Reimagining the telecom industry: simplification

While our telecoms business is simple, the way how we deliver it, and how we operate it, is far from simple. We have let complexity grow for a long time and enabled everything built on top of it, all the way to AI. Nothing works without digital infrastructure. The heritage accumulates though, and consolidation compounds history on top of history: hardware, software, data, and people carrying good old knowledge alongside a certain reluctance toward disruption. Some of that caution is correct, because a critical infrastructure provider cannot take large risks, and a network failing is a public safety matter rather than an inconvenience. The result is still stagnation, while the industries that depend on our services have grown tremendously. The enabler does not capture the value it creates. And without value creation there is no capacity to invest in the next generation, which is the conundrum we have to get through.

There is no perfect end state worth defining, because whatever I define today is legacy tomorrow. The only target that survives is an organisation that continuously learns, tests and improves, which takes both people skills and the unglamorous processes that make improvement routine.

Transformation runs from within. The alternative, building a clean new operation beside the old one and migrating customers across, fails in a technology-driven business, because the organisation treats the new thing as an antibody and resists it. At Telia we engaged the whole company, twenty thousand people over four years, rather than putting a few clever people to work at the sideline.

The product portfolio is where I would start, and our own start was humbling. Alison, our CEO and I stood on stage in 2020 and said we would remove half our legacy products within three years. We did not know exactly how many products we had at the time, which is scary yet unfortunately a pattern in our industry: catalogues multiply, systems disagree, and segments hold their own. The audit found north of a thousand products in the systems while we sold a handful. Then comes the hard part. Stop selling the legacy. Migrate the customers you can. Accept that some have no target to move to, that it costs money, that it may cost customers, and that the sales people will not like it. The business case is genuinely difficult, because complexity cost resists quantification, so it takes executive conviction ahead of proof. It paid off. Products are now well below half, satisfaction is up, and cost is coming down.

Culture decides whether that sticks. Launching things is exciting; hiring is exciting; writing new requirements is exciting. Challenging old stuff is much less sexy, and there is far less natural drive behind it. So cleaning out the cupboard has to become something people take pride in, which means measuring it: count the products, count the platforms, track profitability, lift those KPIs where everyone can see them, and report the numbers externally quarter by quarter. The same bias shows up in customer experience, where attention flows to major releases while the pain sits in signing up for fibre or changing a tariff. Part of IT capacity therefore belongs to continuous improvement rather than to launches.

Partners deserve more than a purchase order. Treating an ecosystem partner as a vendor means writing a requirement, picking the cheapest bid that meets it exactly, and waiting. I want partners with a joint incentive, who will tell me when my requirement is wrong. We cut a sprawling partner landscape down to a few and deepened those relationships instead. It extends to teams: there is no difference between someone on our payroll and someone on a partner contract, and the relationships that last come out of forty-eight hours locked in a room during a migration with every rollback ready, finishing it together.

Execution needs a clear direction and a tolerance for the plan failing anyway. We were explicit about which products, processes, systems and partners we were betting on. Then you test for months, prepare plans A through F and G, and something unexpected still happens. What carries it is a team resilient enough to work from the information actually in front of them.

The interface between humans and machines is mega narrow band. Speech is slow, vision is slow, typing is slow. Widening it is the problem Neuralink is attacking, and I have worked at that boundary myself, coupling neurons onto a silicon dioxide field-effect transistor to measure an action potential. For our industry, growth has come from communication between people. The next stage is communication between everything, living and non-living, which needs low latency, security and bandwidth. We will only invest in that if we unlock the value first, which brings it back to simplification.