Most digital transformation efforts do not fail on the technology. They fail on everything around it: data that lives in six systems and agrees with itself in none, a process nobody wants to change because the current one protects someone’s job, and a budget that funds the project once and then treats it as finished. Starting well means being honest about that from the first week, because the tools are the easy part.
What transformation means once you strip the brochure off it
At its core this is putting digital technology between your organisation and the work it does, so operations run faster and cheaper, customers get a smoother experience, and, once the data is finally usable, you can reach revenue you could not reach before. That is the payoff. It is hard because those three goals pull on different parts of the company and rarely arrive together. Efficiency is an operations project, experience is a product and design project, and new revenue is a strategy bet, so a programme that promises all three at once tends to deliver none of them.
The order that actually works
Teams that get somewhere start narrow and sequence on purpose. First they fix the data foundation, because analytics and AI are only as good as the plumbing beneath them, and a clear set of metrics to steer by is worthless if the underlying numbers disagree across systems. Then they automate one painful, high-volume process end to end, prove the saving, and use that win to fund the next step. Only after a couple of those do they attempt the customer-facing redesign, which is the most visible piece and the one most likely to stall if the back office is still manual. Big-bang programmes that rebuild everything at once are usually the ones quietly cancelled in year two.
Two failure modes account for most of the stalls. The first is buying a platform before fixing the process, so the new tool automates a broken workflow and locks it in place, now it is expensive to change and still wrong. The second is treating integration as a detail: the new front end demos beautifully, but it has to read from and write to a core system built decades ago, and that seam is where the months disappear. Budget for the integration and the data cleanup as the main work rather than the rounding error, and the rest of the plan starts to look realistic.
Why now is a fair question, and the honest answer
The reason to move is not that the technology is new. Cloud, analytics, and machine learning are mature and commoditised now, and that is precisely the point: the barrier to entry has dropped, so a laggard has fewer excuses and a shrinking head start to defend. The pandemic compressed a decade of change in customer behaviour into a couple of years, and the bar is now set by the best digital experience a customer has had anywhere, not by the norms of your industry. The cost of waiting is not a missed trend, it compounds: every year on manual processes and disconnected data is a year competitors spend learning from theirs.
What the Portuguese cases actually teach
The Portuguese examples are useful less as trophies than as patterns. EDP’s move into renewables and smart grids works because the digital layer serves a measurable operational goal in generation and distribution, rather than digitisation for its own sake. Farfetch treats the platform as the product, so personalisation and a clean checkout are core engineering and not a marketing veneer. NOS and Millennium BCP both went at the customer channel, the app, online banking, self-service, where the payoff is visible and the volume justifies the build. Fidelidade automated claims and stood up self-service portals, unglamorous back-office work that strips out cost and friction at the same time. Sonae digitised the supply chain and went omnichannel, which only pays because the logistics underneath were rebuilt to match. The common thread is not ambition, it is that each tied the technology to a specific operational result.
Starting without stalling
If you are at the beginning, resist the roadmap that lists every technology you intend to adopt. Pick one process where the pain is measurable, fix the data feeding it, automate it, and bank a number you can put in front of a board. Change management is not a workstream you bolt on at the end, it is most of the job, so bring the people whose work changes into the design early or watch them route around the new system the moment you look away. We help organisations in Portugal and beyond do this the sequenced way: strategy work that produces a real first step instead of a wish list, process automation that pays for itself, the data and analytics groundwork underneath it, and the change management that decides whether any of it sticks. The organisations that transform are rarely the ones with the biggest plan, they are the ones that shipped a small, real change and kept going.