Contents

Introduction

Change is hard. Successful corporate initiateves are hard. A commonly quoted figure is that 70% of such initiatives fail.^[1] McKinsey appositely calls it, “Losing from day one,” and cites data that will leave the reader depressed.^[2]

In this white paper, we will first review some of the commonly cited reasons for the frequent failure of corporate initiatives, and why—even when they succeed—they are often misguided from the outset.

Next, we return to first principles and examine how corporate initiatives are inherently tied to solving problems that:

• Are hard – worse than hard, as we shall see!
• Modifying the corporation – although systems fight back against attemps to modify them

We then draw conclusions and show how the most commonly cited reasons for failure are, in fact, epiphenomena, i.e. surface-level symptoms of deeper dynamics.

Along the way, we will briefly touch upon potential solutions. However, the scope of this white paper allows only for two illustrative examples.

Current state of knowledge

The most frequently cited reasons for failure fall into the following categories:^[3]

• Lack of a clear strategy and/or execution plan
• Neglecting the human factor (communication, clarity, motivation, culture, fear of consequences, etc.)
• Weak governance (inadequate structures and processes)
• Lack of focus and/or insufficient resources—time, staff, expertise, funding, or leadership commitment. This may arise from an overload of initiatives or from a lack of genuine commitment to the initiative in the first place.

BCG takes a notable approach by combining pragmatism with wisdom.^[4] They emphasize:

• Timing matters:

  • Transformations should begin from a position of strength, not desperation
  • Long-term orientation is essential

• Cost-cutting alone does not create high-performing companies

• Effective transformation requires strong governance and process

McKinsey, similarly pragmatic, notes the following:^[5]

• There are no shortcuts: “the main differentiator between success and failure was not whether an organization followed a specific subset of actions but rather how many actions it took.”
• Even successful initiatives often fail to realize much of their full potential
• About 20% of this lost value occurs “after implementation, once the initiatives have been fully executed.”

McKinsey further highlights the importance of a fact-based assessment, the need to clearly articulate goals across all organizational levels, and the value of assigning top performers to lead initiatives.

Discussion of the currently accepted wisdom

Achievements

At first glance, the reasons for failure and the recommended solutions seem well understood.

Most of the observations and advice regarding successful corporate initiatives are valid. The problems identified do help explain failures, and an organization that applies these lessons should, in principle, improve its odds of success.

Much has already been achieved, and some contributions (as noted earlier) offer genuinely fresh insights.

Gaps

But if the issues are so well understood, why do corporations continue to fail? Are they simply unwilling to succeed?

Here again, McKinsey’s findings speak volumes:^[6]
Our survey results indicate that companies’ transformation efforts remain stuck. The 30 percent success rate hasn’t budged after many years of research.

Worse still, the report adds: We now know that even successful transformations still leave value on the table.

The central critique of current wisdom, then, is simple: it does not fix the problem.

Synthesis

Going beyond the pros and cons, the state of the field can be synthesized as follows:

• The achievements thus far are impressive
• The lack of robust, peer-reviewed research might seem like a weakness—yet we will see that this is actually a consequence of the systemic nature of the problem.^[7]
• The prevailing recommendations, while insightful, have not succeeded in improving the overall success rate
• Therefore, something fundamental is missing

What is missing is the topic of the rest of this white paper.

Returning to the basics: wickedness and systems

Let us take a step back. Corporate initiatives have two essential features:

Firstly, the problems they seek to address are wicked and imponderable.

Secondly, the target of these initiatives—the organization—is a system, and thus greater than the sum of its parts.

We will show that these features explain the patterns already observed in the literature. That is, the observations are merely surface effects—epiphenomena—of deeper dynamics.

Wickedness and its consequences

Description

The problems targeted by corporate initiatives are wicked and imponderable:^[8]

• Wickedness:

  • Incomplete, contradictory, and shifting requirements; even defining the problem is a challenge
  • Time-sensitive
  • One-shot in nature

• Imponderability: There are many wrong solutions, but, by nature of the problem, it is difficult to define what a “right” solution would be. The problem is fuzzy.

