Let me start with the uncomfortable truth I share with every leadership team that asks me to "run a digital transformation." Most of these programmes do not fail because the technology was wrong. They fail because they started with the technology at all. Someone bought a platform, announced a transformation, and then went looking for problems the platform could solve. I have spent more than a decade watching that sequence play out, and it ends the same way almost every time: a large invoice, a frustrated team, and a business that works roughly the way it did before.
So here is the roadmap I actually use, stripped of the buzzwords. It is not glamorous, but it is the version that survives contact with reality.
Start with the outcome, not the technology
Before I let anyone talk about AI, cloud or data platforms, I make the team finish one sentence: "In twelve months, this is a success if ______." Not "we migrated to the cloud." Something a CFO would recognise — faster quote-to-cash, lower cost-to-serve, a churn number that moved, a product shipped in weeks instead of quarters. If you cannot name the outcome, you are not ready to buy anything yet.
The reason this matters is that the outcome dictates the entire sequence. A company trying to cut cost-to-serve should not start the same way as one trying to launch a new digital product. When leaders skip this step, every downstream decision becomes a matter of taste, and taste is expensive.
Map the current state honestly
The second step is the one people want to rush, and it is the one that saves the most money. Before designing the future, map what you actually have: the systems, the data, the manual workarounds, and the places where work quietly breaks. In almost every business I have worked with, the real bottleneck was not a missing tool. It was three systems that did not talk to each other and a spreadsheet holding the whole thing together.
Be specific about your data, because data is where transformations live or die. Where does it sit, who owns it, is it trustworthy, and can you actually get to it? A brilliant AI strategy on top of fragmented, untrusted data is just an expensive way to automate your mistakes.
Transformation is not about adding new technology on top of a broken process. It is about fixing the process, then choosing the smallest technology that makes the fixed version repeatable.
Sequence in three layers: data, then cloud, then intelligence
When people ask me where AI fits, I tell them it fits third. There is a natural order that keeps you from building on sand.
- Data foundation first. Get the important data into one trustworthy place with clear ownership. Unglamorous, essential, and the thing that makes everything after it cheaper.
- Cloud and platform second. Move the workloads that benefit from elasticity and give your teams a modern place to build. Cloud is a means to speed and scale, not a trophy.
- Intelligence and automation third. Now — and only now — apply AI and automation to the specific outcome you named in step one. On a solid data and platform base, this is where the visible returns finally show up.
Trying to run these in parallel, or in reverse, is the single most common mistake I see. AI on bad data disappoints. Cloud without a data plan just moves the mess somewhere more expensive.
Ship in thin slices, prove value, then widen
I do not believe in eighteen-month transformation programmes that reveal their results only at the end. I believe in thin slices: pick one workflow, one team, one measurable outcome, and ship a working improvement in weeks. A thin slice does two things a big-bang programme cannot. It gives you real evidence instead of a forecast, and it builds the internal belief you will need to fund the next slice.
Once a slice proves its number, widen it — same pattern, next team, next workflow. This is how a transformation compounds. Each slice pays for the confidence and the budget of the one after it, and the business is improving the entire time rather than waiting for a reveal.
Treat people and process as part of the system
The part every technology vendor underplays: a transformation is mostly a change in how people work. The best platform in the world fails if the team routes around it. So I plan for adoption from day one — who is affected, what changes for them, what training and support they need, and how we will know they have actually adopted it rather than tolerated it.
This is also where governance belongs. As you introduce AI and automation, decide early who is accountable for outcomes, how you keep humans in the loop where judgement matters, and how you measure whether the system is behaving. Governance added at the end is a bolt-on. Governance designed in from the start is just good engineering.
What I would do if I were starting on Monday
If you take one thing from this, take the sequence. Name the outcome a CFO would recognise. Map what you have, honestly, especially the data. Fix the process before you buy the tool. Then sequence data, cloud and intelligence in that order, shipping thin slices that each prove a number before you widen them. And treat the people who have to live with the change as the most important part of the system, because they are.
Digital transformation in 2026 is not a technology purchase. It is a discipline: the discipline of turning a real business outcome into the smallest, most repeatable set of changes that deliver it — and having the patience to let that compound.
