Methodology: Search Strategy and Selection Criteria

With the objective of assessing what the rate of business strategy implementation failure/success is, we carried out an extensive review of the literature. First, we tried to identify all publications in scholarly journals at the EBSCO Host Research Databases that present estimates for this rate. Several search strings, including strateg* and fail*, strateg* and success*, strateg* and implement*, transfor* and fail* were applied to keywords, titles and abstracts of the publications. In the search strings above, the asterisk sign (*) represents wildcard characters. Second, within this first set, we identified all papers from business journals. All those publications from the list that were not actually publications in the business area even though they mentioned the search terms in the keywords, title or the abstract were omitted from further analysis. Third, we analysed the abstracts of all publications on this final list in order to assess their relevance for our research. We have considered only relevant studies that could present a percentage of failure (or of success) on business strategy execution consistent with the definition of failure presented previously. Fourth, for those publications considered to be relevant, we analysed the full text in order to determine whether an estimate of the failure rate was provided. Fifth, bibliographic references in the selected papers were also used as a source to identify papers or other evidence not captured in our electronic database search. It is relevant to note that the studies dealing with this issue have been authored by academics and practitioners, including consulting companies, and that not all of these studies have been published in academic journals. Therefore, a search strategy based exclusively on evidence documented in academic journals would be incomplete. Consequently, our sixth step consisted of additional searches carried out on the Internet search engine Google, on the websites of major consulting companies and on several national library on-line catalogues (England, United States, Ireland, Scotland, Canada, Australia and Portugal), which allowed the identification of some additional and relevant studies. Unfortunately, however, some of these studies were not available for consultation and it was not possible for us to gain access to either a hard or an electronic copy. Interestingly, many of these unavailable studies were authored by consulting companies (Arthur D. Little, 1992; A.T. Kearney, Reference Kearney1992; Prospectus Strategy Consultants, 1996) and were abundantly quoted, even by reputed academic researchers. Finally, we have also contacted by e-mail the consulting companies, the individual authors of the reports, when their names were publicly available and the authors who have quoted those studies. In total, more than 45 e-mails were sent. In spite of all the efforts made to obtain copies of the studies, most of these efforts proved unfruitful. Many of the companies and authors contacted replied, but we did not succeed in obtaining the required information either because the studies were no longer available (e.g., A.T. Kearney, A.D.L., Prospectus) or because the companies were unable to assist individuals with specific research requests (e.g., B.C.G., McKinsey).

Therefore, the literature reviewed in this paper includes all the academic studies that have met the search criteria above and the consultancy studies that were relevant and available for consultation. The latter account for 45% of the studies analysed. The results of this search are presented and discussed in the next section.

Strategy Implementation Failure/Success Rates

Although the literature on the topic of strategy implementation failure/success is not in short supply, the existing studies are mixed in terms of their features (e.g., failure rate estimated, amount of effort involved in the estimation, complexity and quality of the methodology used, unit of analysis, criteria adopted to define success and research strategies adopted) and this requires special care in comparing their results. The most significant features of the studies considered for this research are summarised in Tables 1 and 2. Table 1 lists studies that have focused on general business strategies, and Table 2 lists studies that have focused on specific business strategies. Although the former table aims to be exhaustive, in the sense that it shows all the studies that our search strategy identified, the latter, aims to be illustrative of the variability of the available estimates.

Table 1 Studies estimating general business strategy implementation failure rates

Notes.

^a Study by a consulting firm or by authors associated with consulting companies.

^b The study was not available on-line. We did not receive replies to our e-mails or the replies were negative.

na=information not available.

Table 2 Studies estimating specific business strategy implementation failure rates

Notes.

^a Study by a consulting firm or by authors associated with consulting companies.

^b The study was not available on-line. We did not receive replies to our e-mails or the replies were negative.

^c In the same year, in a study by O’Brien and Voss (Reference O’Brien and Voss1992), the authors concluded that most British organisations were having problems developing TQM. However, they noted that most UK organisations were in the early stages of developing a total approach to quality, that is, in the beginning of implementation.