For example consider the problem of how to respond to a competitor’s new product launch:

• Wickedness:
  • Even defining what we seek to achieve is a challenge
  • There are more unknowns that knowns
  • The situation shifts, sometimes day-by-day
  • The firm only gets one shot at solving the problem
    -* Imponderability:
  • There are many courses of action that the firm can take.
  • Some of them are clearly wrong
  • There is no meaningful sense in which there is a right solution. To verify this, try defining what the right solution might be. A finance-based definition such as profit maximization (within legal, societal, and ethical bounds) leaves us rudderless until it has been operationalized. How to use it to guide decision-making? After the fact, how to use it to determine whether the action already taken was the right one? Even if there were such a thing as a “right” solution, it would not even matter. The firm will never know if it chose it.

Any attempt to define the right solution is highly theoretical and not much use in practice.^[9]

Although related, wickedness and imponderability are distinct.

Typical sources of imponderability include:

• Known unknowns: Influential factors whose outcomes are uncertain (e.g., whether a client’s industry will enter a recession)
• Unknown unknowns: Influential factors whose very existence is unforeseen (e.g., the 2019 Covid outbreak, which might possibly have been predictable but whose consequences were unpredictable.)
• Chaos: Systems that are deterministic yet, in practice, unpredictable (e.g., weather, dice, the economy)
• Dilemmas: Situations where every option has critical downsides
• Problem formulation: When even defining the issue is problematic (an aspect of wickedness)

Consequences and explanatory power

The most pressing and valuable problems organizations face exhibit the characteristics above.

Consider the consequences of the following on enterprises’ ability to identify answers to its problems/opportunities:

• There are many wrong answers to the organizations’ problems/opportunities, and if right, an answer is only relatively right
• Initiatives are time-sensitive and one-shot
• The firm does not know what the correct problem to address is:
  • Yet it must focus its resources
  • Yet it must act. (Failing to act is also an action.)

In this context, recall the following points made by the common wisdom.

Given that corporate problems are wicked and imponderable, and that organizations are complex systems with homeostasis and non-linear effects, is it surprising that:

• Lack of a clear strategy and/or execution plan is a common cause of failure?
• Lack of focus easily arises, and frequently derails initiatives?
• Even successful companies struggle to capture the full potential value of their initiatives?
• A fact-based assessment is crucial to ground these fuzzy, wicked challenges?
• Change is hard? And even addressing the above points has not improved the success rate of corporate initiatives?

Anticipating: what to do about it?

We will see below and in the examples that high-impact leverage points can be found, which multiply the impact of an intervention. This is characteristic of systems.

At this point, the situation may appear bleak. No wonder so few initiatives succeed. And yet…

Anticipating on the below, we shall see that the way forward lies in recognizing the systemic nature of organizations. By doing so, we unlock the potential for interventions that yield outsized returns, which in turn give room for maneuver to unlock further outsized returns.

The enterprise is a system—and the resulting consequences

Description

A system is a set of interconnected or interdependent elements working together to achieve a common goal.

By definition, it is more than the sum of its parts.

Take, for example, a house. It is made up of various materials, but its structure transforms it into something with the functional purpose of shelter.^[10]

Now, consider the enterprise.

It is a nested system of systems. It comprises people—each a biological system—who are embedded in departments and process flows, both formal and informal.

These departments are part of broader structures, forming the organization, which in turn belongs to larger systems like:

• Supply chains
• The company’s governance framework: board (with its politics), shareholders, stock marekets, etc.
• The regulatory framework and its actors, in the case of regulated industries
• The broader economy, whether national or international
• The broader society, via customer preferences and perception, public scrutiny, etc.

Consequences and explanatory power

Consequences

This white paper cannot explore every implication of systems thinking. However, let us highlight a few key points.

First, systems have goals. These may differ from the intentions of leadership or designers. Examples:

• A for-profit company functioned, in practice, to offer lavish salaries to executives, even at the cost of driving the business toward bankruptcy.
• A humanitarian NGO focused more on pleasing donors than helping beneficiaries.

