Analytics, data science, data decision making,... are topics that have been on the top of all c-levels agenda and have been a very important trend in management for quite some years, now, as the data economy starts building its way and becoming one of the most important industries.
We can leverage data to enhance the quality of our decisions in our daily lives, and use analytics to predict, to foresee and to anticipate trends, consumer behaviours, and pretty much any pattern.
Although analytics has been such an important aspect of prediction in management and we have been using it to try to predict almost everything, we were not able to use all the data and analytics do predict the impact of COVID-19.
Advanced Analytic Models have been under pressure for some months, now, especially for businesses using these kinds of models of data analytics, such as Machine Learning, which is influenced by the real world. So, what happened was that all of the sudden all these models were also affected by this uncertainty in data, in inconsistent behaviour and in missing data points and became showing some cracks, forcing humans to step in and set them straight.
Throughout the years, business analytics has adapted and evolved, going from a more information-driven results to optimization-driven. However, there’s still a long way to go and Pedro Amorim and Bernardo Almada-Lobo, Co-founders of LTP Labs and guests of our most recent Beyond Now Webinar, point out six limitations that may be the reason why analytics wasn’t able to predict the COVID-19 pandemic:
1. Single-scenario and punctual projections of what will happen
2. Static yearly plans with limited or non-existent updates along the year
3. Poorly detailed plans without the adequate level of granularity to estimate outcomes
4. High reliance to what happened in the past to predict the future
5. Solely dependency on internal data, neglecting other potential sources of information
6. Lack of skill to use advanced analytical methods such as sentimental analysis to drive decisions
“It’s true that these limitations weren't a pandemic consequence. They were already there and with the pandemic these issues surfaced”, says Bernardo Almada-Lobo.
Humans respond to crises in different ways. Therefore, behavioural insights are of critical importance. This includes knowledge about what drives behaviour and awareness of changes in these drivers. Organizations dealing with constantly changing conditions will need to understand where they are now and plan how to move forward in a dynamic environment. This includes gaining insights into customer behaviours, which impact products and pricing, optimizing supply chains, and continually updating revenue forecasts.
But how can we use analytics to deal with this uncertainty, you may ask? For Pedro Amorim, the answer is very clear: experiment. “In a complex system if you don't experiment you change something you cannot see further about causality. We need to use the statistics to learn about the causality and not make random assumptions without testing”. Companies must embrace the uncertainty, confronting it head-on and building it into the company decision-making. According to Delloitte, the best way to do it is:
- Consider uncertainty under different time horizons
- Identify the most salient uncertainties for your industry and business
- Use the uncertainties to envision multiple different futures
- Seek diverse perspectives
- Incorporate scenarios into decision-making
- Distinguish implications of different decisions
- Make choices and monitor
By 2020, it’s estimated that 1.7 MB of data will be created every second for every person on earth.
As the pandemic continues, the sociological and economic effects become evident, and thus, many nations, businesses and individuals are exploring coronavirus recovery strategies based on data analytics. In the last months, consumer behaviour started shifting rapidly and significantly altered normal patterns. This mostly impacts business forecasts, such as trends or KPIs forecasts that depend on volatile market conditions. For Bernardo Almada Lobo, given this current volatile environment, there is now a strong belief from managers, that data-driven decision making is now a must-have, in order for companies and businesses to strive. It is now mandatory to do better, faster and more informed decisions, and for that analytical tools and skills are even more critical.
For Pedro Amorim, the essential skills that you need in your company is for the CEO to know how to run a data mining model. “You will need everyone in the organization to understand and prototype these types of models”. On the other hand, for Bernardo Lobo, it all comes down to the data scientists to understand about the business, so they cannot really just hide behind the technology.
To conclude, experts believe that leaders need to be confident and insightful, constantly monitoring and re-evaluating the evolving situation. Continuous, rapid decision-making based on accurate, data-driven analytics and simulations will help leaders meet, and exceed, the expectations of their people and customers.
