Building a digital function that improves itself

Most digital teams plateau. Not because the people are wrong. Because the function itself does not learn.

A self-improving digital function is rare. It is also a recognisable thing once you see one. The work changes more than the people. New analysts join and absorb the standard quickly. Senior leaders reinvest in the parts of the operation that will compound. The team gets quietly better even when the obvious problems have been solved.

Five elements separate a self-improving function from a busy one.

A capability framework that is actually used

Not the document the L&D team produces and nobody opens. A working framework names the capabilities a particular role needs at three or four levels of seniority, and is used in hiring, in performance reviews and in promotion conversations. The framework is the spine of how the function grows. We deliver this kind of framework as part of team build engagements, because without it the rest of the work has nowhere to land.

Without it, capability decisions get made on instinct, and instinct disagrees with itself across the leadership team.

Hiring gates that protect the standard

A self-improving function hires slowly and selectively. There is a clear bar, the bar does not move under pressure, and senior leaders feel personally accountable when an offer is made. Each hire raises the average. Each below-average hire is felt for years.

The single best test is whether senior leaders would be willing to lose a candidate they like rather than lower the bar. If the answer in practice is no, the function is not protecting its standard.

Hiring patterns that compound

Beyond the bar, three hiring patterns separate functions that compound from functions that just grow.

Hire one level above where you think you need to. The role you are filling today is rarely the role this person will need to do in eighteen months. The function will change underneath them. Hiring one level above the brief is how you get someone who is still a strong fit when the brief moves.

Optimise for slope, not intercept. Polish at interview is overrated. Trajectory is underrated. The candidate who has visibly grown across the last three roles will almost always outperform the candidate who has plateaued at a more impressive title. Slope wins over time.

Hire for the discipline you currently do worst. It is tempting to hire more of the people you already know how to manage. The function compounds when you hire the discipline you are weakest in, even though it costs you more management attention in the first six months. Strength tends to look after itself. Weakness only fixes through deliberate hires.

These patterns matter most when a team is scaling fast, which is when they are also hardest to hold to. A clear capability framework and a senior reviewer on every offer are the cheapest way to keep them honest.

A leadership cadence that talks about the team

Most leadership reviews talk about the work. Self-improving functions have a separate, recurring conversation about the team itself. Who is at risk of underperforming. Who is ready for stretch. What capability is missing. What the next two hires should look like. This conversation runs monthly at minimum and is treated with the same seriousness as performance review.

Without it, talent decisions get made reactively, usually too late.

The leadership cadence template

For senior leaders building this rhythm for the first time, a simple monthly template works.

1. People at risk. Who has been underperforming, struggling or quietly disengaging? What is the plan for each, and who owns it?
2. People ready for stretch. Who is ready for a bigger remit, a different scope or a step up in seniority? What is the path?
3. Capability gaps. Where is the function thinnest right now? Is the gap a hiring problem, a development problem or an operating-model problem?
4. Next two hires. What roles are coming, what should the brief look like, and who is sponsoring each?
5. Promotions and exits. Decisions made, decisions parked, decisions overdue.

Sixty minutes, monthly, no slides. The single most useful hour a senior leadership team will spend each month, and the one most likely to be cancelled when calendars get tight.

Learning loops in the daily work

The team writes things down. Decisions, optimisations, what worked, what did not. Not as bureaucracy. As infrastructure for the next analyst to onboard quickly and the next decision to start from a higher base.

The artefact does not matter. A wiki, a shared playbook, annotated dashboards. What matters is that the team treats memory as part of the operating model, not an afterthought. Embedding this discipline is one of the quieter outcomes of a team enablement engagement.

Role rotation, not just promotion

The functions that compound rotate people across adjacent disciplines. A paid analyst spends a quarter close to the planning team. A planner sits with measurement for a project. A junior leader runs an optimisation programme outside their channel.

Rotation creates senior leaders who can lead the function, not just a discipline. It also surfaces operating-model issues that a single discipline would never see.

The starting question

If the function does not yet do these things, the first question is not how to install all five. It is which one is most painful right now.

For most teams, the honest answer is the leadership cadence. It is the cheapest to install, the most visible to the team and the one that unlocks every other element on the list. A monthly leadership conversation that is genuinely about the people, not the work, changes the trajectory of a function faster than almost anything else.

For teams scaling fast or rebuilding after a structural change, the conversation usually starts with team build and digital excellence: designing the function from organisation through to operating rhythm, rather than retrofitting capability into a structure that no longer fits.

The rest follows once it is running.