DCs=developed countries; IVJs=international joint ventures; LDCs=less-developed countries; na=information not available; TQM=Total Quality Management.

The information in these tables is organised in five columns. The first column indicates the author(s) and year of the study and it is listed chronologically. The research method used to estimate the rates of failure/success and the variables against which such rates were assessed are described in the second and third columns, respectively. The fourth column indicates the estimated rate of failure presented by each study. Finally, the last column records some additional comments on each study.

The most appropriate conclusion that can be drawn from the analysis of Tables 1 and 2 is that it is difficult to provide accurate estimates for rates of failure of strategy implementation. The studies carried out so far by researchers and management consulting firms have obtained mixed results regarding the success and failure rates of business strategy implementation. In fact, as can be seen from the fourth column of the tables, the range of variation of the estimates is remarkable. If we first analyse the studies that focus on business strategy implementation in general, we can verify that the estimated rates of failure have ranged from 28 to 90%. When we turn to the studies that have focused on the implementation of specific business strategies this range is even wider. While some studies have obtained rates of failure as low as 7–10% (e.g., Taylor, Reference Taylor1997; Walsh, Hughes, & Maddox, Reference Walsh, Hughes and Maddox2002), others have obtained rates of failure as high as 80–90% (e.g., Voss, Reference Voss1988, Reference Voss1992; A.T. Kearney, Reference Kearney1992). Therefore, although it can be claimed that up to 90% of strategic initiatives fail, as this is the upper limit of the estimates provided in the literature, this is likely to be an overestimation.

Two major reasons support this view. First, most of the higher estimates presented in the literature come from consulting firms (e.g., Kiechel, Reference Kiechel1982, Reference Kiechel1984; Judson, Reference Judson1991; A.T. Kearney, Reference Kearney1992; Prospectus Strategy Consultants, 1996; Hackett Group, 2004a, 2004b; Dion, Allday, Lafforet, Derain, & Lahiri, Reference Dion, Allday, Lafforet, Derain and Lahiri2007). Although we were unable to access the scientific rigour of some of these studies, as it was not possible to obtain details regarding the robustness of the research methodologies used and the results achieved, it has long been recognised that some overestimation may have been committed by consulting firms (Powell, Reference Powell1995). Overestimated failure rates can be used to the advantage of consulting firms, namely as a marketing strategy to convince customers of the importance of adopting their services (Powell, Reference Powell1995). Second, the results on the tables seem to suggest a downward trend on the estimates of failure, indicating that the percentage of strategic initiatives that fail has decreased over time (see Figure 1), a likely result of the scientific progress made in this field over the past two decades and its inclusion in business education programmes. In particular, the identification of obstacles to strategy implementation and a better understanding of the ways they interact with each other, made by both researchers and practitioners (e.g., Alexander, Reference Alexander1985; Ansoff & McDonnell, Reference Ansoff and McDonnell1990; Kotter, Reference Kotter1995; Beer & Eisenstat, Reference Beer and Eisenstat2000; Kaplan & Norton, Reference Kaplan and Norton2001) might have played an important role in improving the rates of failure over the years. Therefore, although some of the higher estimates could have been appropriate and reflect the true dimension of the problem one or two decades ago, they are likely to be outdated nowadays. This is also the case because time since adoption of a new strategy contributes to a better internalisation of the elements of that strategy and consequently to a better performance (Powell, Reference Powell1995; Prajogo & Brown, Reference Prajogo and Brown2006). Considering that some strategies and some management tools have been in practice for a long time, it is likely that familiarity with these strategies and tools has increased leading to the accumulation of knowledge and, consequently, to more successful implementations (Taylor & Wright, Reference Taylor and Wright2003). Several other explanations can be offered to justify the downward trend on the estimates of failure. For example, companies may follow successful early adopters, benefiting from their experience, thus resulting in the improvement of failure rates. Companies may also have become more aware of the need to carefully customise new strategies or management tools to their characteristics and to the contexts in which they operate, instead of blindly adopting general undifferentiated strategies and tools. Independently or in combination each of these factors might help explain the apparent improvement in failure rates.

