In today’s world, more data is being created from more systems than ever before. The walls between those responsible for managing the data and those who need to use it to generate insights need to be broken down. By moving from traditional models to a more democratic one, we can level the playing field.
There’s nothing new about data-driven decision-making. Due to increased volumes of data generated and decision-makers need for information, the term Business Intelligence (BI) gained traction in the 1980s. This surge of data and the need for answers resulted in the increasing usage of data warehouses, ETL processes, and visualisation. Data analysts were hired to support these processes and provide answers to those who sought them. Due to this traditional model, information was difficult to get to decision-makers as the development of custom static reports involved a lot of business logic and long lead times.
When organisations recognised the benefits of having these reports they may have disestablished the data monarchy and introduced a self-service business intelligence tool. Almost every organisation now has some form of self-service tool even if isn’t used by everyone. This is a good step forward, but just learning the tool presents a steep learning curve.
Some organisations have broken down the walls further and embedded data analysts in teams that require the most insights. The analyst reports to their functional manager and has a dotted line to the centralised data or business intelligence function. Analysts who are embedded gain an understanding of the stakeholders they serve and the subject matter they handle. It seems like a win-win. The tradeoff in this model is that the tools remain in the hands of a few. Despite being empowered to provide excellent service, analysts may become a bottleneck with increasingly complex requests.
In a work environment where the analyst, functional manager, and business intelligence manager are in constant communication, work should flow smoothly. By working together with their peers, analysts can form a community while solving problems and becoming subject matter experts.
Communication ultimately poses the greatest risk. If this communication fails, there will be data anarchy. There could be a situation where existing or new teams are hired to handle the increased number of requests. Without an established relationship with the corporate business intelligence team, bad practices might creep into the organisation. Analysts might be moving data out of approved systems, releasing statistics without approval, or becoming a “shadow BI” function.
An organisation’s management or structure may also pose problems if it does not provide analysts with the tools they need. Training or tools that are restrictive could lead to workarounds if neither are provided. At best, this may be bad practice, and at worst, it could compromise security. Complying with privacy legislation is more important than ever, so getting this right needs to be a priority.
The aim of a data democracy is to get insights to those who need them quickly while ensuring security and governance. Training, tools, and trust make this possible. Instead of simply serving as a ‘data vending machine’ or resource, a data analyst should serve as an educator.
It does not mean that everyone should have access to everything. To play by the rules, teams must judge when to share results and when to seek guidance.
The way to start your data democracy the right way is to create governed datasets and train power users in best practices. This will help the teams find creative solutions while giving them an appreciation for where their data comes from. While on-the-spot training can help new users get through the first few months, it is not scalable. It is much more scalable to train power users to champion the use of data in their teams.
Some strategies to get started could be:
- Grouping datasets so teams can access only what they need.
- Preparing short videos with clear instructions on how to get started and use the datasets.
- Making use of a data dictionary that describes the dataset and provides examples.
- Training power users together and providing them with new information regularly.
- Providing data visualisations that encourage exploration and can be repurposed.
- Setting clear expectations about data governance responsibilities and putting a framework in place.
Insight-seeking organisations should prioritise this shift in culture. Although it may not be a quick process, the move toward data democracy will be worthwhile.
It is possible to achieve data democracy without sliding into anarchy. When data silos are broken down, teams can better understand their customers, products, and industry.
- 3 Best Practices for Becoming More Self-Sufficient with Self-Service
- Don’t expect a data democracy if you don’t participate
- Death to Dashboards, Really?
- Building a Data Democracy