---
title: "Does 70% of change fail? Tracing the claim"
description: "Where does the 70% change-failure figure come from? Trace it from Beer and Nohria to Hughes, then ask five questions before you quote it."
url: "https://staging.khoi.co.uk/en/blog/70-phan-tram-thay-doi-that-bai"
lang: en-GB
author: "Lê Quốc Khôi"
date: 2026-11-03
updated: 2026-11-03
topics: ["Change management"]
tags: ["Change management", "Change failure rate", "Theory E and Theory O", "Evidence-based management"]
---

# Does 70% of change really fail? Tracing the claim

## Key takeaways

- The sentence "about 70% of all change initiatives fail" sits in the opening paragraph of Beer and Nohria's Harvard Business Review article (2000) (Beer and Nohria, 2000, p. 133). According to Hughes (2011), the article cites no empirical evidence for it (Hughes, 2011, p. 453).
- Hughes reviewed five published instances of the figure, found no valid and reliable evidence behind it, and says he cannot establish for certain where it began (Hughes, 2011, pp. 451, 460).
- The meaning shifts with each repetition: from an "unscientific estimate" that 50 to 70 per cent of organisations doing reengineering missed their intended results (Hughes, 2011, p. 452) to a "brutal fact" about all change initiatives.
- Hughes does not say change is easy to get right. What to do with the figure: ask for the denominator and the definition of failure, and never use it as a justification or a threat.
- I read Hughes and By (2005) in full. For Beer and Nohria I read only the opening paragraph, so "the article has no evidence" is Hughes's conclusion.

A supplier sends a 25-person print shop in the Midlands a proposal for a new order-management system. Page one reads "about 70% of change initiatives fail" and prices a twelve-week support package to match. Where does that number come from, and what should the owner ask before signing? The shop and the proposal are hypothetical; the chain of citations below is real.

## Where does the "70% of change fails" figure begin?

Nobody can point to a certain origin, including the person who went looking. Mark Hughes (2011) reviewed five published instances of a 70% failure rate and wrote that it is impossible to say with certainty where the figure started (Hughes, 2011, p. 460).

The most quoted wording is in the opening paragraph of Michael Beer and Nitin Nohria's "Cracking the code of change": "The brutal fact is that about 70% of all change initiatives fail" (Beer and Nohria, 2000, p. 133). Hughes notes that the failure rate was often referenced back to that sentence, including in a textbook and in a refereed journal (Hughes, 2011, pp. 452, 455).

The earliest of Hughes's five is much narrower. In 1993 Hammer and Champy wrote about reengineering and offered an "unscientific estimate": as many as 50 to 70 per cent of organisations that undertake it "do not achieve the dramatic results they intended" (Hughes, 2011, p. 452). Hughes suggests this estimate may be why the specific figure of 70% was chosen, and he presents that as a possibility, not a finding (Hughes, 2011, p. 460).

## Did Beer and Nohria give evidence for the figure?

According to Hughes, no. The article cites no empirical evidence for "about 70% of all change initiatives fail"; its basis is the authors' 40 years of experience studying corporate change (Hughes, 2011, p. 453).

Hughes also searched *Breaking the Code of Change* (2000), the book the two authors edited, for anything close to support. He found two pieces, neither of them original data (Hughes, 2011, p. 453):

- Pettigrew cites Bashein and colleagues (1994) for around 70 per cent of business process reengineering interventions missing expectations. That estimate rests on Hammer and Champy's "unscientific" one.
- In chapter 21, Beer writes that approximately 70 per cent of total quality management programmes are *perceived* by top management to have fallen short of their aspirations at launch. The supporting reference is his own earlier speculation of 75 per cent (Spector and Beer, 1994).

