6 min read
Almost every bank I speak to has a digital banking strategy. Very few have one that a board could stress-test in an hour.
What usually exists is a transformation programme with a strategy label attached: a roadmap of platform replacements, a channel modernisation plan, a list of AI pilots, and a target operating model slide that nobody has costed. Each element may be sensible. Together they rarely answer the only question that matters to a board — how does this change the economics of the bank, and by when?
Here is the framework I use when helping boards and executive teams pressure-test a digital banking strategy. Five parts, in order. The order matters, because most strategies fail by starting at part four.
1. Decide what the bank is for
Relevance, not product, is what keeps a bank alive. Before any technology decision, the strategy has to state plainly which customers the bank intends to be indispensable to, and which it does not. A universal bank and a specialist lender need very different digital estates, and the difference should be visible in the capital plan, not only in the brand narrative.
The test: can each executive describe, in one sentence and without reference to a channel or a system, who the bank is choosing to serve better than anyone else?
2. Fix the economics before the interface
Digital strategies are still too often judged on front-end experience while the cost-to-serve behind it is untouched. The economics live in the processing layer — onboarding, affordability, underwriting, servicing, complaints, financial crime — where the hand-offs, rework and manual decisions sit.
The strategy should name the handful of end-to-end journeys that carry most of the cost, state the current unit cost of each, and state the target. If a strategy does not contain unit economics, it is a roadmap, not a strategy.
3. Treat AI as an operating-model decision
AI is now a board conversation rather than a technology conversation, and rightly so. The question is not which models to use but which work the institution intends to stop doing with people, which work it will do differently, and what governance holds the result together.
That places three obligations on the board: a clear view of model risk and accountability under existing supervisory expectations, a workforce plan that survives contact with the productivity assumptions in the business case, and a funding model that anticipates inference costs scaling with volume rather than falling to zero.
4. Rebuild the core in slices, not in programmes
Core replacement remains the single most reliable way to consume a decade of change capacity. The alternative that has worked repeatedly is progressive decomposition: lift the highest-value capabilities — payments, onboarding, pricing, data — out of the core one at a time, run them on modern rails, and let the legacy platform shrink toward a system of record.
Each slice must be independently valuable. If a step in the plan only pays back when a later step lands, it is a dependency, and dependencies are where multi-year transformation quietly dies.
5. Govern with a small number of honest metrics
Boards cannot govern digital strategy through programme RAG status. A workable set is small: unit cost per key journey, straight-through processing rate, time-to-decision, digital share of servicing, income per customer, and change capacity consumed versus released. Six numbers, reported the same way every quarter, will surface drift faster than any programme report.
What this changes
UK banking strategy is converging. AI, wealth, productivity and capital returns appear in nearly every set of results. When every institution is telling a similar story, the strategy document stops being the differentiator.
What differentiates is whether the board can see, in numbers, that the bank is becoming structurally cheaper to run and more relevant to the customers it has chosen. A digital banking strategy that cannot show that is a communications exercise. One that can is the most valuable document the board will read this year.