Change that sticks

How to make operational change outlast the project.

Operational change fades when it’s run as a project with an end date, bolted onto unchanged leadership habits and held in one person’s head. It sticks when your team builds the capability to run the new way of working – and to keep adjusting it – without the person who introduced it.

By Simon Wakeman, Principal Consultant & Fractional COO
· 11 minute read

The project ended. The change didn’t last. You’ve probably seen it happen: a new planning process, a sharper handover between sales and delivery or a monthly review everyone agreed was overdue. It earns real enthusiasm at launch, then thins out week by week until, a few months on, it’s simply not happening any more. Nobody decided to stop. The usual conclusion is that the idea was wrong and a better one is needed.

Usually the idea was fine. This guide is about what makes change hold. It’s written for founders of scale-ups of roughly 15–75 people, where everything is still close enough to the founder that one person leaving, or one busy month, can undo a year of work.

Why don’t operational changes stick?

Most changes fade after launch rather than failing at it, for three connected reasons.

The change was run as a project. It had a start, a plan, a launch and an end date. Once it was “done”, everyone went back to normal work. But in a growing business there isn’t really a normal to go back to. The team keeps changing, clients keep changing and the tools are changing how work gets done almost month by month. So the project closes, the business keeps shifting, and twelve months on a fresh change project starts – usually larger than the last.

The new practice was grafted onto an unchanged system. Retros, new meeting formats and fresh operating model diagrams are the visible part of a change. Underneath, decisions are still made the same way, by the same people, and leaders still behave as they did before. The new practice sits on the surface, looks different for a while, and slowly loses to the old habits beneath it.

One person was carrying it. Every change has a champion. If the champion is the only one who really understands why it exists and how it works, the change leaves with them. That might be a founder who loses interest, a manager who moves on or an outside operator whose contract ends.

There’s a fourth reason that sits under all three. Teams in scale-ups are usually stretched delivering the day job. They have no spare capacity to get good at anything new, so the new thing never becomes second nature before the next busy period knocks it over.

What does change that sticks actually look like?

Change that sticks isn’t a business frozen into a better shape. It’s a business that has got better at changing.

The classic model of change is usually traced back to the psychologist Kurt Lewin: unfreeze the organisation, change it, then refreeze it in its new shape. It’s a useful way to think about a single, contained change. But a founder-led scale-up never really refreezes. Headcount grows, roles shift, new tools arrive and the structure that worked at 20 people creaks at 40.

So the aim is different. Organisation design firms such as The Ready and NOBL argue that change is now constant, not occasional. Their answer is to design organisations that keep adapting, rather than lurching from one big change programme to the next. Mallard shares that view. The design you land on today matters less than your team’s ability to keep improving it once the formal work is over.

A useful way to think about the ability to change is as a muscle. Regular use builds it up. Exercise it once a year and it weakens. A change has really stuck when three things hold:

  • Your people can run it. At least two people understand the new way of working well enough to teach it, and they could adapt it if the business changed.
  • Your practices hold it. It has a home in your regular rhythm – a meeting, a review, a playbook – rather than relying on someone remembering.
  • Your technology supports it. The tools reinforce the new way of working instead of quietly pulling people back to the old one.

That’s the same people, practices and technology model that sits behind a business operating system. Change that only touches one of the three rarely survives.

Why does the team slip back to the old way when things get busy?

Because busy is when habits win. Under pressure, people do what they know, and what they know is the old way.

Three things make the slide much more likely.

Nothing was taken away to make room. A team at full stretch has no room to take on anything new, however sensible. If a new weekly review is added on top of every existing meeting, it will be the first thing cancelled. Clearing space is part of the work of changing, not a side task. Before adding anything, decide what stops: a recurring meeting, a report nobody reads or a handful of low-value projects.

Leaders didn’t change first. Nobody in the business will change faster than the leadership team does. If the founder still skips the new meeting, still makes the call in the corridor or is three weeks behind on the new process, the team learns that the rule doesn’t really matter. They notice double standards very quickly.

The rhythm only runs when things are calm. The weekly operating meeting is the first casualty of a bad month. Yet that’s precisely when you need it – a structured hour to look at what’s actually happening and decide what matters. If your operating rhythm is optional when things get hard, it isn’t really a rhythm.

How do you build capability rather than dependency?

You build it the whole way through, not at the end. Capability transfer that’s saved for a handover week at the end of a project rarely lands, because by then the habits are already set: the expert does the work, and everyone else watches.