Second, systems have components which are themselves often systems—individuals, departments, networks—each with their own goals. These goals can conflict with the system’s overall purpose.^[11]

Third, systems have interconnections. Any intervention causes cascading effects. As a result, interventions often end up achieving effects that are different, and often even contradictory^[12] to the original intent.

Among the effects that make it difficult to plan and foresee the effects of an intervention are:

• These cascading ripple effects are often non-linear. As a result, doubling effort does not double results. For instance, in the case of a cut-off feedback loop, doubling effort may well completely switch off the intended effect.

• Chaos: non-linear effects characterized by “stretching and mixing”^[13] often exhibit chaos, rendering the effects unpredictable.

These feedback loops are a key reason why interventions are wicked problems.

Fourthly, systems have homeostasis. Any non-trivial system will necessarily have some degree of stability. A system that is completely unstable will rapidly cease to exist. This stability implies that the internal relations of the system are set up in such a way as to protect the system’s status quo. This is called homeostasis—and is more informally referred to as resistance to change.

There are other features of systems that have just as powerful an impact as the above examples; however their treatment goes beyond the scope of this white paper.

Explanatory power

These system features directly explain the failure patterns observed in practice and highlighted by the common wisdom.

Indeed, given the systemic nature of organizations—with their cascading effects, homeostasis, human subsystems, resistance to change, etc.—is it surprising that:

• Ignoring the human element often leads to failure?
• Lack of governance (structure and processes) contribute to failure?^[14]
• Lack of focus easily arises, threatening success?
• Cost-cutting alone, which treats the firm as if it were the sum of its parts, does not lead to success?
• Sustaining the improvements is hard, and much value loss occurs after implementation, as the organization seeks to return to homeostasis?

The existing advice addresses symptoms but not root causes. This explains why success rates remain stubbornly low despite decades of research.

Anticipating: what to do about it?

At first glance, systems seem to complicate everything. But that’s only if we treat them as collections of parts, forgetting that it is a system.

If we instead engage with them as systems, they reveal leverage points—small, high-impact areas for intervention.

Whereas in linear systems we might expect a proportional return on effort, systems offer multiplier effects.

For example, many systems have feedback loops. These often act to stabilize the system, but the seasoned practitioner can often use them to amplify positive change.

We shall illustrate this with examples.

Examples of how to make initiatives be successful

Basic case study (as a warm-up)

The situation

Company X was rapidly growing. Its staff were overworked to the point of collapse, and despite this were barely managing to get the necessary minimum done. As a result, the company was hiring as fast as possible, but the problems continued to get worse.

Initial (failed) solutions

Naturally, the company sought to fix the situation by hiring even more for all of the overworked departments. This is a typical quantitative solution to a problem that is, in essence, not quantitative at all, but a systems problem.

Of course, resources being scarce, the hiring drive focused on the really “important” departments, namely, the production-related departments. The less important support departments, such as HR, received no resources for hiring.

Mysteriously, the situation only got worse.

The solution

When we health-checked the company, it turned out that there was a bottleneck if one of the “minor” (as this company saw it) departments. The HR department had not grown fast enough to keep up with the needs of the organization.

As a result, HR-related problems were flooding the organization, disrupting work:

• Hiring was delayed and disrupted for departments that had urgent needs for more staff and missing skillsets
• HR-related tasks such as handling vacation timing and the minutiae of hiring were ending up falling on the managers of all the other departnents. This resulted in headaches, interruptions and the loss of valuable time of already over-worked individuals in vital positions.

This in turn caused technical debt to accumulate in all departments, hiding the root cause. Now, it seemed that the problem was that the entire organization was overworked. It was no longer apparent that the company-wide overload and technical debt had its roots in the HR department being overworked.

