THANK YOU FOR SUBSCRIBING
Follow the Money as Roadmap for Data Analytics
By Hiek van der Scheer, Chief Analytics Officer, Aegon
Data analytics provides businesses a huge opportunity.
It goes without saying that the opportunities with data analytics are huge. Even if you don’t believe the promises that consultancies make, and you take the quote ‘data is the new oil’ with a grain of salt, you cannot deny that data already has a huge impact on business models. Uber, Airbnb and Netflix are the obvious examples: They are not transportation, hospitality or entertainment companies but data analytics-driven companies.
However, data is useless unless an organization is able to leverage these with (advanced) analytics to make better or faster decisions. Examples can be found across industries and across the value chain.
Why is data analytics an opportunity NOW?
The answer is quite simple:
Digitization of organizations
Making vast amounts of data available, which can be leveraged with analytics. This holds for the classical digitization of processes but also in new areas like health data (with health trackers), IoT and open source services (like PSD2).
Easy-to-use analytics tooling is available. To mention a few: DataRobot, H2O, Domino Datalab, Dataiku, and Sagemaker; all enabling automated machine learning. Be aware, a ‘fool with a tool is still a fool’ but these solution make ML much more accessible to a wide range of analysts and organizations.
Data savviness has increased in organizations. The possibilities of data for new business models based on data are increasingly known. You don’t have to study econometrics or computer science to embrace data analytics.
Lack of alignment with the business
Many research reports indicate that organizations are still unable to capture the potential of data analytics. A major reason is the lack of alignment between data analytics and the business. It is not the unwillingness that causes this misalignment.
On one hand, the business in typically unable to clearly articulate what they expect from data analytics. Often they expect the silver-bullet while the business challenge might be too complex to completely solve with only data analytics Or they mistrust data and the analytical insights at forehand, and rely on their professional experience and knowledge.
Only with an open mind-set and an iterative process between business and analytics, the best solutions arise
On the other hand, many data scientists get energy from building the most advanced analytical solutions without thinking about implications of using it in real-life. Only with an open mind-set and an iterative process between business and analytics, the best solutions arise.
Boosting the effectiveness of data analytics
Although the above findings are pretty universal and persistent, there are ample opportunities to make analytics work for your organization.
First, focus on people over technology. The challenge of alignment between data analytics and business is a people issue. Ensure that the technical people have sufficient soft-skills to listen to the business, translate the business opportunities and challenges into technical analytical requirements. Moreover, schedule regular meetings on progress and findings to enable the business to leverage this. If your organization is at a low analytical maturity level, don’t hire top-notch data scientist who are solely interested in the latest ML applications. Instead, hire analysts who can wrangle data, provide rudimentary insights and work with the business to use these insights for better decision making.
Second, focus on impact instead of the most sophisticated models. Although this sounds trivial, it is easier said than done. Most data scientist are more interested in building advanced models using the latest technology then ensuring that a model is integrated in an end-to-end process with a proper process to monitor performance and maintain the model. How to focus on impact?
Regularly define priorities for data analytics with management team. Reviewing the impact of past and current initiatives as well as estimating the impact of potential new initiatives is a key aspect of this process.
Be realistic of the contribution of data analytics for the problem at hand. In some cases, e.g. re-targeting models for digital marketing, analytics can be fully descriptive. While in other cases, e.g. decline payment due to fraudulent claims, require a human to review the case. The balance between analytical models versus human depends on the nature of the decision, the available data, the complexity of the decision etc.
Take an end-to-end view on the usage of the analytical insights. Actually capturing impact is as important as building analytical insights. E.g. predicting customer churn can help increase retention only if the organization takes actions on the prediction; a dynamic pricing model is extremely useful but only if it can be applied in the market.
Third, embedding analytics in the organization requires top-management support. They have to lead by example, foster understanding and conviction, build a re-enforcement mechanism and develop skills:
• Connecting the analytical teams with leadership teams helps to create mutual understanding of the barriers to fully capitalize on analytics.
• Sharing success cases will help to show-case the opportunities of analytics
• Trainings at all levels in the organization creates awareness, conviction and willingness to experiment.
Finally, analytics is not a one-off intervention but a transformational journey. Like with any transformation, it takes time and stamina to be make it successful. Or even better phrased: “There are eminently complex questions to be answered to fully utilize analytics. These issues will be resolved not by taking a conceptual approach, but by embracing an experimental approach. Those companies that take the plunge, test, reproduce their successes on a larger scale and learn from their failures could very well establish an unbeatable lead over their more hesitant competitors.”