A better pattern runs in three stages, repeated for each part of the change:

  1. Do it together. The person leading the change does the work with the team, explaining their reasoning as they go rather than just the outcome.
  2. Let the team do it with support. The team runs it while the leader watches, coaches and debriefs. This is where most of the learning happens, and it’s usually slower than doing it yourself.
  3. Let the team do it alone. The leader steps out and checks back later. If things wobble, the fix is more coaching, not stepping back in.

It’s slower at first, and that slowness is where the learning happens. A founder or outside expert who jumps in gets a quicker answer today and a team that’s a little less able to find it tomorrow. People can’t take accountability for things they don’t feel competent to do, so competence has to come first.

This matters as much for outside help as for founders. An interim manager or consultant who does the job brilliantly but teaches nobody leaves a gap the day they go. We explore that pattern in the interim who became the dependency. The test of any outside operator is what your team can do without them, not what they did while they were there.

What should you write down, and who is it for?

Write down enough for someone new to run the work without asking you – and write it while you build the change, not afterwards.

Documentation that’s produced at the end of a project tends to describe what was intended rather than what actually happens. It’s also the first thing to get cut when time runs short. Documenting as you go is cheaper and more accurate, because the questions are fresh and the people doing the work can check it.

Three kinds of writing do most of the work:

  • Playbooks for repeatable work. How a new client is onboarded, how a quote is built, how a month is closed. Write them for the newest person in the team, not the most experienced. If a capable new starter can follow it, so can everyone else.
  • Decision logs. What you decided and, more importantly, why. Context that lives in a few people’s heads makes every change feel like a shock to everyone else, and it disappears when those people leave. The pattern is common enough that we wrote about it separately in when your best person leaves, so does the process.
  • Ownership. Each playbook and practice has a named owner, a role rather than a personality, and a date when it gets reviewed.

There’s a newer reader to write for too. If your team uses AI tools or agents, remember that they have no memory of how your business works unless someone gives it to them. Clear, current playbooks let your people and their AI assistants run a process the same way every time. Without them, every prompt starts from scratch and someone has to feed in the history by hand.

Keep it lean. The goal is a minimum viable operating system, not a manual nobody opens. Past a certain point, more documentation is often a way to avoid harder conversations about how people are actually working.

How do you keep change going once the project is over?

Give change a rhythm. Change that depends on one person’s energy to drive it slides back the moment they’re distracted. Change that has a regular slot in the calendar keeps going.

A few habits make continuous change normal rather than disruptive:

  • A regular moment to notice what needs to change. A quarterly review of priorities and ways of working, plus team retrospectives in between. The rhythm matters more than any single change you make through it.
  • Structure that’s openly temporary. Put a date on your org chart, such as “our structure until March”. It changes how people hear the next adjustment. A small change becomes an update rather than a verdict that the last one failed.
  • Smaller changes, more often. Adjust one team or one process at a time instead of saving everything up for a big reorganisation. A small adjustment is simpler to explain, and simpler to reverse if it doesn’t work.
  • Experiments for reversible decisions. Most changes to processes, meetings and tools are easy to undo. Let the people doing the work try them for a set period, then decide together whether to keep them. Save careful, slower thinking for the changes that are hard to reverse, such as a restructure or a senior hire. And if almost every experiment “succeeds”, they probably weren’t real experiments.
  • A plan for what people lose. People rarely resist change itself. They resist losing something – some control, a familiar routine, the feeling of being good at their job. Work out who stands to lose what, and you can plan around it rather than being caught out when people push back.

How do you know if a change has actually stuck?

Measure what keeps running after the person who introduced it has stepped back – and whether the work got better, not just different.

Adoption is the tempting measure. Everyone is using the new tool. Every team runs the new meeting. But adoption tells you very little about whether anything improved. A team can follow a new process faithfully and still make the same mistakes.

Better signals are about the work itself:

  • Are decisions sticking, or getting reopened a week later?
  • Is there less rework and fewer dropped handovers?
  • Are the same problems coming round less often?
  • Does the founder get pulled in less?

Then apply a simple test. Three months after the person leading the change steps back, is the new way of working still running, still owned and still being improved by the team? If it is, it has stuck. If it has quietly drifted, the change was being carried, not built.

Agree these signals before you start. Otherwise it’s very hard to tell useful change from churn, and very easy to declare success because the launch went well.

Does constant change wear people out?

It can, which is why the aim is steady adjustment rather than permanent upheaval.

Some fast-growing tech companies reorganise every few months and treat it as normal. Most founder-led scale-ups don’t need that pace, and many of their people would find it exhausting. But there’s a sensible middle ground between a structure that stays untouched for two years and constant churn.