The situation of overwork of each department itself caused extra work for downstream departments, via several mecahnisms such as the following:

• Department A is late finishing a task, due to being overwokred
• Consequently, the downstream Department B, which handles the next step, has to wait. For a limited amount of time, it had less to work on than its full capacity. Paradoxically, this is a frequent phenomenon in overworked organizations!
• Idle staff reflects badly on a department head, so busy-work was created for Department B. This busy-work required input from other departments, resulting in further work overload for the other departments
• Once Department B receives the work from Department A, it is even more late on its work and overworked than it otherwise would have been, due to the time lost waiting for Department A
• To make things worse, the tasks arriving from upstream into Department A, tended to clump, with multiple tasks arriving at some periods and very few at others.^[15]

In parallel to all of this, given that the root cause was an overworked HR department, the attempted solution of increased hiring made things even worse… which in turn increased pressure to hire more!^[16]

Once the true root cause was identified, the fix also became clear: break the vicious cycle, by fixing the indigestion. The firm’s project up-take would be temporarily limited to what HR could handle.

This strategy is not a strategy of reducing revenue. On the contrary, once the indigestion was resolved, and the various issues described above disappeared, much capacity would be freed up. The organization’s ability to take on projects would end up being much higher than in the current situation.

This approach was tested via a pilot project. Once validated, the following was decided:

• Include HR in decision-making meetings about accepting new projects
• Temporarily put the manager who was next-in-line for the CEO position in charge of the HR department, at 25% of his time. Failure in this role would imperil his chances of becoming CEO.
• Set up a Project-Management Office driven by the CEO to track progress

The HR department now had the ability to limit the uptake of new projects to the volume that it could handle. As a result, HR-related tasks stopped flooding the other departments. This, combined with the temporary reduction of projects, allowed the various departments to catch up on technical debt, and sort out all of the above-described issues that were limiting capacity.

As a result, the organization ended up able to handle a much larger volume of projects, without the cross-organization overwork and exhaustion.

This example illustrates that, although systems make it harder to achieve corporate initiatives, they also introduce opportunities: leverage points where a small interventions can lead to outsized results.

More sophisticated case study

The situation

Company X is a niche bank. Forgetting that the enterprise is a system, this bank went on a cost-cutting spree and, among other things, fired a significant proportion of its tellers.

When its net income turned negative, it realized—alas, too late!—that these tellers were, in fact, necessary. An internal investigation revealed that:

• The negative net income was caused by a drop in revenue
• There was no apparent drop in number of customers (footfalls)
• However, there were long lines at the branches. The current number of tellers were not managing to serve customer demand.

It was further to be expected that, as customers realized that they would not be served in a timely manner, they would switch to the competition.

Rehiring the tellers was not an option. Those that had been let go had, mostly, already found jobs. As for hiring completely new staff, that was not a workable solution either. Indeed, the tellers who had been let go had a valuable qualification: they knew how to run the bank’s quircky, buggy and complex legacy teller IT interface. It would certainly have been possible to train new staff; however it was not feasible to hire and train them in sufficient numbers within a sufficiently shoft time frame to avoid massive loss of customers.

Initial (failed) solutions

Given the situation, the bank decided to simultaneously re-engineer and automate the teller process. They purchased expensive equipment for this purpose.

However, there was no improvement.

They brought in a consultancy to assess the situation. The consultancy:

• Found that the new machines were barely being used
• Concluded that the issue was resistance to change

On this basis, the bank implemented various change management initiatives, complete with communication, incentives, governance structures, etc. to encourage tellers to use the machines.

This failed. Although there was a marked increase in machine usage, processing time actually went up, and the number of customers that tellers could serve went down.

The interested reader may wish to pause for a moment and consider what the solution might be, both before and after reading the following complementary information.

The process in question is a two-step process, as outlined in the diagram below.

graph TD
    A["Customer Arrival"] --> B1["Step 1: Machine <br> (45s)"]
    B1 --> C1["Step 2: Teller <br> (Vol/100)"] --> D["End of transaction"]
    A --> B2["Step 1: Teller <br> (Vol./100)"]
    B2 --> C2["Step 2: Machine <br> (45s)"] --> D
    B2 --> C1

The new machine can do both steps, and takes 45 seconds for each step. However, for regulatory reasons, there must be at least one human touchpoint in the transaction. Thus, step one can be done by a machine and step two by a teller, or vice versa, or both steps can be done by a teller. But it is not possible for both steps to be completed by a machine.