Figure 1 Business strategy implementation failure rates. Note: For this figure, we used the rates in Table 1. When two rates were given in any one study, we used the average for the figure.

It seems therefore reasonable to assume that the current rates of failure are well below some of the estimates often quoted in the literature. However, if this is the case, what is then the real percentage of strategies that fail? Although there have been several studies on this issue in the past two decades, our view is that the current state of affairs does not allow a robust estimate to be provided. Several reasons can be advanced for this.

First, the studies discussing the success/failure rate of strategy implementation vary considerably in the amount of effort put into the estimation of the rate. In some of these studies, the estimation of the rate of failure/success was their main objective (e.g., Golembiewski, Reference Golembiewski1990; Park, Reference Park1991; Wilkinson, Redman, & Snape, Reference Wilkinson, Redman and Snape1994; Pautler, Reference Pautler2003; Makino, Chan, Isobe, & Beamish, Reference Makino, Chan, Isobe and Beamish2007). In other studies, this objective was part of a broader research agenda (e.g., Beamish, Reference Beamish1985; Voss, Reference Voss1988, Reference Voss1992; Taylor, Reference Taylor1997; Nutt, Reference Nutt1999; Walsh, Hughes, & Maddox, Reference Walsh, Hughes and Maddox2002; Taylor & Wright, Reference Taylor and Wright2003; McKinsey, Reference McKinsey2006), while in others the rates of success/failure were presented as complementary information in an introduction or as an aside (e.g., Gray, Reference Gray1986; Harrigan, Reference Harrigan1988a; Hall, Rosenthal, & Wade, Reference Hall, Rosenthal and Wade1993; Mohrman, Tenkasi, Lawler, & Ledford, Reference Mohrman, Tenkasi, Lawler and Ledford1995; Lewy & Mée, Reference Lewy and Mée1998a, Reference Lewy and Mée1998b; Sila, Reference Sila2007). An implication of the effort put into the estimation of the rate in each study is the consequent impact on the complexity of the computational method used. In some studies, the computation is very simple (e.g., Beamish, Reference Beamish1985; Harrigan, Reference Harrigan1988a; Sila, Reference Sila2007), while in others it is much more complex and demanding (e.g., Golembiewski, Reference Golembiewski1990; Park, Reference Park1991).

Second, these studies are not easily comparable because the criteria used to define success/failure are very distinct and consequently, they can account for some of the differences between estimations. It is possible to distinguish between ‘technical success’ and ‘competitive success’ (Voss, Reference Voss1992), between ‘success as process ease’ and ‘success as process outcomes’ (Bauer, Falshaw, & Oakland, Reference Bauer, Falshaw and Oakland2005) and, similarly, between ‘implementation success’ and ‘organisational success’ (Hussey, Reference Hussey1996; Mellahi & Wilkinson, Reference Mellahi and Wilkinson2004). The higher rates of failure estimated may depend on a stricter sense of success adopted by researchers.

Estimates of technical success and success as process ease may be higher than estimates of success as process outcomes or organisational competitive success in the marketplace, since more internal and external contingencies can affect the latter types of success. In Tables 1 and 2, we have reported mainly failure rates from a ‘competitive success’ or an ‘organisational success’ perspective. Even so, the studies in the tables are not easily comparable because in some cases researchers relied on managements’ perceptions to derive an estimate of success/failure (e.g., Beamish, Reference Beamish1985; Gray, Reference Gray1986; Voss, Reference Voss1988, Reference Voss1992; Taylor & Wright, Reference Taylor and Wright2003), whereas in others, they used more objective measurements (e.g., Golembiewski, Proehl, & Sink, Reference Golembiewski, Proehl and Sink1981, Reference Golembiewski, Proehl and Sink1982; Golembiewski, Reference Golembiewski1990; Hall, Rosenthal, & Wade, Reference Hall, Rosenthal and Wade1993; Pautler, Reference Pautler2003; Makino et al., Reference Makino, Chan, Isobe and Beamish2007). Furthermore, some studies have used a single criterion to define success/failure (e.g., Gray, Reference Gray1986; Walsh, Hughes, & Maddox, Reference Walsh, Hughes and Maddox2002; Sila, Reference Sila2007), whereas others have used multiple criteria (e.g., Golembiewski, Reference Golembiewski1990; Park, Reference Park1991; Wilkinson, Redman, & Snape, Reference Wilkinson, Redman and Snape1994; Mohrman et al., Reference Mohrman, Tenkasi, Lawler and Ledford1995).

