact than before. It is becoming effort.
Actively comparing. Using less. Trust going. Retention spend can still work here.
Decision made, move underway. Winning them back costs more than they were worth.
Transformation milestones move whole cohorts from settled to irritated. Almost all retention effort is aimed at stage three, which is late — and the cheapest intervention available is not spending more there, but not creating stage two in the first place.
For months, nobody can see the whole customer
During migration — which for a major programme runs into months or years — customer data sits partly in both systems. Support staff cannot see the full picture in either. Customers get partial answers, are asked to repeat things you should already know, and are told their account is in transition.
Each conversation is survivable. The accumulation is corrosive. Customers who go through months of this form a view about whether you are competent, and the view outlasts the migration by a long way.
The fix is to design the migration so that one system shows the complete picture throughout, even if it is the old one. Partial-in-both is the most expensive option and the most commonly chosen, because it gets treated as a data integration problem when it is a customer problem that happens to need a data solution.
Three things to check against your own data
- Look at churn by cohort over the last eighteen months. Has it risen among customers with bespoke arrangements, premium tiers or long tenure? If so, standardisation has cost you the people who were paying for not being standard.
- Compare provisioning and fault resolution times either side of each milestone. If they lengthened, even briefly, the experience got worse whatever the technology can theoretically do.
- Who owns the customer’s experience across the whole programme? If the answer is the programme manager or the customer experience team, it is nobody — neither has authority over the decisions that actually shape it.
Bundles come apart when their pieces migrate separately
Bundled services work — for you and for the customer — because billing, provisioning and account management are joined up. That integration is the product.
When migration moves the components at different times, the bundle fragments. Billing shows separate lines for a while. Commitments at bundle level get vague. Discounts get misapplied. Nobody on the phone can explain quickly why the bill looks different this month.
The customer reads that as you being unreliable commercially, and starts wondering what the bundle is actually for. The retention advantage the bundle created is exactly what erodes. The fix is to sequence migration around what the customer buys rather than around what is technically convenient — components that share a bundle move together, or with a deliberate interim arrangement.
Nobody owns the customer’s experience across the programme
The most structural of the nine. There is a technical architect. There is a programme manager. There are workstreams for billing, network, operations, service, marketing. Each makes sensible decisions inside its own boundary.
What does not exist, in most mid-market telecoms transformations, is anyone who owns what the customer experiences across all of it — how pricing, provisioning, support, channel and bundle decisions add up into one journey.
That absence is what the other eight describe. Every one of them is a locally sensible decision with a consequence nobody owned. The fix is to name someone at executive level with authority across the workstreams rather than inside one of them. It is not another workstream; it is a constraint on all of them. Most programmes do not have it, the frictions accumulate, and the churn data turns up long after the team has moved on.
What this means
These nine are not retention problems and they are not customer service problems. They are what happens when a transformation treats the customer’s experience as a downstream consequence rather than as the thing being designed.
Every one of them is a locally rational decision. None is wrong on its own. Together they produce an experience nobody chose, and the evidence arrives months after the programme team has dispersed — which is precisely why the lesson rarely gets learned.
The economics are stark in telecoms, because every point of churn is worth real money and winning someone back after they have gone is usually impossible. Designing the customer experience into the programme costs a fraction of what losing the cohort costs.
Which is the same argument as everywhere else, in a sector where it is unusually easy to measure. Change the work before you change the technology — and be clear about what the customer will experience before anyone signs off the configuration.
Is your programme protecting retention, or quietly eroding it?
Forty-five minutes with your leadership team, online, at no cost. We bring a working model of how firms in your sector make money and where AI has moved those numbers. You tell us where it is wrong. Then we tell you what we think — including “don’t invest”, if that is the honest answer.
Book a free assessment45 minutes · online · no cost · nothing to prepare
Continue reading
10 Reasons ERP Programmes Stall in the Mid-Market
They rarely fail on the software. The order things were decided in is usually the issue.
For chief executives9 Hidden Operating Frictions Reducing Customer Profitability
Profitability lost through fragmentation most firms have not yet noticed.
For retail11 Hidden Costs Quietly Undermining Retail Transformation
Where complexity grows faster than the discipline meant to contain it.