In the coming days, most of us will be ending a year-long cycle that started when the pandemic did. The news media sector was quickly impacted, with a drop in the volume of advertising investment and an increase in subscriber churn. The impacts were still seen in March and April, almost as a proxy for the size of the socio-economic problem perceived to be within the entire economy.

Faced with an environment of extreme uncertainty, the RBS Group structured itself based on its data-driven culture, organising an action plan to reduce the crisis’ impacts. With a focus on generating results, we use data to quickly identify problems and their causes, and we propose effective solutions to generate value.

In 2020, during the evolution of our data-driven journey, we had the opportunity to learn important lessons about how to effectively use the data-driven culture during times of uncertainty.

Track key and control metrics

In digital transformation businesses, the use of objectives and key results (OKRs) or a North Star metric is common. The objective is to direct all efforts toward objectives aligned with the strategy.

However, in times of crisis, a mature environment of control metrics is also relevant to anticipate problems and causes. Tracking control metrics and focusing on stages of business processes helps to predict impacts before they have a consequence on key metrics.

An example that summarises this view is seen by managing the subscriber churn metric. At Grupo RBS, we monitor control metrics, such as the number of cancellation intentions, which allows us to anticipate impacts on our churn rate. It also allows us to evaluate and take action to increase our ability to retain subscribers.

For environments of high uncertainty and volatility, predicting behaviours is quite complex. However, constant monitoring of the correct indicators has a positive effect on the rapid recognition of problems. Identifying deviations in behaviour quickly allows for a more complete analysis of the causes and a reduction of the final impacts. Monitoring the results daily is able to guarantee speed on most topics.

Create methodology for analysis of causes

Having the speed to identify the problem is as essential as understanding it. The variation in metrics tells us where the problem is, but very little about its cause. Having models developed for cause analysis allows for quick identification and understanding of the business problem. During a crisis, these models may lose their grip, but their adaptation to the new context is faster than their creation.

Following the example of churn, our defined analysis models are organised to evaluate dimensions of the cancellation intention metric (cause, channel, etc.), but also to understand the profile of subscribers (length of service, price paid, product, engagement, etc.). In this way, we were able to apply the same logic in the new context, quickly extracting the main causes to share with teams to build the action plan.

Track prediction models

Tracking and predicting key user behaviours is a useful science for maximising conversion. On the other hand, models can also be transformed into metrics. For example, a propensity-to-cancel model can be used to generate an indicator of the number of subscribers likely to cancel, helping to predict churn.

Another application may be to use the model to understand a variation that already exists in churn. For example, an increase in cancellations without an increase in the propensity to cancel may indicate an exogenous cause for the variables used by the model.

Build scenarios

With an increase in uncertainty and difficulty in predicting behaviour, the construction of scenarios seeks to bring sensitivity to decision makers. Looking at bands of variation reduces the discussion based on feeling, allowing all areas to prepare themselves to act in a more or less energetic way in the face of problems. Building scenarios is an investment in planning that reverts to the success rate of actions and the speed of reaction to changes in reality.

Data-driven thinking and acting

In the old business model, decision-making was based on how the main leaders felt. Many companies have a culture informed by data, as they are unable to end the cycle using the insights to guide the decision.

Companies with analytical maturity act in the opposite direction, reducing feeling in times of uncertainty and making decisions based on data. In addition, using an agile mindset enhances the data-driven culture. Conducting test-and-learn cycles allows you to validate hypotheses and implement the best actions.

Our main case for adopting a culture driven by data occurred with the annual increase in newspaper subscription prices. Due to the uncertainty of the economic environment, we did not know whether our price increase strategy would have a positive impact or be offset by the sharp increase in subscriber churn. Therefore, we created an A/B test to check price sensitivity. We segmented groups of subscribers according to their propensity to cancel and subscription profile, evaluating the results for each individual group.

This test allowed us to validate hypotheses with speed and mitigate the risk in the strategy for the rest of the subscribers. Our price increase had a 20% gain in addition to the budgeted scenario, even in the face of an unprecedented crisis.

Conclusions

We can say that a culture driven by data has allowed the Grupo RBS to adapt faster and gain intelligence to identify and mitigate the impacts of the crisis. These learnings will be important for future uncertainty scenarios, but they also helped us develop new capabilities in our data-driven culture so we are even more effective during the next business challenges.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .