Twos & Twos
Introducing Igor Ansoff
A long time ago, in a world yet far, far away from Covid, I presented a keynote at a conference in Mumbai, India. Although it and its organizers remain among my most favorite, four days in around-the-clock high heat and humidity took their toll on the ice-veined Scandinavian that is yours truly. As it happened, the route back took me via Switzerland. The crisp, cold, mountain morning air that greeted me as I exited the aircraft at Zürich airport still remains fresh in my mind; I could, at last, breathe again.
Moving from Blue Ocean Strategy to the Ansoff matrix had, hand on heart, a similar effect. Although it is far from perfect, the lack of general fuckwittery was genuinely refreshing.
Many will, of course, be familiar with the 2x2 matrix also known as the product market expansion grid. Created by applied mathematician and strategic management thinker extraordinaire Igor Ansoff in 1957’s Harvard Business Review article Strategies for Diversification, it has been a staple of business education ever since, and still features prominently in marketing and strategic management texts. That some expected it to be featured in my original list was therefore not particularly surprising.
However, there are a couple of reasons why it failed to make the cut.
At its core, the Ansoff matrix is a planning tool that details four ways of growing a business:
Before we explore the options that the grid provides, it is worth briefly mentioning what Ansoff called the underlying ‘basic concepts’ so that we interpret them as originally intended.
Two terms are of particular importance. The product line refers to both the physical and the performance characteristics of individual products; the size and weight of, say, a car would be the former and its acceleration, top speed and fuel consumption the latter.
The product mission, meanwhile, is a description of what the product is supposed to do which, by crucial extension, defines its market. This particular use of the word mission, Ansoff argued, was more more useful than any concept of a ‘customer’, since customers can have many different missions, each requiring a different product. Personally, I am not entirely sure how much the distinction helps - clients who use the matrix often struggle with this interpretation of markets.
Either way, the two aspects of product mean that a ‘product-market strategy’ can be defined as ‘a joint statement of a product line and the corresponding set of missions which the product is designed to fulfill’, with four different approaches to execution:
1. Market penetration involves an effort to increase sales of already existing products to already existing customer bases or new similar customers.
2. Market development is about finding new markets for existing products with at most minimal adaptation.
3. Product development is retaining the current mission but developing new products within it, in other words, creating new products for existing markets.
4. Lastly, diversification entails a simultaneous departure from the present product line and the present market structure, i.e., new products into new markets.
The more we move down and to the right in the matrix, the higher the risk. Or to put it perhaps more accurately, the farther the company moves from its ‘comfort zone’ – what it is currently doing and therefore knows to work – the higher the uncertainty.
(For the purposes of clarity and later reasoning, the difference between risk and uncertainty is that risk concerns an outcome that is unknown but a probability distribution governing said outcome that is known. Uncertainty, on the other hand, is characterized by both an unknown outcome and an unknown probability distribution.)
Ansoff’s argument stands in stark contrast to the claims of Kim and Mauborgne who argued that the act of creating new product-market spaces somehow minimized risk (while at the same time maximizing opportunity). Probably needless to say, I am inclined to side with Ansoff on this one, at least within a short-term timeframe.
Understanding what options might be on the corporate table is undeniably important for any strategic decision-making process. Although the matrix is not all-encompassing (as some rather naïvely have suggested), it can doubtlessly clarify thinking and classify objectives. As a result, it remains a popular tool not just with executives tasked with devising future growth paths, but also marketers developing tactics.
At the same time, as one might expect given its longevity, Ansoff’s work has garnered a fair bit of scrutiny and criticism over the years. Yet while some of its flaws are obvious (the lack of competitive considerations, for example), commentators habitually miss the mark and simply add to the original design; from four (Ansoff) to seven (de Waal) to eight (Pleshko and Heiens) all the way to sixty-four (Varadarajan). This shows not merely a misunderstanding of Ansoff’s intentions (the grid was never supposed to be context free, and he was very clear on its limitations), but also a general lack of understanding of complexity. Rules of thumb are not easily replaced by rules of (wrongly perceived) precision.
No, the actual flaws in the matrix are significantly more fundamental.
Firstly, as has been detailed by John Dawes, it suffers from an overreliance on subjectivity. In the original article, Ansoff uses the example of an airplane manufacturer that adapts its passenger jets into cargo transportation to illustrate market development. Although one could argue that this is indeed a new product mission and thereby market (remember, the product mission defines the market), the client list could be more or less the same; many if not most aviation companies also offer airfreight. It would therefore also qualify as a penetration strategy.
To make matters more confusing still, if we accept, as Ansoff suggests (and others such as Kotler have since argued), that new products can be incrementally new, then the altered airplanes could be considered a product development strategy if the customers remained the same, or even diversification if they did not! In other words, we end up in a situation where the original example could, depending on personal interpretation and/or preference, fit into any of the four boxes or all of them at once.
Secondly, Ansoff makes a clear distinction between what the company can know and what it cannot know:
So far we have dealt with diversification forecasts based on what may be called foreseeable market conditions – conditions which we can interpret in terms of time-phased sales curves. Planners have a tendency to stop here, to disregard the fact that, in addition to the events for which we can draw time histories, there is a recognizable class of events to which we can assign a probability of occurrence but which we cannot otherwise describe in our present state of knowledge. One must move another notch up the scale of ignorance in order to consider these possibilities.
Many businessmen feel that the effort is not worthwhile. They argue that since no information is available about these unforeseeable circumstances, one might as well devote the available time and energy to planning for the foreseeable circumstances, or that, in a very general sense, planning for the foreseeable also prepares one for the unforeseeable contingencies. In contrast, more experienced military and business people have a very different attitude. Well aware of the importance and relative probability of unforeseeable events, they ask why one should plan specific steps for the foreseeable events while neglecting the really important possibilities.
This, he argues, means that strategists should make a point of creating ‘flexibility’. Diversification could help to that end. However, despite the accuracy of this observation, his conclusion that one still somehow would know how to diversify is a non-sequitur. If one knows what to do, then there would be little uncertainty.
If we are to be technical about it - and feel free to skip this part - anyone acting within a complex adaptive system such as a market has to deal with what are called adjacent possibles. As long-time subscribers will know, they can be defined as what can come to be. Each adjacent possible that becomes a reality enables a whole new set of adjacent possibles, and it is impossible to know what these might be ahead of time.
To illustrate, the creation of Meta was enabled by the creation of Facebook, which was enabled by the creation of social media, which were enabled by the creation of the internet, which was enabled by the invention of the personal computer and so on. A person living a century ago could not have predicted the rise of the internet any more than a person living fifty years ago could have predicted the price of advertising on Facebook today; the invention of the internet did not cause Facebook to rise, it merely enabled it.
The principle of the adjacent possible means that no matter how much data we collect or how sophisticated models we use, we will never be able to predict the future. As we cannot know what new adjacent possibles the current adjacent possible might enable, there is no way of knowing the sample space (all possible adjacent possibles).
This presents a fundamental problem to Ansoff’s reasoning which is based on mathematical calculations to determine course of action. As noted, the more we move down and to the right in the matrix, the higher the uncertainty. When it comes to diversification, not only do we not have any historical data upon which to model, we do not know the sample space. Without it, it becomes impossible to calculate the requisite probability values (proving my earlier point that any risk would be unquantifiable and therefore strictly speaking not risk but uncertainty). Immediately, a central part of the exercise falls down. The grid turns from predictive to descriptive, from option inducing to option informing.
Given its limitations, then, the best way to use the Ansoff matrix is together with a contextual understanding of the business and an already set strategic aspiration. It is for this reason that it failed to make the original list of concepts; it should not be considered until later in the process.
To be clear, I do not have any issue with it in principle – it is a good way of visualizing what one can do – but despite Ansoff’s best efforts and Kotler’s claims to the contrary, it tells us little about what one should do or why.
Next week, we are going to look at Bruce Henderson’s growth share matrix to see whether that fares better. Until then, have a lovely weekend.
Onwards and upwards,