There’s a theory that the gargantuan porn industry is the hidden force behind many technological advances. VHS won over Betamax in the video wars (remember?) and streaming got the research dollars to become viable because the porn barons and baronesses committed to them.
While this idea makes for great copy, one could argue that there’s a more potent hidden force than porn: the military.
The Internet itself, let’s not forget, started out as Arpanet, funded by the Advanced Research Projects Agency (Arpa), part of the US Department of Defense. The good news is that the US and other military organisations are ploughing money into how data can be used to develop and fine tune combat strategies, and make the outcomes more predictable.
It should come as no surprise that Darpa, the renamed Arpa, is one of the many players in the quest to analyse data for purposes of military strategy.
For example, as The Economist reports, the Major Combat Operations Statistical Model (MCOSM), jointly developed at the Naval War College and the Naval Postgraduate School in California predicted that Russia’s move to take Kyiv would fail. Its creator says that MCOSM says seven out of 10 forecasts proved correct. This tally is likely to improve as more predictions are correlated with what actually happened — like so much in the software world, ceaseless iteration is the key to success.
The more one reads about what underpins these software programmes, the more one sees the potential for chief executives and boards in formulating and monitoring business strategy — and the pitfalls.
At this stage, it’s as well to remind ourselves that the very concept of strategy is one that business ultimately borrowed from the military.
What are our goals? How do we reach them? Do we have the resources we need?
These questions have always been hard to answer, but in today’s global and fast-moving business world, it is even harder to do so. As GIBS Professor Manoj Chiba points out, in business there are a large number of interdependent variables that are changing daily. Change is happening in unpredictable ways, and most companies will have to globalise in some way to compete—for example, Amazon opening its doors here in South Africa means all local companies are actually competing globally.
The current obsession with data stems from the insight that the ability to collect and process large amounts of data could help executives understand — and deal with — these changes. But while the benefits of using data operationally is well understood and most companies are on the road to using it for these purposes, research from McKinsey indicates that artiicial intelligence (AI) is most widely adopted in service operations, product development and marketing/sales, with strategy and corporate finance lagging.
When it comes to benefits, however, strategy and corporate finance are right up there, with 76% of respondents seeing moderate or significant value from AI, only slightly behind the highest figure of 80%. However, of that 76%, only 45% report significant benefit as opposed to manufacturing, where 57% report significant benefit.
To summarise: AI does deliver benefit in the strategy and corporate finance category, but its full potential in terms of both benefits delivered and overall adoption could be significantly improved.
Getting to grips with data
Reading the The Economist article, one gets a real sense of what the crucial issues are when it comes to using AI in strategy, and they tend to coalesce around data and the multiple challenges and opportunities it encapsulates.
At base, companies need to develop a much greater and deeper understanding of how data works (yes, it has a life cycle) and what it is. Antonei Badenhorst, the divisional director for data enablement at 4Sight, comments that in order for data to be valuable in the strategy context, it has to be broad in scope, and of good quality. The last point takes us back to the venerable, but perennially true, data maxim: “Garbage in, garbage out.”
Strategy data needs to be global in scope, Professor Chiba says; it must also include real-world data, including unstructured data such as photographs, says Sameer Jooma, who is the executive lead for trader ecosystem at Standard Bank. For example, Jooma says, getting data about Africa’s informal economy, particularly in the retail sector, is often difficult and one needs to rely on alternate methods of data collection. He has found photographs to be one of the best forms of alternate data due to the richness of information contained in them. The photographs are usually obtained either by an on-the-ground field force or crowd-sourced through “citizen data gatherers”.
Professor Chiba makes the important point that simply acquiring more and more data does not necessarily lead to better business decisions: it’s vital that companies are intentional about what data they collect and how good it is. “There’s a sweet spot between too much data and too little,” he says.
Goldilocks proves her worth once again.
What’s hidden in the algorithm?
