SDA Professor of Strategic and Entrepreneurial Management
In a recent article, David Collis makes an interesting attempt at combining the two old and seemingly-opposed views of the strategic process: the top-down analytical approach that implements a plan carefully chosen by the top management of the firm and a bottom-up emergent approach that shifts gears continuously in reaction to what managers on the ground are experiencing. His conclusion is that start-up companies might benefit from using the traditional analytical process to set an initial vision and strategic direction that later can be refined and adjusted by using typically emergent Lean Start-up tools. All in all, a worthwhile point for entrepreneurs to consider.
What Collis does not address though is how the deliberate and emergent approaches can also be combined within the strategic process of more mature and established companies. The majority of those firms are very adept at strategic planning but notoriously clueless at managing emergent strategy. In a way, the increasing interest in corporate entrepreneurship might be seen as another chapter in the periodic realization that mature firms do not know how to innovate their business models and products within the rigid frames of their planning process. So how can those firms become more open to emerging internal entrepreneurship?
The power of experimentation
One crucial factor concerns the tools that executives use in the strategic planning process. For decades, business schools and strategy consultants have been promoting frameworks and tools that belong almost exclusively to the deliberate approach (think Porter’s frameworks, etc). This is not a criticism -after all, it is a lot easier to teach structure than chaos- but it certainly has left strategic thinkers with an incomplete toolkit. On the other hand, emergent strategy tools do exist. For instance, the entrepreneurial world today is rich with concepts like prototyping, validation (aka, experimentation) and quick iteration. In this article, I explain how one of those tools, experimentation, could help the strategy process and why all managers should be familiar with it.
Strategic analysis is characterized by the implicit or explicit formulation of a number of assumptions. The traditional tools to manage those uncertainties are 1) to analyze past data and 2) to recall past situations that can serve as analogies. Such focus on the past, while valuable, is also a big source of inertia. Think about the classic Kodak story and how the assumption that the imaging industry was all about the razor/blade revenue model blinded the company’s executives to the different nature of the digital imaging world.
So whenever strategists identify an important assumption in their mental model, instead of thinking harder or discussing longer, why don’t they test it directly? Why not construct an experiment that can validate or reject their hypothesis? For example, Pfizer experimented with three different channel alternatives (pharmacies, employers and online) in a small-scale controlled market environment to learn about the best business model for its smoke-quitting Nicorette product. Contrary to their intuitions, Pfizer executives learned that the online channel generated the most customer interest and used that insight to renew their strategic approach for the product.
The stages of systematic experimentation
Systematic experimentation would make the early stages (i.e., diagnosis stages) of the traditional strategy process look more like this:
1. Visualize your business model. Of course, this can be a mental visualization but it often helps to explicitly draw it on a piece of paper, especially if this is done by a group.
2. Extract key hypotheses from every part of your business model. For instance, are you looking at the right segment? Do customers in that segment have the needs you think they have? Are those needs best satisfied with your value proposition? Can your company efficiently develop and manufacture that product? Are you using the right channel to deliver it?
3. Rank those hypotheses in terms of importance. Some of them might make or break your business while others are secondary.
4. Devise experiments for as many of those hypotheses as time and budget permit, starting with the most impactful ones. From online ad/link tracking, landing pages or mock sales to customer co-creation, split testing or minimum viable product (MVP) prototyping, there is a wide range of tests that serve different needs.
5. Use the results to update your thinking -changing or confirming your intuitions-, and devise new experiments until you are confident enough to commit to a specific strategy.
Obviously, this sequence is not meant to be a rigid prescription but rather a discipline that strategists need to incorporate in their work at certain moments. For instance, executives in large firms – who are naturally wary of committing to untested strategic paths that are significantly different from the status quo- might find experiments particularly attractive during periods of high uncertainty.
On the other hand, one reasonable objection to the systematic use of experiments in strategic thinking is that they are expensive and time-consuming. While there is no doubt that a gut feeling is cheaper and more immediate than an experiment, the latter is becoming exponentially less onerous, thanks to the development of big data analytics.
Use the power of big data
Big data analytics promises companies to improve their decision making by exploiting massive amounts of information. For example, a pharmaceutical company used neural networks –a big data technique- to model the relationships among more than 50 variables involved in a production process, a number impossible to analyze jointly with traditional tools. The new methodology revealed five of those variables as the key drivers of efficiency in that process, which led to their optimization and subsequent improvement of yield by more than 30%. Although the potential of big data is high, managers are still learning where and how to deploy its power. So far, it has been most widely used in the marketing and sales areas, with retailers such as Amazon or Tesco, for example, collecting billions of customer data points every month to better tailor individual offers, prices and promotions.
Strategic analysis, on the other hand, is still a largely unexplored area for big data, in terms of running experiments as well as the traditional analysis of past data. Both types of tools would undoubtably benefit from access to large samples and quick results. Think about market segmentation, for instance. Time and again, new entrants are able to disrupt incumbents’ businesses simply by virtue of identifying and targeting a new segment of customers, which begs the question: Why were incumbents not able to see that segment before? One answer is that they were probably not running enough experiments to test their understanding of the market. For instance, Luminar –a data analytics company that specializes in the U.S. Latino market- constantly runs market tests on vast amounts of data from thousands of sources in order to develop sophisticated insights on very granular subsegments of that market. A similar approach, albeit at a smaller scale and done internally, could add great value to the market analysis step in the strategy process of many companies.
In summary, the process of formulating or revising a strategy needs more testing tools. So far the main constraints seemed to be that experiments were expensive and time-consuming and that managers did not know how to design and interpret them. Thanks to big data, the only serious constraint is the latter. Time for managers to learn how to run strategic experiments.
*I would like to thank Markus Venzin for his comments