As for the processing speed of a teller, it depends on customer volume. As a rule of thumb: given a certain monthly branch volume, at current staffing levels, the teller’s speed (in seconds) was found to be roughly the monthly branch volume divided by 100.

The solution

The solution that resolved this problem can be found at the end of this white paper, in the section entitled Spoiler.

Spoiler: solution to the second case study

Please read the second case study before reading this section.

This situation illustrates the kind of paradoxes that can arise because the firm is a system composed of subsystems.

To understand what is going on, let us assume for the sake of argument that once the machines are purchased and the processes re-engineered, the tellers are fully on board with the initiative and want to use the machines as much as possible. Zero change resistance.

Under this rosy assumption, surely the initiative should be crowned with success—or should it?

Let us check. In this rosy scenario, the initial situation is as follows:

graph TD
    A["Customer Arrival"] --2,000 transactions--> B1["Step 1: Machine <br> (45s)"]
    B1 --> C1["Step 2: Teller <br> (2000/100=20s)"] --> D["End of transaction"]
    A --2,000 transactions--> B2["Step 1: Teller <br> (2000/100=20s)"]
    B2 --> C2["Step 2: Machine <br> (45s)"] --> D

Thus, each transaction takes 65s to process:

• 45 seconds on the machine
• 20 seconds with the teller

This represents a significant improvement over the 80s that were needed prior to the purchase of the machines.

However, let us further assume that the line supervisors are smart and competent. This assumption is not so unlikely, especially since we’re describing a rosy scenario.

Being smart and competent, they observe that the teller’s volume of transactions is quite low, and that as a result, they are currently faster than the machine. At current volumes, full manual processing would take just 40 seconds. Thus, the process can be sped up by fully manually handling some transactions. After shifting 400 transactions / month to full manual, and rebalancing, this gives:

graph TD
    A["Customer Arrival"] --1,800 transactions--> B1["Step 1: Machine <br> (45s)"]
    B1 --> C1["Step 2: Teller <br> (2200/100=22s)"] --> D["End of transaction"]
    A --2,200 transactions--> B2["Step 1: Teller <br> (2200/100=22s)"]
    B2 --> C2["Step 2: Machine <br> (45s)"] --> D
    B2 --400--> C1

Note how:

• Each of the manual steps takes 2 seconds longer
• The switch poisons the old paths: by adding 2s to each manual step, it has added 2s to each of the semi-manual paths. These now take 45+22=67s.
• The fully manual path takes 22+22=44s, which is an extra 4s longer than before. This is still much faster than the semi-manual paths.

Thus, even now, the fully manual path remains more attractive. A smart supervisor will want to shift more transactions over to the fully manual path.

Let us take a step back. What we see is that each batch of 400 shifted transactions increases processing time in the fully manual path by 4 seconds and increases the semi-automated paths by 2 seconds.

We can immediately grasp that there is a point beyond which further shifting is counterproductive. The manual path ceases to be faster.

What is easy to miss, however, is that before reaching that point, the original (semi-manual semi-automated) paths may already have been poisoned so much that all paths are now slower than the original setup.

This is exactly what happens in our current (rosy) scenario. Shifting transactions over to the manual path turns out to always be more attractive, until there are no transactions left to shift:

graph TD
    A["Customer Arrival"] --0 transactions--> B1["Step 1: Machine <br> (45s)"]
    B1 --> C1["Step 2: Teller <br> (4000/100=40s)"] --> D["End of transaction"]
    A --4,000 transactions--> B2["Step 1: Teller <br> (4000/100=40s)"]
    B2 --> C2["Step 2: Machine <br> (45s)"] --> D
    B2 --4,000--> C1

But wait! The result of all this optimization is that now the processing is taking 80s, instead of the initial 65s!

