The most-cited number in organisational change is that roughly three quarters of change programs fail. The original source is a 1995 study by John Kotter, later echoed by McKinsey. Whether the exact figure holds or not, the pattern behind it does.
Most organisations do not need another centrally designed change program. They need a system that makes it normal to evolve practices, learn from what works, and remove what gets in the way, until better work becomes the default.
The failure rate stays high because many established change methods still treat change as a deterministic move from the current culture to an "ideal" future culture. Diagnose through centrally designed surveys, design the "target culture", then roll it out.
This fails for three reasons.
First, these methods underestimate how difficult it is to achieve and sustain change. Many programs never reach the root causes of bad practices, the unwritten rules that keep them in place. Unlearning old ways takes time and practice.
Second, they tend to discard the current situation in favour of a new future, throwing out the best of what already exists.
Third, they lack the ground-up nuance and flexibility needed for uncertainty. When you implement truly new ways of working, you cannot predict upfront what will and won't work. Fixed programs struggle with that reality. And you invite resistance when frontline employees are not involved in designing and testing the changes that shape their day-to-day work.
I'll focus on this third point: managed evolution. Creating autonomous, evolutionary loops of trial and error that gradually build better ways of working.
A simple loop: identify tensions, define practices, experiment, then scale what works.
Step 1: Start from friction, not from an "ideal culture"
Ask teams: "What stands in the way of your best work?"
You will get concrete friction, not abstract values. Friction is where practices live. Cluster the answers into a small set of themes so you can see patterns. Use categories such as: Purpose, Authority, Structure, Strategy, Resources, Innovation, Workflow, Meetings, Information, Membership, Mastery, Compensation.
Source: Brave New Work by Aaron Dignan
Pick one area that is not too difficult. Then brainstorm one new practice that might reduce the friction.
If needed, find inspiration in Daniel Coyle's The Culture Playbook or Aaron Dignan's Brave New Work. Tip: use AI as a sparring partner. Prompt: "Act as if you are Daniel Coyle and suggest new practices to help us improve [meetings / decision-making / workflow] so it becomes easier for people to do their best work."
Step 2: Define the experiment, then run it safely
Define:
- Where you will try it
- When, and for how long
- How you will evaluate success
- What "safe to try" means: what can go wrong, and what you will do if it does
Communicate it as an experiment, not a mandate. The goal is learning.
Step 3: Scale by removing barriers (this is where top management comes in)
If the experiment works locally, it still won't spread by itself. The blockers are often structural: decision rights, policies, KPIs, budget rules, IT constraints, governance, incentives.
This is where senior leadership earns their keep: remove the barriers that prevent a better practice from scaling. Once those barriers are gone, the proven practice will spread. Then move on to a slightly harder tension.
Word of caution: if top management only "sponsors" but does not actually remove constraints, this turns into a collection of local pilots. Lots of motion, little change. Step 3 is the difference between isolated experiments and real evolution.
Making this scalable without turning it into another program
Steps 1 to 3 can be supported by AI assistants instructed with specific skills: synthesising inputs from individuals and teams, suggesting candidate practices, spotting patterns across the organisation, and surfacing likely scaling barriers for senior management to discuss and decide on. They can work through input via Teams, Slack, or email, and feed a steady cadence into the management rhythm:
- Here is what is getting in the way of best work
- Here are the experiments running
- Here is what seems to work
- Here are the barriers that require leadership action
This turns managed evolution from something that works for one team into something that works across the organisation, without building another program around it.
Try it
Pick one team. Ask the Step 1 question. Design one practice, run one safe experiment, then bring the first real scaling barrier to leadership. That is where the real work starts.
Part 2 will go deeper on why many change approaches discard the strengths of today's culture and what to do instead. Part 3 will cover the root causes of bad practices and the unwritten rules that keep them in place.
This post was inspired by Aaron Dignan's Brave New Work. The experiments themselves can be drawn from Daniel Coyle's The Culture Playbook, a practical catalogue of small, repeatable moves that shape group behaviour.