Horizon scanning, risk sensing, predictive analytics, forecasting, insight. All of these terms imply some kind of mystical, deep learning, cutting edge technological solution; and technology plays a huge part in the ability to get better at prediction and spotting issues before they become crises. However, no model or technical solution that anyone has discovered takes into account the often unpredictable nature of human behaviour.Appropriately employed technology can give you the ‘what’ far more efficiently and effectively than humans, but it isn’t yet good enough to explain the ‘why’ or the ‘what next’; and that is the power of meshing powerful technologies with human beings. Which is what can make the difference in a crisis situation. There are five key components to prediction:
- Look at the problem for a long time; become an expert. The longer a particular problem is analysed both in terms of technical and analytical effort and the period for which data exists, the more able one is to move into the predictive space. This calls for identifying and assessing which issues are the biggest threats to your organisation and tracking them over time. Moreover becoming an expert during the ‘Identify and Assess Risks’ and ‘Prevent and Prepare’ stages of a crisis will translate into greater effectiveness during the ‘Respond and Recover’ stage.
- Trends over time are critical and the details underpinning them can be difficult to spot retrospectively; risk sensing and prediction is not merely the extrapolation of historical data. Technologies that have been trained and refined over a period of several months to collect and prioritise data over a year will be far more effective than a ‘quick look’. The analyst that has been examining a particular sector or problem for a long time will be far more likely to spot the weak signals that indicate an emerging issue.
- Become a tech expert; make it work hard for you. Technology can do a huge amount of the time-consuming heavy lifting that used to be the preserve of analysts. But you have to invest time and effort in understanding how and why it works, and then how to make it better. Underpinning every amazing tech platform is a team of people supervising its learning and pointing it in the right direction; to make it work for you, you have to do some work for it first.
- Shorten the forecasting timeframe and narrow the problem set; ask the right question. Consider these two questions:
- How will political interference in regulation in Africa increase the likelihood of a supply chain related crisis over the next 20 years in Africa?
- Will the imminent clampdown on unethical practices in the South African manufacturing sector result in both supply chain disruption and an increase in the likelihood of a reputational crisis in the next 2-4 weeks?
- The first question is broad in scope, relies on a significant number of assumptions and it is unlikely that the customer is going to take the author to task in 20 years’ time for an inaccurate prediction. However, a well thought through and evidenced answer will give emerging risk or strategy functions in a business an idea of what the future holds. The second question has narrowed the geographical, thematic and temporal scope down considerably. A data heavy, well researched answer by a subject matter expert analyst will prompt an immediate business decision; and its accuracy can be evaluated and measured, and the analysis revised fairly rapidly. When responding to a crisis, when the aim is to keep your business running, there might not be much time to articulate a deeply considered, narrow question; but taking the time and having a start point to work from, will inform the decision makers at speed, with accuracy and will support the response and recovery process.
- Train for it. Forecasting is part science, part art. Careful practice in controlled environments and critically revisiting previous predictions will help narrow down mistakes and re-train the mental predictive model. Prediction is not crystal ball gazing; it is about weighing up possible outcomes and providing a judgement on how these may affect the customer.
- Challenge convention; invite disagreement and creative tension. History is littered with examples of institutional bias and groupthink resulting in poor analysis. Enabling a creative tension with appropriate emotional support (the freedom to make mistakes and challenge convention) within a team of analysts can help avoid the trap of groupthink; and the combined brainpower of a team can be frighteningly powerful. However, working alone has its benefits too by avoiding the trap of self-righteousness complacency that teams can fall into – “we are of one voice therefore our work is done”.
Corralling a suite of technologies and the appropriate skills in the same place is a challenge; prediction isn’t taught in many places. Emerging stronger from a crisis doesn’t happen by accident. To discuss any of the above please contact Al Bowman.
Thinking Fast and Slow by Daniel Kahneman
Super-Forecasting; The Art of Science and Prediction by Philip Tetlock and Dan Gardner
Al Bowman, Senior Manager, Deloitte Managed Services
Al Bowman is developing a suite of risk sensing/intelligence fusion propositions to support the decision making process within organisations. Al’s last military job was commanding officer of the British Army’s Intelligence Fusion Centre responsible for the enabling the prioritisation and leading on the coordination and delivery of global intelligence and security risk advice to the Army.