Third, different studies have used different research strategies to estimate the rate of success/failure of strategy implementation. Some researchers have adopted a case study approach (e.g., Voss, Reference Voss1988, Reference Voss1992; Hall, Rosenthal, & Wade, Reference Hall, Rosenthal and Wade1993; Lewy & Mée, Reference Lewy and Mée1998a, Reference Lewy and Mée1998b; Nutt, Reference Nutt1999). Others have employed a survey method (e.g., Beamish, Reference Beamish1985; Wilkinson, Redman, & Snape, Reference Wilkinson, Redman and Snape1994; Mohrman et al., Reference Mohrman, Tenkasi, Lawler and Ledford1995; Walsh, Hughes, & Maddox, Reference Walsh, Hughes and Maddox2002; McKinsey, Reference McKinsey2006; Makino et al., Reference Makino, Chan, Isobe and Beamish2007; Sila, Reference Sila2007), while still others have used a combination of methods (e.g., Gray, Reference Gray1986; Harrigan, Reference Harrigan1988a; Charan & Colvin, Reference Charan and Colvin1999; Taylor & Wright, Reference Taylor and Wright2003). It is well know that, while some research strategies allow statistical generalisations to be made, others, like case-based research, only allow analytical generalisations.

Fourth, the unit of analysis varies considerably from one study to another. Some researchers have considered as their unit of analysis a single project, such as developing a new product or launching quality circles, which may be seen as part of wider strategic initiatives (e.g., Nutt, Reference Nutt1987, Reference Nutt1999; Voss, Reference Voss1988, Reference Voss1992; Park, Reference Park1991; Lewy & Mée, Reference Lewy and Mée1998a, Reference Lewy and Mée1998b; Hackett Group, 2004a, 2004b; Lawson, Stratton, & Hatch, Reference Lawson, Stratton and Hatch2006, Reference Lawson, Hatch and Desroches2008). Other researchers have focused on business-wide strategic initiatives, which may in turn be decomposed into several smaller projects (e.g., Kiechel, Reference Kiechel1982, Reference Kiechel1984; Harrigan, Reference Harrigan1988a, Reference Harrigan1988b, Reference Harrigan1988c; Mohrman et al., Reference Mohrman, Tenkasi, Lawler and Ledford1995; Walsh, Hughes, & Maddox, Reference Walsh, Hughes and Maddox2002; Pautler, Reference Pautler2003; McKinsey, Reference McKinsey2006; Sila, Reference Sila2007).

Fifth, some studies prove very difficult to obtain/access, in particular those undertaken by some management consulting firms such as A.T. Kearney, Arthur D. Little, McKinsey, Prospectus and Booz Allen Hamilton. Therefore, any conclusions taken from the estimates they have produced without a proper understanding of the context, methodology and results obtained might lack legitimacy and scientific rigour. In spite of this, it is common to find researchers (e.g., Holder & Walker, Reference Holder and Walker1993; Mintzberg, Reference Mintzberg1994: 25, 284; Smith, Tranfield, Foster, & Whittle, Reference Smith, Tranfield, Foster and Whittle1994; Zairi, Reference Zairi1995; Dow, Samson, & Ford, Reference Dow, Samson and Ford1999; Korukonda, Watson, & Rajkumar, Reference Korukonda, Watson and Rajkumar1999; Kaplan & Norton, Reference Kaplan and Norton2001: 1; Walsh, Hughes, & Maddox, Reference Walsh, Hughes and Maddox2002; Sterling, Reference Sterling2003) that quote the results of these studies not because they have read the original work but because these estimates have been quoted by other researchers or in well-known journals such as The Economist or The Wall Street Journal. Unfortunately, this has lead some of these studies to be widely misquoted and misunderstood (Taylor, Reference Taylor1997).

