On strategy, good and bad (mostly bad)

Good strategy is hard. I’m sure of that, but you don’t have to take it from me. Richard Rumelt makes that point repeatedly in his superb book, Good Strategy Bad Strategy. You’ve likely seen the consequences of leaders and companies who lacked good strategies. Maybe you’ve been called upon to create a team strategy or a technology strategy, or expect to in the future. In this post I’ll summarize Rumelt’s views on strategy and interpret them through my own experience.

Recently a member of my team had been writing a strategy document, for the ML Infrastructure team of which they are a leading member. And they’ve been a bit stuck. They had written a fairly long but incohesive document, and they asked for my advice with how to proceed.

Reading through their document, it contained plenty of insight. The author has so much relevant knowledge, including the technical depths of all systems in the team’s wide scope, strengths and weaknesses of the dependent and client teams that we collaborate with, and the direction and patterns of our engineering organization and product development as a whole. The document overflowed with the fruits of many lessons from years in this setting.

Yet the author knew that something was missing in the document, which was likewise clear to me when reading it. The author didn’t know which parts of the document to extend or create and which to shrink or discard. There was some wisdom, but no strategy; the author hadn’t laid out a direction for the team or any actions towards it. He hadn’t done so because he didn’t yet know what we should do nor what he wanted to accomplish with the document. He had a good handle on our situation: trends that would become important, our strengths and weaknesses, constraints we were likely to face, and many of the trade-offs and options for us to consider. He had a strong foundation, but still needed to craft a strategy out of that.

In discussing the purpose and construction of strategy with him, I sought out insightful and practical sources about strategy as a whole. That led me to finally reading Good Strategy Bad Strategy by Richard Rumelt.1 This book will henceforth be my first recommendation for people who want to better understand strategy.

Good Strategy Bad Strategy—or GSBS—is a superb book.2 In a short and immensely readable 300-odd pages, it lays out exactly what the title promises. Throughout its descriptions and entertaining examples it has utmost clarity of purpose. The book is very quotable because of its precise and excellent prose, yet is also unquotable in the sense that I could share it in its entirety because it contains very little waste. You might as well read the whole thing. I’m not one to tell anyone what to read or how much to read—there is no shortage of excellent writing out there and we all have our own preferences (including, for many, the preference to not read books at all)—but if you have some interest in evaluating strategy or crafting your own, you are likely to find Good Strategy Bad Strategy worthwhile.

In GSBS, strategy is “a cohesive response to an important challenge”. In the context of a business as a whole—which is the main setting in the book—the challenge could be about how to succeed financially as a business and earn customers in a competitive industry. If your setting and scope are, instead, a specific team, the challenge might be a specific technological hurdle, or it might be the need to choose between initiatives given finite people and time. In the absence of scarcity we just do everything we aim to do, in which case we don’t have or need a strategy. We might have plans or goals, but no strategy.3

Strategy is scarcity’s child

Rumelt provides a clear foundation of a strategy, which he calls the kernel. The kernel has:

• A diagnosis

• A guiding policy

• A set of coherent actions

You have to diagnose the scenario, pick an “overall approach to cope with or overcome the obstacles identified in the diagnosis”, and lay out an initial set of immediate actions that suit that approach. The diagnosis has to highlight the most essential challenges, simplifying away secondary concerns.

Rumelt points out that some views of strategy only encompass the first two bullets. He calls that a mistake. It’s a mistake that I’ve learned in my own career. It is always more useful to explain the first iteration of actions that you will take.4 If you don’t understand those, then you might not have considered the feasibility of your guiding policy and its ultimate effectiveness at tackling the challenge. Furthermore, since you will usually communicate your strategy, the coherent actions help explain the guiding policy and they “bring concepts down to earth”.

You might be like me, first taking the concept of strategy seriously through chess or other games, where we have complete knowledge (or at least clarity on which knowledge is known and which is not) and a relatively discrete action space. In the settings in GSBS, the action space is much more complex. There might be too many permutations to thoroughly evaluate. Rather than seeing strategy as the act of choosing from a known set of options, Rumelt instead equally privileges the search process through options. Strategy is a process of design. Design in the broad sense of understanding the needs and crafting a solution with elements that must be “arranged, adjusted, and coordinated”.

