2020 in the rearview mirror

With a decidedly "unprecedented" year behind us, we thought it would be interesting to recap the main types of data science and engineering projects that kept us busy in 2020.

?? Understand what's going on

As soon as the COVID crisis hit, many organizations were eager to make sense of what was happening.

Real-time/crisis reporting

It's hard to remember today how the month of March unfolded, but the lockdown happened rather quickly. So, the first focus was to provide a measure of what was going on.

Information that used to suffice at regular intervals (e.g., monthly dashboards) was now necessary daily or almost real-time.

Speeding up reporting from a monthly to a daily basis is easier said than done. The data flows, and the infrastructure required to move to real-time reporting are very demanding.

Armed with a good dose of ingenuity, we focused on business-critical indicators. We set up automated processes to provide business users with a faster, more actionable overview of their operations.

Identify what changed and what stayed the same

2020 was a year with extreme variability. Imagine you are a retailer located at the intersection of business and residential areas. In 2020, you could have lost many customers, but you might also have gained many new ones. What happens if lost customers don't return? What happens if new customers don't stay loyal once things go back to normal?

Analytics play a significant role in figuring out what drives large swings in customer behavior. Understanding those fluctuations is the key to efficient promotional activities and logistical planning.

Another consequence of this challenge was figuring out the relevant time frames for customer profiling. Should we still rely on 2019 data, or is there a "new normal" to factor into forecasting?

? Seek the quick win

In many sectors, the mood was not one for experimentation. Many projects were laser-focused on very narrow "quick wins" to make sure all euros invested in data projects contributed to a bottom line already suffering under pressure. Some of the fast results projects we worked on were:

• Customer acquisition: what actions can we launch in the next couple of weeks to quickly boost customer acquisition?
• Loyalty: how do we limit customer retention initiatives to those with the highest likelihood of profitability?
• Process: what short-term optimizations can we still generate in terms of speed, quality, or cost?
• Cash: how can we improve our liquidity position by anticipating fraud and understanding what's behind our Accounts Receivables?

? Build scenarios

An analytics project's goal is rarely about delivering certainty but rather about evolving from uncertainty towards a calculated risk.

Never has this been truer than in 2020. Some of the most challenging projects we worked on this year revolved around scenario planning.

How can an organization's top management define a strategy with so many uncertainties, both internally and externally? We, therefore, asked questions such as:

• What are the sales forecasts if our retail branches' operations remain limited until 2022?
• What happens to the nature of products and services consumed if unemployment rises by more than 1%?
• How can I anticipate employee well-being/absenteeism if remote working becomes the norm?

Starting from brainstorming sessions with the entire organization, we iterated versions of a decision tree to select the most likely scenarios. We then modeled the variations within the internal and external factors that impacted the organization's KPIs.

?? Lay the groundwork for future success

For those who lived by the motto "never let a good crisis go to waste," 2020 was a great year to plant the seeds of future data-driven successes. To name a few:

• Define a proper analytics roadmap so that the organization is ready to act as soon as things go back to normal.
• Double down on automation in a context where physical presence is increasingly irrelevant.
• Invest in data engineering as an enabler for advanced analytics projects. In this case, we also practiced what we preached since we also seized the opportunity to invest in our own Data Engineering practice (Psst! We're hiring ??).

?? What then for 2021?

When asked to do the impossible, we often say that we are data consultants, not fortune tellers.

But one thing is certain: data is more than ever at the core of many successful business transformations. The sooner you consider analytics a core component of your future successes, the more likely your organization will overcome "unprecedented" events on the path to success.