What makes regular change tolerable is predictability. People cope far better with change they know is coming, through a rhythm they understand, than with a big reveal after the founder has disappeared for a few weeks to redesign everything. Even so, change you expect is still change. Ask your managers who’s struggling. A few strong performers will want steadier ground under them, and giving them that is part of good design, not a concession.

How do you rescue a change that’s already fading?

Pick one recent change that hasn’t held and work through three questions with your leadership team.

  1. What did it stop? Write down what your team stopped doing to make space for it. If the answer is “nothing”, that’s probably why it faded. Choose one thing to stop now.
  2. Who else could run it? Name who could keep it going if its champion left tomorrow. Where there’s only one name, pair a second person with them this month.
  3. Did we change too? Ask honestly whether the way the leadership team makes decisions and runs meetings has changed in the last six months, or whether you’ve only asked everyone else to change.

Then book a recurring slot – quarterly at the least – whose only job is to review how you work and decide what to change next.

How does Mallard make change hold?

Mallard works with founders whose businesses have outgrown how they run, and transferring capability to the team is part of the work from the start rather than a handover at the end.

In an Improve sprint, a senior Mallard operator works inside your business alongside your team for three months to fix the biggest bottleneck end to end, with success measures agreed up front. Your team builds the change with the operator rather than receiving it, so they can run it – and keep improving it – without outside help. Mallard stays on for 90 days after a sprint to make sure the change holds.

If you’re not sure which change to tackle first, the best first step is a free 30 minute founder review call.

Some commonly asked questions from founders…

Why do new processes fade after a few months?

Usually because nothing else changed around them. The new process sits on top of the same leadership habits, the same overloaded calendars and the same way of making decisions, so the old habits win as soon as the team gets busy. Fading rarely means the process was wrong – more often there was no room, no rhythm and no owner to keep it alive.

How long does it take for operational change to stick?

Longer than the project that introduced it. A new way of working usually needs at least one busy quarter, run by the team without outside help, before you can say it has held. That’s why the useful test isn’t whether something launched, but whether it’s still running – and still improving – three months after the person who introduced it has stepped back.

Should operational change be led by the founder or the team?

Both, in different ways. The founder and leadership team have to change their own habits first, because the team copies what leaders do rather than what they announce. But the people who do the work should design and run as much of the change as possible. Change that’s done to a team rarely survives; change the team helped build usually does.

How much should we document, and won’t it just go out of date?

Document the minimum someone new would need to run the work without asking you, and write it while you build the change rather than afterwards. Give each document an owner and a review date. Playbooks that are used every week stay current because people notice when they’re wrong. Documents nobody uses go stale whatever you do.

How do I make sure change survives a key person leaving?

Make sure the change never lived only with them. That means at least two people who can run each new practice, the reasons behind key decisions written down where everyone can find them, and ownership that sits with a role rather than a personality. If one departure would unravel it, the change hasn’t stuck yet – it’s just being carried.

Is continuous change exhausting for a team?

It can be, if every change arrives as a surprise. Smaller, regular adjustments made through a known rhythm are usually far less draining than a big reorganisation every couple of years. Tell people the structure is temporary, make room before adding anything new and check in on who’s finding it hard. Some people do their best work with more stability, and that’s fine.

How do I measure whether a change has worked?

Measure the work, not the adoption. Everyone using the new tool or running the new meeting tells you very little. Look instead at whether decisions are sticking, whether there’s less rework and whether the same problems are coming round less often. Agree those signals before you start, so you can tell useful change from churn.

Does AI make operational change harder to make stick?

It raises the stakes. AI tools change month by month, so how you split work between people and AI will need revisiting often. That’s manageable when reviewing how you work is already routine. When it isn’t, each new AI tool risks turning into its own disruptive change project. Treat every AI decision as provisional and agree up front what would make you look at it again.

How do I stop outside help becoming a new dependency?

Write the handover into the brief from the first day. Agree what your team should be able to do without outside help, name who will own each piece of work and check progress against that throughout. If nobody inside the business is learning the job alongside the outside operator, the dependency has simply moved from you to them.

How does Mallard make sure change holds after an Improve sprint?

Mallard treats capability transfer as part of the whole sprint rather than a handover at the end. In an Improve sprint, a senior Mallard operator works inside the business alongside your team for three months to fix the biggest bottleneck end to end, with success measures agreed up front. Mallard stays on for 90 days after a sprint to make sure the change holds.

Stop being a bottleneck in your own business

Book your free 30 minute
founder review call