## Who cited whom, and what did each say?

The table covers the five instances Hughes reviewed, plus By (2005), which I read separately. The middle column paraphrases the original wording. The last column gives the evidence offered, according to Hughes (according to me, for By).

| Year and source | Wording (paraphrased) | Evidence offered |
|:--|:--|:--|
| 1993, Hammer and Champy, book on reengineering | As many as 50–70% of organisations doing reengineering miss the dramatic results they intended | Called an "unscientific estimate" (Hughes, 2011, p. 452) |
| 2000, Beer and Nohria, Harvard Business Review | About 70% of all change initiatives fail | No empirical evidence cited (Hughes, 2011, p. 453) |
| 2005, By, Journal of Change Management | A failure rate of around 70% of all change programmes, citing Balogun and Hope Hailey (2004) | No underlying data in the paper (By, 2005, p. 370) |
| 2008, Kotter, *A Sense of Urgency* | After years of study, an estimate that more than 70% of needed change is not launched, not completed, or finishes late, over budget and short of its aspirations | Hughes: far more ambiguous than "70% of initiatives fail" (Hughes, 2011, p. 454) |
| 2008, Bain (Senturia and colleagues) | With each survey, 70% of change initiatives still fail | No research cited; tracing back reaches a 2002 piece that itself cites Beer and Nohria (Hughes, 2011, p. 454) |
| 2009, McKinsey (Keller and Aiken) | Only 30% of executives surveyed called their change programmes completely or mostly successful | Cites a survey of 1,546 executives (Hughes, 2011, pp. 454–455) |

## How does the number change meaning as it travels?

It swells, from an estimate about one method into a "brutal fact" about every initiative.

The people involved have corrected it. In 1995 Hammer, with Stanton, wrote that the 50–70 per cent estimate was a simple descriptive observation that had been misrepresented and distorted into a normative statement. They added: "There is no inherent success or failure rate for reengineering" (Hughes, 2011, pp. 452–453).

Kotter folds three different outcomes into one figure (Hughes, 2011, p. 454). His 1995 article, going by the summary on the Harvard Business Review site, states no failure rate: he reports watching more than 100 companies try to remake themselves, a few very successful, a few utter failures, most in between with a tilt towards the lower end (Kotter, 1995). Hughes notes that he did not claim only 30% of change programmes succeed (Hughes, 2011, pp. 454–455).

McKinsey shows best how much the definition matters. Hughes compared the 30% with the original 2006 report, in a section headed "Success is the norm". There, 38 per cent said their transformation was completely or mostly successful at improving performance, and 30 per cent said so about improving organisational health. About a third said "somewhat" successful on both counts, and about one in ten said completely or mostly unsuccessful (Hughes, 2011, p. 455). So the 30% matches one yardstick, and the group reporting failure was about one in ten, not 70%.

## What does Hughes conclude, and what does he not say?

The 70% story exists as a popular narrative, but Hughes finds no valid and reliable empirical evidence for it (Hughes, 2011, p. 451). He goes further: any inherent failure rate is hard to defend, because change is ambiguous, context-dependent, judged differently by different people, unfolds over time and resists measurement (Hughes, 2011, pp. 451, 461).

What he does not say matters too. He does not claim that most change succeeds, and he offers no replacement figure. Knocking down an unsupported number does not make change easy.

By's paper, in the same journal, shows the figure running the other way. By cites Balogun and Hope Hailey for a failure rate of around 70 per cent, then suggests that such a poor rate indicates a lack of a valid framework for managing change (By, 2005, p. 370). His recommendations then call for methods to measure the success rate of change initiatives to be designed (By, 2005, p. 378). My reading: the figure serves as a premise, and only afterwards does the paper ask for a way to measure it.

## What should a manager do when someone quotes the number?

Do not use it as a reason to win budget or as a threat to staff, and do not wave it away merely because it lacks a source. Ask five questions, drawn from the places where the citation chain above breaks.

1. 70% of what? How many initiatives, in which industry, in which year? Bain's version cites no study, and McKinsey's 30% is one yardstick in a survey of 1,546 executives.
2. What counts as failure? Missing the original goal, finishing late, overspending, or not lasting. Kotter folds three outcomes into one figure.
3. Who judged? Top management's perception, the people doing the work, or financial data. Beer's TQM figure is a perception held by top management.
4. When was it measured? Six months in and five years in can give different verdicts; time is one of Hughes's five reasons.
5. Does the chain reach original data? If you follow it back and meet only an "estimate" or an opening sentence, it is an opinion, not a statistic.