A profound issue is that software requires many assumptions to be made when the software’s algorithm is constructed, and again when populating it with data. This means that there is a huge dependence on the domain knowledge of the developer who creates the algorithm, and then on the current knowledge and, above all, judgement of the person inputting the data. For example, inputs for a military program would include an assessment of both parties’ specific military skills, such as reconnaissance, firepower, mobility, logistics and so on — areas where there is no firm answer. Such military software could also include estimates of mental or cultural factors such as, for example, an assumption that combatants from democracies would be more creative than those from authoritarian regimes, which discourage personal initiative.
In other words, particularly when it comes to strategy and predictive modelling, hard data rests on a web of assumptions whose accuracy is less certain. The strategist is supremely at the mercy of the software’s inbuilt assumptions and the estimates fed into it.
Two things follow. One is that a much more iterative process is needed when using data to support strategic decision-making. The second follows: the strategic horizon has to become much less distant to allow this kind of approach — not a bad thing given the increasing speed at which business moves. But it does surely involve a totally different approach to strategy than the one to which boards and executive teams are accustomed.
A new approach to strategy?
This iterative process of strategic decision-making from one “Aha!” moment to the next suggests that strategy will become evolutionary rather than a programme set at intervals. It also fundamentally disconnects strategy from a particular individual or group. Ajay Lalu, the chief executive of Q-HOP, a company that develops data-driven solutions primarily in the retail industry, says that large corporations experience too-frequent changes of leadership that inevitably give rise to a strategic review. However, if the strategy is the result of an ongoing process of constantly being recalibrated in response to data, it’s possible to have a strategy that is genuinely long term in that it does not depend on who is on the team making it, and that is also flexible enough to respond to short-term shocks while maintaining course.
“You need to get data expertise onto boards, and you need to understand the difference between long-term and operational strategy,” Lalu observes. “Boards will need to spend more time understanding the nature of the data sets available and how they can be used.”
And, as Badenhorst points out, one needs to understand just what methodology is being used to create the insights; each methodology or data model has pros and cons, and these must be well understood.
One could speculate that a growing dependence on data will spell the end of the long-time domination of chartered accountants in the executive and on boards, with engineers becoming more prominent.
Be that as it may, access to data talent is already a challenge. Other challenges relate to the need to allow data to travel freely across the organisation — the corporate tendency towards silos is fatal here.
Another angle comes from Derek Wilcocks, group chief informaton officer of Discovery, a company that has always prioritised data. He does not draw a clear distinction between strategy and operations: because the company is so product-oriented, strategy begins there. In a sense, products are the strategy, he argues. As a highly entrepreneurial company, Discovery relies on data to develop and validate ideas; for example, Discovery Bank was only able to develop personalised products with a better cost structure because it had decades of data on which to draw.
To conclude, it’s clear that the use of data to create and monitor strategy is in its infancy and is likely to lead to significant changes in how companies operate, many of them unclear at the moment. But, as the Ukraine war is demonstrating, the potential is too great to ignore. Perhaps the last word should go to Professor Chiba: “Data is just an enabler — it guides strategy, it doesn’t make it,” he says.
Understand what data can do for you
A stimulating introduction to data and the kind of “Aha!” moments it can identify is Don’t Trust Your Gut by Seth Stephens-Davidowitz — it’s an entertaining and thought-provoking read. Although he does not address business strategy directly, the examples he uses clearly show how data can deliver insights with the potential to spark profound rethinking of strategies.
For example, we all want to do the best for our children and think we know what we should do. In fact, the data shows that most of the decisions parents make have much less effect than we imagine, and that the most important decision is often fluffed, presumably because nobody really recognises how important it is. That decision is the neighbourhood in which you raise your children, because it has the most impact on his or her future success in life. There characteristics determine the neighbourhoods most likely to increase a child’s chance of success:
- Percent of residents who are college graduates
- Percent of two-parent households
- Percent of people who return their census forms.
Such adults are likely to be smart, have stable lives, and be active citizens. This knowledge would profoundly change one’s parenting strategy.
Gallery