The paradox lies in that:

• The process has been optimized, or so it seemed
• Yet the current processing time has worsened to 40+40=80s, the same level as before automation
• And yet if a transaction were shifted over to one of the semi-automated paths, it would take even longer: 45+40=85s

Thus, now we have a situation where:

• All transactions handled manually, losing the benefits of purchasing the machines
• Still, no supervisor would shift traffic back to the machines, because the semi-automated route now takes 45 seconds for the machine plus 40 seconds for the teller—totaling 85 seconds, which is worse than the 80s fully manual route.

This is a classic case of systems behavior: an apparently beneficial intervention results in unforeseen effects. (This paradox also arises in traffic control, where it is referred to as Braess’ paradox.)

The solution was to:

• Congratulate the supervisors for proactively seeking improvements;
• Explain the paradox to them; and
• Forbid any further use of the fully manual path.

Doing so led to a successful acceleration of the process, bringing the average processing time down from 80 seconds to 65 seconds—a nearly 20% improvement.

This improvement translated directly into a revenue increase (although slightly less in terms of percentage, since the bank had other income sources as well).

Due to operating leverage, this led to a much larger increase in net income—far beyond what a simple ~20% improvement in process efficiency would suggest.

This example illustrates our point that, while systems make it harder to implement corporate initiatives, they also create opportunities. Small interventions can lead to outsized results.

Appendix: some examples of the literature on failure of corporate initiatives

Various articles published by Harvard Business Review cite quite diverse factors. For instance, Kotter (1995) cites:

• Insufficient sense of urgency
• Lack of a powerful guiding coalition
• Vision that is too complicated or too vague
• Undercommunicating
• Behaving in ways that are antithetical
• Insufficient planning
• Not getting quick wins
• Not enforcing new norms
• Not promoting people in line with the initiative

As for Deuschel et al (2024), they cite the following key causes of failure:
“1. First, they don’t clearly define the type of change their organization or departments are facing, which means they can’t clearly define their specific objectives.
2. Second, they often apply a one-size-fits-all strategy to their situation and don’t assess what sort of balance it calls for between exploration (innovating and risk-taking) and exploitation (optimizing existing processes).
3. And third, they don’t consider who will drive the change and how employees’ differing skills and readiness for change will affect success.“

Further perspectives from BCG and from McKinsey are described in the body of the text.

References

Beer, M., & Nohria, N. (2000). Cracking the code of change. Harvard Business Review, 78(3), 133–141.

Bucey, M., Schaninger, B., VanAkin, K., Weddle, B. (2021). Losing from day one: Why even successful transformations fall short. McKinsey & Company (People & Organizational Performance and Transformation Practice, online)

Deuschel, N. T., Langan, R., & Rodríguez Gómez-Rico, C. (2024, December 9). 3 reasons change initiatives fail — and how to ensure yours succeeds. Harvard Business Review (online)

Kotter, J. P. (1995). Leading change: Why transformation efforts fail. Harvard Business Review, 73(2), 59–67.

Meehl, P. E. (1967). Theory-Testing in Psychology and Physics: A Methodological Paradox. Philosophy of Science, Volume 34 , Issue 2 , June 1967 , pp. 103 – 115.

Meehl, P. E. (1992). Theoretical risks and tabular asterisks: Sir Karl, Sir Ronald, and the slow progress of soft psychology. In R. B. Miller (Ed.), The restoration of dialogue: Readings in the philosophy of clinical psychology (pp. 523–555). American Psychological Association.

Reeves, M., Gruss, C., Ellmer, K., Job, A., Bouslov, G., Catchlove, P. (April 2024). Five Truths (and One Lie) About Corporate Transformation. BCG (online). This study is an extension of BCG’s earlier work, “ The Truth About Corporate
Transformation,” published in MIT Sloan Management Review.

Stewart, I. (2002). Does God play dice? : the new mathematics of chaos (2nd ed.). Blackwell.

Footnotes