Many effective strategies are more designs than decisions—are more constructed than chosen

Furthermore, a strategy must have some simplicity to it. It cannot be a rule so complex or a plan so exhaustive that it cannot be understood and implemented by everyone involved. So not only does the strategy setter have to design a feasible direction out of a multitude of options, but they also have to choose which complexities to sacrifice for clarity. They must “absorb a large part of that complexity and ambiguity, passing on to the organization a simpler problem—one that is solvable”.

People don’t always succeed in creating a good strategy. Not only do they not end up with a good strategy, but they create something else that is recognizable in its bad patterns. Rumelt calls this “bad strategy”.

The elements of bad strategy are:

• Fluff

• Failure to face the challenge

• Mistaking goals for strategy

• Bad strategic objectives

While good strategy is usually clear, stripped of any unnecessary complexity or distractions, bad strategy has the fluff of “gibberish masquerading as strategy concepts or arguments”. Bad strategy might entirely fail to address the essential challenge of the situation. Instead, it might have goals that are “just statements of desire rather than plans for overcoming obstacles”, and the objectives might not only fail to face the challenge but also be infeasible.5

You’ve all seen it. Ambitious goals not backed by clear steps to achieve them. Absence of logic—I mean this not cruelly, to call the logic poor, but instead to describe cases where logical reasoning wasn’t part of the pitch at all—about choosing a direction over another. Wrapped up in metaphors. Slides that show positivity and empty slogans but no explanations.

Rumelt highlights three leading reasons why people can choose to create bad strategy. They might be “unwilling or unable to make choices among competing values”, or following a “template-style” exercise conducted without inspiration, or they might fall into the trap of assuming that success follows naturally from “a positive mental attitude”.

There is difficult psychological, political, and organizational work in saying “no” to whole worlds of hopes, dreams, and aspirations

I’ve seen every type of bad strategy. Sometimes leaders, rightly or wrongly, decide to postpone the major changes needed to face their central challenge.6 Sometimes the fluff of a bad strategy intentionally hides a lack of agency; the business might be in stasis or slow decline without any real advantages over their competition.7 There might not be much of a need for good strategy: a naturally strong position “can obviate the need for sophisticated design-type strategy”, or in absence of a great opportunity we should wait and “seek coordinated policies only when the gains are very large”. Fluff instead of good strategy can be a morale exercise.

Even when a good strategy is needed, which is often, many people aren’t equipped to find one. You first need a clear diagnosis, understanding your setting. Consider my motivating example of a colleague who had deep domain knowledge, insights earned from extensive experience, and a thorough understanding of the mandates and personnel of other teams. Even then it was hard enough to form a strategy. Imagine trying in the absence of those advantages. Many people just don’t know enough about their domain to diagnose their essential challenge and know their options. They might also lack the design creativity—along with the aforementioned domain knowledge—to construct a guiding policy and a set of coherent actions. So when forced to write a strategy, they write something vacuous.

A blue-sky objective is usually a simple restatement of the desired state of affairs or of the challenge. It skips over the annoying fact that no one has a clue as to how to get there.

Rumelt is critical of template methods along the lines of “vision, mission, values, and strategies”. Visions are often infeasible, missions and values might be empty platitudes or flat-out deceptive, and a good strategy does not inherently follow from those. Good strategy is not so easy as to be found through a simple formula. It takes deep understanding and analysis.

GSBS mostly talks at the scope of business strategy, but most of us will try to create strategies at smaller granularities. A team’s strategy, a product strategy, a hiring strategy, a tech debt strategy, a cost strategy. Those can be useful as long as there is enough scope for strategy in the area. But sometimes the granularities are too small, with no practical constraints or not actually containing important challenges. Sometimes the scope is downstream and reactive to something else more important. The tail doesn’t wag the dog.

