Hand an integrator the work, never the knowledge of how you charge
Operators that outsource their billing platforms often give away, without noticing, their own understanding of how customers are charged. The build, testing and running can sit with an integrator; the people who understand the rating rules and their history should stay on the operator's payroll.
An operator that outsources its billing platform to an integrator usually believes it has bought a service. In my experience it has often also sold something, without noticing: its own understanding of how it charges customers. My position is that a telecom operator can hand an integrator the build, the testing and much of the running of its business support systems, but should keep on its own payroll the people who understand its rating rules, its discount logic and the history behind each exception. When that knowledge leaves, the operator can no longer judge the integrator's work, price its changes or explain its own bills.
I spent twelve years at an integrator before I became a client, and I do not write this as a critic of integrators. They bring scale, method and people who have done a migration before. Most of the billing programmes I have seen succeed had a strong integrator in them. The problem lies in what clients choose to keep.
What we had handed over without noticing
When Tolvarra and Lonmara Mobile merged in 2023, both ran their business support systems through outsourcing agreements. Lonmara Mobile's was the more extensive. Over a decade it had moved first its development and testing, then its support, and finally its configuration of tariffs and discounts to its integrator, Halvecor Solutions, under a managed-service contract priced per change request. By the time of the merger Lonmara Mobile employed eleven people in billing and rating, and Halvecor's team on the account numbered about 180.
The arrangement looked efficient. Changes were delivered, the platform was stable and the service levels were met. The cost became visible only when we began planning the consolidation. To move mobile customers onto one platform, we needed to know exactly how every legacy tariff and discount behaved, including those created years earlier for single corporate customers. Our eleven people could describe perhaps a third of them. The rest were known to Halvecor engineers, some of whom had moved to other clients, and to documents that described what had been requested rather than what had been built.
That gap had two consequences. We could not challenge the integrator's estimates for the migration, because we did not know enough to say whether forty days was reasonable for a given change. And we could not explain, without asking Halvecor, why a particular customer's bill looked as it did. A managed service had become a dependency on knowledge we no longer held.
Which roles came back and which stayed out
In early 2024 I proposed to our executive committee that we bring the design, configuration and acceptance of rating and billing rules back in-house, while Halvecor kept platform engineering, regression testing and infrastructure operation. Over eighteen months we created 138 roles: configuration analysts, billing architects, revenue assurance engineers and a small team that owns the catalogue of tariffs and discounts. Fifty-two of the people we hired had worked on our account at Halvecor, and we agreed the transfer with Halvecor's management rather than recruiting around it.
The test for each role was simple to state and harder to apply: would we need this person to tell whether the integrator's answer was right? A billing architect passed that test. A regression tester, valuable as the role is, did not, because a well-defined test suite can be specified and checked from outside. We also left out of the plan the work where an integrator's scale is a genuine advantage, such as performance testing for large bill runs and the platform's own upgrade cycle.
Our finance director opposed the plan, and his reasons were sound. It added permanent headcount at a time when our owners expected merger costs to fall, and it replaced a price per change with salaries we would pay whether or not changes were requested. The operating partner at Ostrevan, our larger shareholder, preferred a renegotiated outsourcing contract with stronger knowledge-transfer obligations. We tested that option seriously. It would have been cheaper by about €2.2 million a year. It would also have left the knowledge on the integrator's side of the contract, documented but not employed.
The cost a fixed price hides
In its first full year the in-house team cost €3.1 million more than the equivalent change requests would have. The lead time for a new tariff, from approved design to customers being able to buy it, fell from 41 working days to 12. The integrator's estimates for migration work fell by about a fifth once our architects could question them line by line, which recovered much of the additional cost within the consolidation programme alone. And when a migration wave overcharged 7,900 households in November 2024, the people who traced the defect in the converged discount logic were our own analysts, two of whom had joined from Halvecor that summer.
There is a fair objection to all of this: that an operator should concentrate on customers and networks and buy everything else. For most enterprise software I agree. Billing is different in telecoms because the bill is where the product ends. It is where a tariff, a promotion, a regulatory obligation and a customer's history meet, and where the customer decides whether the operator can be trusted. The rules that produce it are commercial decisions written in software. An operator that cannot read its own rules has delegated a commercial function, not a technical one.
The approach has limits. A smaller operator may not be able to employ a full billing architecture team, and some will reasonably decide to rely on a vendor's standard product precisely so that there is less of their own logic to understand. That is a coherent choice, provided the operator has simplified its tariffs to match. The weak position is the common one in between: a complex, heavily customised catalogue whose meaning is held by people on someone else's payroll.
I would put one question to any chief executive or board reviewing a business support outsourcing contract. If the integrator left tomorrow, how many of your own employees could explain how a customer's bill was calculated? If the answer is a handful, the contract's price does not show its cost. Some of that cost will appear the next time you need to change, migrate or defend the way you charge.