Six Steps to Effective Change, Thirty-Five Years On

Six Steps to Effective Change, Thirty-Five Years On

Beer, Eisenstat and Spector showed in 1990 why change programmes do not produce change. Their six steps still hold. CLEAR agrees with most of them, disagrees with one, and adds what they left out.

1. The argument of 1990

Beer, Eisenstat and Spector opened with a claim that was heretical in the era of corporate culture programmes: "the fallacy of programmatic change". Most change efforts, they wrote, rest on the belief that attitudes lead to behaviour and behaviour leads to organisational results, so the job is to change attitudes first through communication, training and new values. Their field research found the causal arrow pointing the other way. Individual behaviour is powerfully shaped by the roles people play and the relationships they are in. Change the organisational context, the roles, responsibilities and relationships, and behaviour changes. Attitudes catch up afterwards, because people come to believe in what they have been doing.

The successful units they studied followed a recognisable pattern, which the authors described as "six overlapping but distinctive steps, which we call the critical path."

Three further ideas run through the article. Change should start at the periphery, in units with real business problems and willing managers, not at the centre. Top management's job is to create a climate for change and to spread lessons, not to drive a single programme. And formal structure comes last, because it freezes whatever exists at the moment it is applied.

2. Where CLEAR and the six steps agree

Behaviour follows context

This is the bedrock of both. CLEAR's Leverage step exists because its creator reached the same conclusion from behavioural science that Beer, Eisenstat and Spector reached from field research: the fastest and most durable way to change what people do is to change the system they do it in. Thirty-five years of evidence since 1990, from default effects to the hierarchy of controls, has strengthened rather than weakened this claim. The "fallacy of programmatic change" is the "fallacy of persuasion-first design" in behavioural language.

Joint diagnosis is Clarify

Step one asks for shared diagnosis of a concrete business problem by the people who must change. CLEAR's Clarify step asks for a shared definition of the outcome and measurable key results, built with the stakeholders rather than handed to them. The underlying logic is identical: people commit to problems they have helped define, and a change initiative without a jointly owned problem is a programme in search of a reason.

Start small, at the periphery

Beer and colleagues found change succeeding in plants and divisions, not in headquarters programmes. CLEAR's Experiment step tests interventions at identified leverage points on a small scale before any commitment. Both reject the corporate-wide launch, and both for the same reason: a small unit can learn and adjust, a corporate programme can only succeed or fail.

Lock in last

Step five's insistence that structure, systems and policy follow proven practice rather than lead it is CLEAR's Refine step: scale what the evidence supports, embed it, and start the next cycle. Both warn against the instinct to begin by reorganising.

Change is continuous

Step six, monitor and adjust, is the seed of what CLEAR makes central: the iterative loop. Beer, Eisenstat and Spector placed it last and described it briefly. CLEAR builds the whole method around it.

3. Where they differ

How you find the leverage points

The six steps say that managers should redesign roles, responsibilities and relationships around the business problem. They do not say how a manager works out which roles, responsibilities and relationships matter. In the companies studied, this came from experienced general managers' judgement, sometimes aided by consultants. CLEAR's Leverage step supplies a method: map the system, its flows, feedback loops, dependencies and incentives, and look for the points where a small change propagates. Task alignment tells you what kind of thing to change; a systems map tells you which one.

Testing versus doing

The six steps describe a unit that commits to a new way of organising and makes it work. There is diagnosis and there is monitoring, but there is no explicit stage in which a proposed change is tried in a form small enough to be abandoned. CLEAR's Experiment and Analyse steps add this. The difference matters because the 1990 study, like most change research of its era, was built on successful cases. It tells us what the units that succeeded did. It cannot tell us how many units did the same things and failed, and the authors acknowledged that renewal often took years and was uneven. An experimental step is how a method protects itself from learning the wrong lesson from a success story.

Measurement

Beer, Eisenstat and Spector are concrete about the business problem and vague about measurement. CLEAR's Clarify step requires key results, numbers that will show whether the change worked, before anything is designed. This is not a minor procedural difference. Without key results, step six's "monitor and adjust" has nothing to monitor against, and the judgement of whether a unit has been "revitalised" falls back on the impressions of the people who led the revitalisation.

The role of the top

Perhaps the most striking position in the 1990 article is its limited role for top management: create the climate, start at the periphery, do not drive a programme. CLEAR is less prescriptive here. In some systems the leverage point is at the top, in a policy, an incentive structure or an executive behaviour that shapes everything below it, and a map that found it there would be wrong to ignore it in favour of a peripheral pilot. CLEAR's answer to "where should change start?" is "where the map says", which sometimes agrees with Beer and colleagues and sometimes does not.

What explains the mechanism

The six steps rest on an empirical regularity: context changes behaviour. They do not explain it. Behavioural science since 1990 does. Roles and relationships work because they change opportunity (what is physically and socially possible), because new responsibilities create new cues and consequences, and because performing a behaviour under new conditions reshapes beliefs through experience rather than argument. CLEAR's grounding in that science lets a practitioner predict when task alignment will work and when it will not: when the missing ingredient is skill rather than context, for instance, redesigning roles will fail until the skill is built.

The 1990 claim that context changes behaviour is developed at length in two companion papers: one on behaviour-free design, which takes task alignment to its conclusion by asking whether the behaviour is needed at all, and one on tipping point leadership, which compares a second leader-centred model of change with CLEAR.

4. Side by side

5. What the six steps still teach

It would be easy to read this as CLEAR superseding an older model. That would miss what the 1990 article got right that behavioural methods, CLEAR included, can forget. Beer, Eisenstat and Spector were insistent that change is political and emotional, that managers who cannot move must eventually be replaced, that cohesion is built through shared work on real problems and not through team-building, and that headquarters' appetite for uniform programmes is the single greatest threat to genuine renewal. A systems map does not protect a change leader from any of these. The six steps are a reminder that the system being mapped is made of people with careers, and that a leverage point in the map is often a person who will have to give something up.

6. Conclusion

"Why Change Programs Don't Produce Change" has aged better than almost anything written about organisational change in its decade, because its central claim was empirical and correct: people change when their roles, responsibilities and relationships change, and programmes aimed at their attitudes mostly do not reach them. CLEAR stands on that claim. What it adds is the apparatus the 1990 authors did not have: a method for finding the leverage points, an experimental step for testing them, measurement that makes "monitor and adjust" meaningful, and a science that explains why task alignment works and when it will not. Read together, the six steps and CLEAR are not competing models. They are the same insight, thirty-five years apart, with the later one able to show its working.

© 2026 Erik Bohjort · clear-framework.com