For the hypothetical print shop, one email asking these five questions would show whether the supplier holds data or only a slide. The habit suits you if you want to buy a specific result. It does not suit you if you use "nobody knows the true rate" as an excuse to delay: no solid statistic does not mean any given change is easy.

If the change is already decided and staff are not using it, [four theories of resistance](https://staging.khoi.co.uk/en/blog/bon-ly-thuyet-khang-cu-thay-doi) and [Kanter's ten reasons](https://staging.khoi.co.uk/en/blog/muoi-ly-do-khang-cu-kanter) help diagnose each case in place of a general rate.

## Going deeper: what did I read, and where are the limits of Hughes's analysis?

I read Hughes (2011) in full through an unofficial copy of the article in the *Journal of Change Management*, citing the journal's own page numbers, and By (2005) in full through a PDF of the printed paper retrieved from a web archive. For Beer and Nohria I read only the opening paragraph on the Harvard Business Review site; everything about the rest of the article, including "no evidence", is Hughes's account. For Kotter (1995) I read only the summary and first paragraph. I did not read Hammer and Champy, Kotter 2008, the Bain and McKinsey reports, or Beer's book chapter; sentences from those are Hughes's quotations, paraphrased by me.

The first limit: Hughes reviewed five instances he calls high-profile, not every appearance of the figure (Hughes, 2011, p. 452). The second: he suggests consultants may have deliberately promoted the number, but the wording is "may have", so it is speculation (Hughes, 2011, p. 455).

Hughes calls the case studies in Beer and Nohria's article informative, though they offer no evidence for the figure. He says they present Scott Paper and Champion International as near-pure forms of the two types and ASDA (UK) as an illustration of combining them (Hughes, 2011, p. 453); see [ASDA and Scott Paper](https://staging.khoi.co.uk/en/case/asda-scott-paper-theory-eo).

> **Figure:** [Theory E and Theory O of change](https://staging.khoi.co.uk/en/framework/theory-e-theory-o) Theory E and Theory O of change (Beer and Nohria, 2000; redrawn from the library framework).

Framework detail: [Theory E and Theory O of change](https://staging.khoi.co.uk/en/framework/theory-e-theory-o).

## Frequently asked questions

### Why does a figure with no evidence behind it last so long?

Hughes (2011) says it is hard to establish that too. He points out that promoting the figure gives academics a reason to research, consultants a reason to sell services, and the managers of change a rationale for their own work. He also quotes Beer, who once mentioned in passing an "unconscious collusion" between academics, consultants and managers.

### If I have to mention the 70% figure, how should I word it?

Name the nearest source and the state of the evidence, for example: an estimate repeated from Beer and Nohria (2000), with no underlying data according to Hughes (2011). That tells the audience it is a widely quoted opinion, not a measurement. Do not put it on the first page as a justification.

### How do I know whether my own change project is on track?

An average cannot answer that, since Hughes argues that outcomes depend on the context of each organisation. A simpler habit is to write down three lines before you start: which measure will show success, at what point you will read it, and who will judge. That is my suggestion, not the finding of any source.

### If 70% is hype, should we drop change management?

No. Hughes targets the statistic, not the practice of managing change. He writes that research over the past thirty years has deepened our understanding of the complexities of change, and that this progress has informed both theory and practice.

## References

- Beer, M. and Nohria, N. (2000) *Cracking the code of change*. Harvard Business Review. Available at: https://hbr.org/2000/05/cracking-the-code-of-change (Accessed: 2026-11-01).
- By, R.T. (2005) ‘Organisational change management: a critical review’, *Journal of Change Management*, 5(4), pp. 369–380. https://doi.org/10.1080/14697010500359250
- Hughes, M. (2011) ‘Do 70 per cent of all organizational change initiatives really fail?’, *Journal of Change Management*, 11(4), pp. 451–464. https://doi.org/10.1080/14697017.2011.630506
- Kotter, J.P. (1995) *Leading change: why transformation efforts fail*. Harvard Business Review. Available at: https://hbr.org/1995/03/leading-change-why-transformation-efforts-fail-2 (Accessed: 2026-11-01).

---

Canonical HTML version: https://staging.khoi.co.uk/en/blog/70-phan-tram-thay-doi-that-bai