We’ve covered a couple of prerequisites for developing good strategy: deep domain knowledge and a capacity for design. You need the former to know what is happening and what is possible. The latter is necessary for constructing a good direction and path forward out of the innumerable possibilities. You have to be creative with combining choices together.

You need taste.

Being strategic is being less myopic—less shortsighted—than others. […] Being “strategic” largely means being less myopic than your undeliberative self.

What can we do to be better at strategy? Rumelt suggests techniques for improving our judgement. He breaks it down into three skills:

• Avoiding myopia

• Questioning your judgements

• Practicing judgement

These remind me of techniques that I apply and advice I give regularly.

With written proposals (think: RFCs, PRDs, slideshows, etc.), I want to see a strong alternative approach investigated clearly and honestly, steelmanned instead of strawmanned. That’s one of the significant advantages I’ve seen strong engineers have over others, that occasionally they find a better approach that other people would have disregarded entirely. I try to convince my team that even if their initial judgements are right most of the time, it will eventually pay off to go through the exercise of really investigating an alternative.

After writing out my ideas, I ask myself “if I’m wrong here, where did I go wrong?” or the more forceful “I’m wrong; what did I miss?”. I like a good pre-mortem for critical projects. Rumelt writes that “trying to destroy your own ideas is not easy or pleasant”, but I have to say that I enjoy it. I know I’m not infallible. Being the first and harshest critic of my own ideas is how I form better ones and improve my reasoning.8 It’s all part of the process for me.

Rumelt’s advice on improving judgement is to write down your opinions, because that grants “the opportunity to pre-commit to a position and thereby subsequently evaluate [your] own judgement”. I couldn’t agree more. It isn’t enough to form an opinion in your mind, because you will fool yourself later by selectively remembering your opinion and how much you weighed different elements of it. You have to record your judgement so that you can check it accurately later. I regularly encourage my team to write proposals and not to stop at listing out options. They must weigh those options and recommend a path forward, and then test that view through feedback from informed colleagues. I don’t penalize people for needing to revise that path later; instead they’re rewarded by the success of getting more situations right in the end.

If you make the most of every opportunity to apply judgement, you will get better at it.

If you read more on the topic, you’ll quickly notice that there are many different definitions of strategy. None of them are objectively right or wrong, because strategy is just an abstract concept. Definitions of strategy might be influenced by military history, chess, tabletop or video games, or business case studies. Strategy could mean anything we choose it to mean. Nobody owns the term.

We might as well choose definitions of strategy that are useful. That’s what Rumelt has done: he’s picked a definition of strategy that makes it a useful framework for people who need to make choices. He’s created real boundaries (“to give content to a concept one has to draw lines, marking off what it denotes and what it does not”) and elaborated extensively on what isn’t strategy or isn’t good strategy.

I have often tried to back out an implicit strategy from observing the behavior of a person, team, or company. Even if there is no intentional strategy, is there an underlying implicit strategy? What if there is an intentional choice to not have an explicit strategy, is that itself a strategy? What if there is an explicit and intelligent strategy, but it’s just the aggregation of natural optimizing behavior of each unit of the business? I have wrestled with those riddles. I believe Rumelt in GSBS would say “No” to all of those, or would call the outputs “bad strategy”. He chooses to define strategy in a way that is less philosophically challenging and more practically distinguishable.

Regarding the latter of those riddles, Rumelt explicitly clarifies that strategy must be “imposed”, that it is “an exercise in centralized power, used to overcome the natural workings of a system”. Strategy is not a summary of the firm’s natural and reasonable behavior, but rather an explicit choice and enforcement of direction.9

An important duty of any leader is to absorb a large part of that complexity and ambiguity, passing on to the organization a simpler problem—one that is solvable.

As you increase in seniority in your career, you will more often need to set a direction from among many unenumerated options. You will have to form opinions, write documents, convince others, and explain how to act on those views.

Strategic thinking is a core skill in leadership roles, a skill that you can improve through dedicated practice.