Organizations need qualitative data, an inclusive data strategy, and open communication channels between the IT and non-IT teams to leverage modern data solutions like data mesh. But how do you maintain data quality and build data trust? How do you remove the disconnect between IT and other teams? And more importantly, how do you devise an effective data strategy?
Dacil Hernandez, Director of Data & Analytics at Nagarro, answered these questions in her recent talk with data mesh evangelist Scott Hirleman. She emphasized that tech is easy; it is more difficult to align your people and processes to data needs and create value using the existing data sets.
“Everyone keeps saying ‘business first’ and people first, but they continue to focus on tech instead,” says Dacil. This makes it essential to define what your organization means when it says, ‘business first.’ It is equally important that your business teams, finance, marketing, and IT work together.
Traditionally IT has been considered the ‘data guardian’ in most organizations. But often, non-IT teams are the actual data owners creating a need to identify data owners and educate them on their roles and responsibilities. Data ownership entails much more than just keeping the data. You must also ensure the data is qualitative and easily accessible.
Another challenge organizations face concerning data is that teams are apprehensive about owning the data, as some consider it a liability. “Data ownership should not be treated like a hot potato that no one wants to claim responsibility for,” says our in-house data expert. She adds that it falls on the business leaders to lead data conversations, understand the tech perspective, and further communicate it to their respective teams. Meanwhile, data teams must improve their understanding and capabilities around data ownership.
Once data ownership is established, you can start laying the foundation for a solid data strategy by asking the right questions and stating the data requirements upfront. The teams owning and processing the data must ask the right questions to establish the business need and what dataset serves it best. This should be complemented with forums where all the stakeholders can discuss, debate, and decide their data needs collectively.
Dacil says that talking to potential data consumers will help them understand the context and makes room for suggestions and insights. For instance, there may be instances when an alternate data source addresses a particular need better than the dataset initially considered.
We have distilled some insights from their conversation in this article.
A well-rounded enterprise data strategy ensures that all the stakeholders are involved in its conceptualization and understand their roles clearly. Dacil lists out the following points to remember while building a data strategy.
- Incentivize data sharing: You must lay down engagement rules that encourage transparent and proactive data exchange. A good, top-down data strategy and understanding of the term ‘data-driven’ is critical. An effective data strategy encourages stakeholders to participate in data sharing and processing. To ensure that, you can incentivize teams to share their data with other teams and departments. “You can also create a ‘fear of missing out’ by educating and informing your people on the benefits of data sharing,” says Dacil.
- Data agility: Another critical aspect while drafting a data strategy is making sure that data can be tweaked to meet the business needs promptly to avoid any substantial loss.
- Maintaining data quality: Modern data solutions like data mesh and analytics offer good results only with qualitative data. Dacil brought up an example where a team had data on phone numbers, but most of these numbers weren’t real. So even though there was data, it wasn’t quality data and barely served its purpose.
She suggests it also helps to keep the data value in mind when assessing data quality. “If you have something that’s low quality but high value, then you need to work hard to improve that quality,” says Dacil.
From a cultural standpoint, you need to make sure that identifying data quality issues doesn’t turn into finding faults and placing blame. Instead, you can adopt a feedback-oriented approach and use gamification to make the process fun and interesting. This will also help increase data literacy and trust across the data users and owners.
- Building data trust: Data trust is the key for any organization aiming to become data-driven. Besides maintaining data quality, the IT team must identify any existing data issues within the organization. If the business teams discover data issues every time, it will hamper their data trust. To avoid this, the IT team or the data owners must ensure that they find the mistakes first, identify the root cause and take corrective measures, so those mistakes aren’t repeated.
One small change at a time
If you aren’t happy with your current data strategy, you don’t always need big-bang change. “Instead, you can break your challenge into smaller milestones and take small steps to achieve them,” says Dacil. She adds that defining these milestones helps plan and prioritize the most significant challenges and address them first. It’s best not to attempt to solve everything at once. She strongly suggests bringing in an expert who can help you identify the gaps in the existing data systems and build a roadmap for your organization’s current and future data needs.
To data-mesh or not
It is an exciting time in the world of data with various modern data solutions. However, not every data solution is suited to your needs. There are a few questions you must ask before you decide to invest in data mesh.
Is centralization a bottleneck, or is it just about trying out a new trend? A data mesh approach requires maturity to manage federated governance and decentralized data ownership. So, make sure you understand the reasons behind your data problems. Ask the expert if data mesh is the right way for your organization. It is a must to consider your challenges and maturity levels before deciding on a data solution like data mesh that involves considerable cost and realigning your people processes to the new system.