Crafting Non-Predictive Strategy, Part I: Deep Understanding Beats Prediction

As Milo and I have argued before, the environments and issues businesses deal with are more complex than traditional strategy models admit.   Business issues today display high levels of uncertainty, they can behave non-linearly, and they can be vulnerable to “Black swans”, i.e. low-probability but high impact events that disrupt even the best formulated strategies.  The added difficulty for strategists and managers is that nonlinear environments often appear linear for an extended time period (think US house prices).  As a result, some conclude that what seems to be an essentially linear pattern (prices fluctuate a bit around a ‘long term trend’ but always rise), are linear in reality – before a radical change occurs that completely disrupts previously assumed patterns (e.g. prices fall dramatically).  In short, people often assume an environment is linear and predictable when in fact the continuity we observe is only a particular case of limited duration.  To make matters worse, with many nonlinear systems change is not nicely spread over the years:  most of the cumulative change occurs in one, single – often dramatic – occurrence.  In the language of engineering, some things don’t “fail gracefully” (e.g. a bridge that breaks suddenly instead of bending slowly).

Not a “graceful failure”.

This generates a question:  How can you craft strategy in nonlinear environment?  One (traditional?) approach is to simply ignore or deny the nature of the environment.  Instead, we suggest that you first begin by considering the dimension of time. 

Most of the approaches that have tried to deal with an uncertain future have directed their effort towards making better predictions. Some have assumed that key to successfully dealing with uncertainty is to take a deep dive into the long-term future.  Even when the difficulty of prediction is acknowledged, effort is still devoted to imagining alternative possible futures.  This is the case of scenario planning, one of the most popular of such approaches. When creating scenarios, for instance, oil executives are asked to think about a world where a barrel is $300 and a world where it is $13.  There is no denying that imagining possible futures is useful and interesting, but the experience with scenario planning is that even when several alternative futures are boldly imagined, the actual one is rarely one of them.  This is not surprising: the number of possible scenarios quickly grows exponentially as we look farther ahead in months and years, so why would three or four scenarios bring anything useful?  There is, in fact, a fundamental flaw in such approaches: it assumes that we can somehow successfully imagine the central aspects of the future.  There is ample evidence, however, that this is not true.  We do a terrible job imagining the future, sometimes with dreadful consequences, and scenario planning (developed commercially beginning forty years ago at Shell) hasn’t helped us much.

Other approaches have aimed at delineating key uncertainties about the environment.  These approaches are based on the idea that not all a firm’s environment is necessarily uncertain.  For instance, a firm trying to sell an existing technology in a new market will face market uncertainty, but perhaps not technological uncertainty.  This selective approach to prediction was pioneered by Paul Saffo.  To him, the strategist’s modest goal should be to create effective, rather than accurate, forecasts.  Saffo says effective forecasts need not be completely accurate, but should “define the cone of uncertainty”, i.e. they should effectively “delineate the possibilities that extend from a particular moment or event,”  and “tell you what you need to know to take meaningful action in the present.” Such efforts are sometimes assimilated to a mapping exercise, but the map metaphor is dangerous since maps describe a physical territory that doesn’t change, whereas the business environment changes constantly.  So again, these techniques are useful in a limited way, because they boil down to narrower, more humble prediction.  But, smart or not smart, prediction doesn’t work in truly nonlinear environments.

To return to the question:  How can you craft strategy in nonlinear environment?  Milo and I argue that instead of putting effort into better prediction (no matter how modest), in many cases strategists must  take the opposite approach and learn to focus their effort purely on a better understanding of the present.  In so doing, they shrink their time horizon to a very short span in the near future, and reflect on the likely consequences of their action.   In other words, strategists should concentrate almost wholly on the present environment, on what they can shape now, and on the likely consequences of their immediate actions; they should  reject efforts to peer into an abstract future that may arrive no matter what they do.  They should also build the uncertain nature of their envelopment into the heart of what they do.

This can be done in two ways. First, strategists must focus on mitigating the impact of surprises.  We are not able to predict when certain events will occur, but we can prepare for their occurrence, and take steps to reduce their impact should they occur.  The approach explicitly values prudence, even when it does not seem “optimal”:  raise a coastal nuclear power plant well above sea level, or spread key factories across several continents.  Obviously this approach works only for a certain class of “already-seen” events:  an earthquake, for instance.  It will not work for entirely new events, however.  On the other hand, even highly disruptive events are rarely entirely unique:  the invention of the Internet, for instance, is unique in many ways, but still it is the invention of a new communication technology, and the telegraph was a kind of Victorian internet.  But with nonlinear systems, one is bound to find that even small differences can have huge impacts, so the mitigation of impact, though it should be used, has only limited value.

