Get certainty for the uncertain: Leveraging big data and scenario modelling techniques to prepare for the after-crisis

Authors: Andreas Späne, Dr. Matthias Schlemmer

The future is, by definition, always unpredictable. Currently, however, the future seems even more unpredictable than usual, mainly because COVID-19 is a crisis unlike any we have experienced before. Alongside the need to respond to the crisis, planning for the transition from crisis mode back to business as usual is important. But how is this to be done? More variables than normal are unclear, so planning instead needs be based on well-founded hypotheses of likely events. By building on these hypotheses and the resulting scenarios, it is possible to calculate the impact on an organization. This exercise offers a glimpse into possible futures, providing support to decision makers. The time is now right for this robust data-driven scenario planning.

Much of the uncertainty we are facing lies in the simple fact that the COVID-19 crisis is unprecedented. No crisis in recent decades has had similar implications. Whereas, for example, the financial crisis of more than a decade ago taught us some lessons on how to deal with economic shocks, the current situation has shut down whole industries for weeks, forced supply chains to become more flexible and placed major physical constraints and psychological strains on our workforce.

Consequently, conventional approaches for how to manage one’s business are not necessarily appropriate any longer (see also this link). When projecting even further ahead, from ad-hoc crisis management towards the transition to non-crisis mode, even more uncertainty arises. For example, what about brick and mortar retail stores? What will happen when businesses reopen? Will consumer demand revert to what it was? Will we be able to secure the right amount of stock and maintain supply? Will the same channels be used by our customers? Historical data and experience will very likely prove to be of little value in preparing for that moment, as there are no comparable patterns to be found in such data.

Even more so than usual, the best way to prepare is solid scenario planning. Allowing for ambiguities and embracing uncertainty can help to distill the most likely situations into three to five scenarios, and to create plans for each. Scenarios provide managers with forecasts on how the business may develop in response to changes in some indicators. The scenarios may not only help in deciding on concrete measures, but also in building a clear story to pass on to their employees. This allows for faster decision making, with leaders allocating the necessary resources and focusing on the right issues.

It is imperative to exclude emotion and only use hard data when simulating the specific impacts of various possible scenarios. Scenarios are, fundamentally, complex and interdependent what-if narratives. While the scenarios may not be based on historical data themselves (as outlined above), their impact should be simulated by using data analytic approaches to define and calculate the most relevant variables and triggers. Data sources will be dependent on the scenario outline, but may well include (external) macro-economic and socio-economic data, as well as business-specific (internal) data. However, it should be borne in mind that this exercise can easily become very complex, especially in the case of multi-country operations. For example, store opening scenarios might vary substantially according to regional COVID-19 regulations, routes-to-markets and lockdown timelines.

The necessary assumptions for generating these scenarios are highly dependent on your specific business. A basic structure for such an approach would look like the following:

1. Envision your value chain in its entirety. A typical list of elements may include: Customer operations (including customer demand, sales channels and pricing), Product supply (including operations, supply chain and logistics), Operations (including organization and people), Financial information
2. For each element, formulate some concrete questions to assess the potential impact of COVID-19 (such as “What will the sales channel distribution look like post-crisis?” or “What is likely to happen in the countries in which my suppliers are located?”)
3. Pinpoint the triggers or indicators that will influence the likely outcome to the questions asked.
4. Identify the variables, such as the number of delivery staff, affected by each of the respective answers. Hypothesize, but stay realistic.
5. Build a model (most likely the tricky part!) that allows you to simulate different triggers and observe the change in outcome variables.
6. Decide on the three to five most likely scenarios and prepare meaningful responses.

By using analytics approaches, the weight of triggers and the most significant dependencies can be made transparent to decision makers. In order to do this, however, companies will have to build up the necessary data and analytics capabilities quickly. They need to have the right talent and the appropriate technological infrastructure on hand to carry out such analysis.

We have helped clients with the scenario modelling outlined above. This involves asking the right strategic questions, generating solid hypotheses based on the dependencies, and using data-driven modelling of their outcomes. Ultimately, this helped decision makers identify non-regret moves and plans in response of relevant indicators and triggers.