Ten best practices on becoming a data driven company

It is common sense how important data will be in future. In fact, mastering their data assets is mission critical for many companies. Companies that invest in embedding a holistic data strategy are more likely to succeed in their digital transformation. Many companies have been on that data journey for some time now – It is time to have a look what has worked well. Based on our joint work with our clients we have compiled 10 best practices to share with you within this post. 

By Jens Müller-Iden and Joachim Engesser

1.   Prioritize a data vision which is more than an empty promise

It is quite simple to proclaim ?We will become a data-driven company” and to start respective activities. At the same time other strategic initiatives and/or operational topics will pop up and trade-off decisions need to be taken. And this is the litmus test for the data strategy. Only if the majority of those trade-off decisions are taken in favor of the data strategy there is a good chance for successful implementation. Of course, there is room for shaping your path. But a data vision will only be strong with top management buy-in and the required investments priorities. 

2.   Focus on relevant business use cases and make the impact transparent

In order to get the buy-in for prioritizing the data strategy it is crucial to make the impact transparent. This includes the positive impact of implemented initiatives but also the negative effect of not doing specific things.

The easiest way to do this is to pursue a value-driven approach and strictly link activities to business use cases. Those use cases will help you to achieve immediate impacts. Additionally, it is a superb vehicle to prioritize different data activities among each other (effort vs. value creation).

Based on the business use cases the data scope within the data lake can be gradually increased. This ensures that all data can be managed in an appropriate way (e.g. monitoring of data loads and issue resolution, data quality checks) and the data lake does not turn into a data swamp.

3.   Make your scarce talent happy

It is not really surprising that data scientists are scarce at these days. For many data scientists working on exciting content is even more important than compensation packages or career opportunities. Therefore, it is crucial to build a working environment which attracts talent. It is especially important that these talents are not scattered across the organization. Building a center of excellence for example allows to mutually inspire themselves and to learn from each other. Furthermore, it will be extremely difficult to hire all required experts from the market. Therefore, it is often beneficial to launch a training program for internal expertise development.

4.   Don't get lost in the data governance jungle

The available work within data governance is infinite. Hence, it is absolutely crucial to focus your governance efforts on the most value promising activities. A productive way is to strictly align all governance activities along the prioritized use cases (see #2). For those use cases the governance work is in theory pretty simple: you just need to ensure that the required data is available in an appropriate quality. In practice you will face huge challenges esp. aligning people across various units. Therefore, you need to ensure that you have an adequate mandate and the full buy-in of at least one C-level executive.

5.   Accept deficiencies in data quality

Data quality is by no means unimportant, but it is not decisive for all data entities and attributes. This is highly depending on the respective business use cases. Therefore, it is crucial to understand for which data fields quality matters most. Put your full focus on these fields and accept shortcomings in the remaining ones.

Making data quality transparent is a key driver to achieve a long-lasting data quality improvement. It is good practice to define data quality KPIs; also known as Key Quality Indicators (KQI). A continuous measuring and reporting of these indicators force all stakeholders to work on further improvements. 

6.   Be brave with data privacy - be cautious with security

Obviously, you need to adhere to data privacy regulations (breaking GDPR for example may become very costly). However, we want to encourage you in being brave to get the customer opt-in for the data processing. For example, when registering an account at the large tech giants accepting their broad data privacy terms is mandatory. Another option is to create a “win-win”, e.g. by tying loyalty program benefits to an opt-in.

In any case, customers expect treatment of personal data with high care. Ensure to design and implement a holistic data security concept for minimizing the risk of data leakage and a significant company reputation loss. 

7.   Find smart ways for the necessary legacy architecture transformation

The need for a modern IT architecture (modular, flexible, scalable, fast, value-adding) to become a data driven company is quite obvious. Nevertheless, it is not possible to replace the legacy IT architecture with a big bang. Separating the data in a dedicated data layer is a key enabler for a stepwise transformation. Within this data layer hybrid setups of legacy data warehouses and new data lakes are common, ensuring a manageable complexity and leveraging synergies of pre-built standard solutions. Key data services are exposed, and key data elements are made available for a broad audience.

8.   Scale before the tipping point

Delivering fast and tangible results is inevitable to keep the buy-in for the data initiative. Especially in the beginning this means that architectural workarounds and related technical debts need to be accepted.

In order to avoid a constant increase of the technical debt, dedicated activities for increasing architectural maturity need to be launched. These activities take care that a scalable and sustainable architecture will be built up and technical debts will be deleveraged. Therefore, a large portion but not the full development capacity should be assigned to implement business use cases.

9.Win the minds, hearts and hands of your people

Don't forget the people. You'll need the buy-in of the organization to succeed with your data initiative. This requires in almost all companies a comprehensive change management addressing all three engagement dimensions:

• Mind: Promote the business value and other rational advantages
• Heart: Generate passion and enthusiasm along the journey
• Hands: Provide the necessary training that everyone is able to successfully execute his/her job

10. Do not shy away from learning from others

A lot of companies have designed and implemented data strategies in the recent years. Some best-in class players are already at their second or third main revision, adjusting to experiences and new technical developments. Depending on your data strategy maturity, make use of these experiences to avoid typical pitfalls and fast forward your own approach.

Conclusion: Focus, Focus, Focus

The data playground is huge. There is a high risk that you can get lost in almost every dimension mentioned above. Therefore, it is crucial that all activities you initiate should be focused only on value creating aspects. Data itself has no value. Only if you gather, prepare, analyze or use the data to support a clear business use case you can be successful with your data initiative.