There is always a precedent or analogue – the Victorian Internet

Second, strategists must focus on anticipating the consequences of their own actions.  To do this, they must first master the present.  This is harder than it sounds, and it is surprising how this fairly basic point gets ignored in the heat of the strategic moment.  There is a tendency to rush to action.  In dealing with a nonlinear issue, one should, on the contrary, slow down and take the time to understand the situation.  In depth.  Historians and political scientists Neustadt and May give a striking example of why this matters when they recount how US President Jimmy Carter was called back urgently from his vacations in the summer of 1979 by the news that an elite Soviet commando had been spotted in Cuba. What where the Soviets up to?  US forces were put in alert and experts were assembled in a War room… until someone reminded the participants that there had always been an elite Soviet commando in Cuba, and that this had been agreed upon by J. F. Kennedy in 1962.  In short, the problem did not exist.  A puzzled Russian ambassador told his US counterpart: “How am I supposed to explain this to my government?”

One tool that Milo and I developed for strategists to think in detail about the present – in other words to answer the pretty basic strategic question “What is going on?” –  is a refinement of Neustadt and May’s work.  We call it the “KPUU framework”.  It demands strategists answer and get agreement about four simple questions about the present:  What do we Know (including how did this issue begin)?  What do we Presume? What is Unknown (but could perhaps be discovered by finding the right person or source), and what is essentially Unknowable (e.g. consumer acceptance of chemically-enhanced language learning)?  An open debate about what data goes in each column – especially what is Unknown versus what is simply Unknowable at this moment – uncovers a huge number of assumptions and also exposes strategists’ differing rules of evidence.  This effort to understand more deeply the present is, in our view, more valuable than most efforts to plumb the depths of uncertain futures.

The KPUU Framework – Don’t Take Action Without it

Once the KPUU framework creates an explicit, shared inventory of knowledge about the present, including embedded assumptions, then the question of what it all means (implications) can be asked (we also have frameworks for this process, which we will share in later posts).

In our view, only when one can clearly answer in detail “What is going on?”, followed by the question “What does it mean?” can the various options for “What should we do?” be considered.  Note that the only forward-looking question in this framework for non-predictive strategy is the final one, and it is focused internally:  “What should we do?”, not “What will happen?” or  “What will the world be like?”, etc.

To sum up, time and energy spent on prediction can be more efficiently spent on a deep understanding of the present; i.e. shrink the time horizon you consider. The illusion and danger of prediction will go away, if painfully, and fewer mistakes will be made.

To read more about how to think about the future, read Milo’s post “The fragility if the future… and your strategy”, and my post “Welcome to Extremistan”. On the danger of prediction and forecasting, read “We’ve met the enemy and he is, er, forecasting”.

If you enjoyed this post (and wish to learn more about crafting non-predictive strategy), why not subscribe to our blog?


9 responses to “Crafting Non-Predictive Strategy, Part I: Deep Understanding Beats Prediction

  1. Pingback: Crafting Non-Predictive Strategy, Part I: Deep Understanding Beats Prediction | Data Analytics Use Cases |

  2. Philippe,

    Your KPUU-matrix, with certain re-formulations, could be turned into an inter-active morphological inference model. We employ general morphological analysis for non-predictive strategy analysis under genuine uncertainty.

    Some possibilities are presented in the book: “Wicked Problems – Social Messes: Decision Support Modelling with Morphological Analysis”. Description at:

    BTW: G. W. Leibniz would have completely agreed with you: “The present is pregnant with the future” – we simply don’t properly understand the present.

    T. Ritchey

  3. Dear Tom,

    I have spent more than 30 career years, at a reasonably high professional level trying to come to grips with this issue, and this article has summarized and hit nail on the haed precisely!

    Thank you. We need masses more of your kind and quality of thinking to move civilized society into the truly sustainable zone of economic of truly progessive development.


  4. Philippe, Tom
    Thankyou for the article and comments.
    Can I suggest that the last question be modified from ‘What should we do?” to “What should we do – next?” factoring in order of priority is imperative when one has scarce resources and relies on response cycle time as a means for advantage ( but isnt necessarily easy).
    Kind regards

  5. I like the concept, but I would always combine it with a scenario approach

  6. Pingback: Crafting Non-Predictive Strategy, Part II: Start with who you are « Silberzahn & Jones

  7. Pingback: Crafting Non Linear Strategy, Part III: Acknowledge the Nature of the Problem « Silberzahn & Jones

  8. Pingback: A Simple Framework to Develop Deep Strategic Understanding in Uncertain Environments « Silberzahn & Jones

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s