70% of change projects fail: Bollocks!
Once upon a time in a galaxy far, far away...
‘Sadly, we must report that despite the success stories described in previous chapters, many companies that begin reengineering don’t succeed at it...Our unscientific estimate is that as many as 50 per cent to 70 per cent of the organizations that undertake a reengineering effort do not achieve the dramatic results they intended.’ (Hammer and Champy 1993, p200)
An unscientific estimate. No definitions of success. No investigation of validity of expectations. 70% of BPR projects fail. Sexy stuff, people.In 1995, Professor John Kotter publishes the article “Leading Change” in the Harvard Business Review. Rather than quote studies, he notes he has “observed” over a 100 companies in the previous ten years with success varying. He is circumspect about success and failure rates, noting the varying stages and reasons for difficulty. Kotter’s 1995 work is often referenced as a source. It’s not in this article. The eight-step framework is in this one.In 2000, researchers Michael Beer and Nitin Nohria published “Cracking the Code of Change” in the Harvard Business Review. The article is actually about their work on Theory O and Theory E of change. But the sentence that grabbed the attention of the consulting world was almost a throw away line at the beginning:
‘The brutal fact is that about 70% of all change initiatives fail.’ (Beer and Nohria, 2000, p133).
Nothing to support it, no mention of where this fact has come from, how the figure has emerged to be a “brutal fact”. But it does set up a need for an alternative theory of change (eg Theory E and Theory O).From an academic perspective Mark Hughes published a fascinating challenge to the statistic in the Journal of Organisational Change Management in 2011. From his analysis, many of the subsequent published papers form a version of a set of academic matryoshka dolls. Examination of their proof of the 70% citation inevitably leads to Hammer and Champy and Beer and Nohria. The mind boggles how many times this statistic has set up a justification for the academics following endeavour. Indeed he notes that Michael Hammer distances himself from the original statement
“Unfortunately, this simple descriptive observation has been widely misrepresented and transmogrified and distorted into a normative statement . . .There is no inherent success or failure rate for reengineering. (Hammer and Stanton, 1995, p. 14, cited in Hughes, 2011).
These two sources (Hammer and Champy, and Beer and Nohria) made the curriculum reading lists of pretty much every undergrad and postgrad in the western world. And thus influenced a very large cohort of managers, consultants, project managers and change management practitioners.The figure gets a life of its own, in 2008 in "A Sense of Urgency", Professor John Kotter “estimates” more than 70% of needed change fails. His website states “Thirty years of research by leadership guru Dr John Kotter have proven that 70% of all major change efforts in organizations fail". Yet, I struggle to find any peer reviewed publications by Kotter on the research that led to this. But I fully understand that some-one who researches in the area may be reluctant to challenge this and ask to see the research in order to evaluate the research design. Some sacred cows you don’t touch…From an academic perspective you have a choice at this point. Do you position against famous professors with best selling books and challenge the “unscientific” statement and “estimates”? To challenge Beer and Nohria on the “brutal fact” is to distract from what is a pretty useful theory and contribution to change (Theory X and Theory O). Maybe you need to wait twenty years to do so. It may be more prudent for career progression to stand on the shoulders of giants and build incremental “knowledge” on 70% failure rates.So then large consulting firms and IT vendors get in on the act. Some-where along the line some pretty good studies on project implementation and benefits get further twisted into a persistent myth that 70% of all change projects fail. Statistics like that can be very useful in selling services and products. They create fear. If you don’t use our services you may be in the 70% …that would be bad.Industry heavy weights and thought leaders continue to popularise the statistic with Daryl Conner using it as a big stick to beat up change practitioners and admonish them to do better (why after 30 years are we still having 70% of our change projects fail? We must be culpable). Ron Ashkenas recently used it in the HBR again. This means it must be true.
But it’s not. And here’s six reason’s why:
1. The definition of change project is questionable
A lot of the research studies that reference the 70% failure talk about success of project implementation. Project implementation success is often very different to change management success. Yes, any project by virtue of purpose relates to change – eg it is created to change something, deploy something, and improve something. But not all projects are “change projects”. To assume so is conflation.A change project needs to have a change management methodology employed and change management resourcing. The studies referenced as proof of the 70% statistic do not control for the presence of a change manager or a change methodology. If neither of these were present I would argue that you couldn’t make any statement about change projects being successful or not.Take a look at the studies that do control for change management. Towers and Watson’s Change and Communication ROI studies reveal that organisations that have a change management approach have 2.5 greater financial returns than companies that don’t.In IBM’s 2008 study Making Change Work, it was identified that of the 20% of companies who represent “change masters", their success could be attributed to four factors:
- Realistic awareness and understanding from leadership of the complexity of change
- A systematic approach to change (eg a methodology)
- Dedicated change managers and change resourcing
- Permitting the right investment for change.
To my view, if you don’t have these four factors, I’m not sure you can include in a study about change management success.The notion of “control” in a research design is critical. Finally, earlier this year (and 20 years from the original Hammer and Champy statement) researchers Barends, Janssen, Wouter, ten Have and ten Have publish a marvelous meta-analysis of 563 studies in change in the Journal of Applied Behavioural Science. Only 2% use a case control design, and 13% used control groups.
2. The definition of “success” is questionable.
Looking at some of research quoted success is defined as: did it meet expectations, were benefits realised, was the project delivered in full, on time, on budget.In my experience change success is defined as
- People are using the new technology, policies, and adopting new behaviours
- The business outcomes have changed for the better
You can go further (and should go further) and track metrics at various stages of the change.Change success is rarely measured in absolutes. Things change during the course of an initiative. Sometimes dramatically. Often business sponsors have an unrealistic expectation on what success looks like and when it will happen. It based on personal KPI reporting, not what change really looks like in organisations. If you have change resourcing at a senior level you can reset expectations. If you don’t have some-one who knows change at a senior level influencing these expectations of success you have a senior executive filling out a survey saying that the [change] project failed (an absolute).
3. Success is measured at the wrong time.
There is recognition that successful change takes time – moving up the adoption curve can be a lengthy process. And that depends on the type of change and the type of organisation. "Was the project delivered in full and in time" is simply not a “change success” metric. We know from practice, that culture change can take many years to embed. As change practitioners we need to interrogate expectations of the timeliness of benefits realisation. Benefits realisation is more than in full on time and on budget. For more on this, have a look at Conner Partners paper on Installation or Realization; it’s a great read.
4. The units of analysis are not the same.
The multiple studies reference different types of companies, industries and types of change. Without a proper meta-analysis you can’t make the claim that this is a consistent finding. You are comparing apples, with oranges, tossing in a grape or two and saying the fruit salad is a worrying story. It’s handy that they look similar, but the units of analysis are not comparable. Changing a culture has very different success factors, time frames and methodology to a large-scale system implementation. I take my hat off to Martin Smith for his early efforts at a meta-analysis with “Success rates of different types of Change” in Performance Improvement - this is more like what we need. It is telling though that his concluding comments steer away from a definitive statement about what success looks like during organisational change, and instead makes suggestions to readers on how to use these studies in understanding their own change efforts. The reasoning of this article combined with meta-analytic rigour of Barends et al' paper starts to tell us a lot more about organisational change success.
5. I don’t think I am [that] special, nor my peers.
If this statistic were to be true, I would have 70% of my change initiatives shelved as failures. So would my peers. We don’t. We’re pretty good. I’ll grant you that. But I don’t think we are the outliers here.Change is difficult, don’t get me wrong. It is even more difficult in organisations where sponsors and leaders don’t understand the need for change management. No doubt about that. But is the field of change management fraught with persistent failure. Absolutely not. There is such a wide variety of types of change, scale of change, scope of change that to create a mean is well, mean-ingless.The next time you meet some-one with the title of change manager strike up a conversation. Ask them how many of their initiatives have failed? It is highly unlikely they will say anywhere near 70%. Ask them then about what would have made many of their projects a better success in a quicker period of time. Then you’ll have some useful insight.
6. A Career Limiting Admission for a CEO
Seriously. You want me to believe that 70% of the worlds CEOs have led failed change efforts? Really? Is the talent pool for CEOs that large? I’m not sure they would still be CEOs if that were the case. Even if the surveys are anonymous, some-where there are 70% of company boards looking at poor performances from their CEOs. I struggle with that.
A call to action
Practitioners:
When some-one uses this statistic, call them on why they think it is true. Have they read an influencer or delved into the empirical research? How was success defined? Was the presence of change management support included? Be informed and responsible in your use of the statistic. Please don’t use this statistic to suggest that change management is difficult or risky to do. That’s just plain wrong.While I don’t agree with Daryl Conner’s view that change practitioners have culpability for the 70% failure statistic, I do think his 23 questions in Physician Heal Thyself are excellent. Create a community event where you focus on these questions – collectively lift the quality of change management practice. We cannot and should not shy away from improving change success rates.
Researchers:
Mark Hughes has made an excellent start with his paper on “Do 70% of all Organisational Change Efforts Really Fail?. This unpacks why it is a myth. But let's get to the real answer. There is much, much more to do. There is ontological opportunity in addressing understanding the social construction of management myths. Eric Abrahamson’s Managerial Fads and Fashions: the Diffusion and Rejection of Innovation (1991) will be useful as a starting point. There will be more in the critical management literature.With regards to epistemology, Barend’s et al’s 2013 paper is impressive. One of their implications for further research is to conduct more replication studies. So there is argument for epistemological contribution by doing more like this. Replication studies are high risk though from a publishing perspective. This may be better suited to an honours student (Australian academic pathway) It’s a tough one. Given the lack of quality in OCM research when it comes to success rates, I would argue that there are a series of research studies that involve control case designs, focusing on a specific type of change with each study. So find 30 cases of culture change – control for methodology, resourcing and include time series collection of data. Then do it with restructures, and then systems implementations. Then we build a body of knowledge.But above all, regardless of the design be clear on face validity: Start with qualitative research on practising change managers. Talk to them and their sponsors on how they define change success. Build your surveys using those definitions and constructs. Then look at the reliability. Use that research on different industries, different types of change. Control for what differs. And then please make sure it gets into a HBR! (yes, I know…) Or share the working papers with the MBA students. Get it out there
Vendors:
Do your studies on the relative difference that change management makes. When you use fear as motivator you run the risk of freaking the customer out and they run away from the whole concept or become paralysed (Fight, flight and freeze). And nothing gets changed at all. Better to maintain status quo because 70% of change projects fail anyway…. Post-script. Timing hey? Just before hitting publish, I come across Jason Little’s post on the same topic. A week ago. It’s a great read. Jason shares more about what the studies tell you, but there are very similar themes to this post. With less snark and frustration ; -) To my delight, Heather Stagl has also taken it on earlier too. And Gail Severini has initiated a terrific discussion in the OCP group with some great insights coming out and pointed me to Barend’s at al, and Smith’s papers. This post is improved for her comments and viewing of the original draft.So it looks like I’m in good company – would it be too optimistic to say we are at a tipping point?