Finally, it is not always easy to distinguish what is fact and what is fiction in some of the estimates offered in the literature. In particular, there seems to be no scientific grounds behind some of the estimates. For example, Mintzberg (Reference Mintzberg1994: 25, 284), Kaplan & Norton (Reference Kaplan and Norton2001: 1), Burnes (Reference Burnes2004, Reference Burnes2005), Raps (Reference Raps2005) and Sila (Reference Sila2007) quote several sources for the rates of failure they mention in their papers (e.g., Kiechel, Reference Kiechel1982, Reference Kiechel1984; Judson, Reference Judson1991; Dooyoung, Kalinowski, & El-Enein, Reference Dooyoung, Kalinowski and El-Enein1998; Beer & Nohria, Reference Beer and Nohria2000; Waclawski, Reference Waclawski2002; Sirkin, Keenan, & Jackson, Reference Sirkin, Keenan and Jackson2005). However, a detailed analysis of these sources shows that they did not carry out an estimation of the quoted rates of failure. They claim their estimates were based on ‘Interviews’, ‘Studies’, ‘Experience’, ‘The Literature’ or ‘Popular Management Press’, rather than on solid empirical evidence. On other occasions, the sources of the estimates are incorrectly interpreted (e.g., Kaplan & Norton, Reference Kaplan and Norton2001, in interpreting the findings of Charan & Colvin, Reference Charan and Colvin1999). We also found evidence of studies incorrectly identifying their sources (e.g., Dyason & Kaye, Reference Dyason and Kaye1997) and studies not identifying their sources at all (e.g., Jantz & Kendall, Reference Jantz and Kendall1991; Neely & Bourne, Reference Neely and Bourne2000; Becer, Hage, McKenna, & Wilczynski, Reference Becer, Hage, McKenna and Wilczynski2007).

Unless these factors are accounted for, any attempts to present estimates for the real success/failure rates of strategy implementation are doomed to fail or are of little practical value.

Independently of the ‘real’ success/failure rate, and despite success rates that seem to have improved over time, it is reasonable to conclude that the number of strategic initiatives that fail is still considerably higher than would be desirable. This suggests that organisations either need better implementation guidelines or need to make better use of the existing ones. The need for better implementation processes has been widely acknowledged by researchers (e.g., Dean & Bowen, Reference Dean and Bowen1994; Mockler, Reference Mockler1995; Barney, Reference Barney2001; Hickson, Miller, & Wilson, Reference Hickson, Miller and Wilson2003) and research on how to avoid implementation obstacles and improve implementation has been underway for many years (e.g., Stanislao & Stanislao, Reference Stanislao and Stanislao1983; Alexander, Reference Alexander1985; Ansoff & McDonnell, Reference Ansoff and McDonnell1990; Kotter, Reference Kotter1995; Beer & Eisenstat, Reference Beer and Eisenstat2000; Miller, Wilson, & Hickson, Reference Miller, Wilson and Hickson2004; Stadler & Hinterhuber, Reference Stadler and Hinterhuber2005). It is, therefore, imperative to assess the extent to which these guidelines account for some of the improvements achieved as well as to understand the reasons why so many initiatives still fail.

Although efforts should be made to reduce failure rates, it is important to emphasise that failure can be seen as an important part of the strategic learning process within organisations (e.g., Mintzberg, Reference Mintzberg1987; Krogh & Vicari, Reference Krogh and Vicari1993; Sitkin, Sutcliffe, & Schroeder, Reference Sitkin, Sutcliffe and Schroeder1994; Edmondson, Reference Edmondson2011). Unintended past mistakes and deliberate strategic experiments can both generate useful lessons (Wilkinson & Mellahi, Reference Wilkinson and Mellahi2005), which may prove highly advantageous in the marketplace (Krogh & Vicari, Reference Krogh and Vicari1993).