Leaders shared lessons learned on their varying paths to data maturity, including: When, and how, to relinquish control; the importance of outlining the analytical process; and how encouraging a culture of reuse and repurpose can help to accelerate the journey to maturity.
When to relinquish control
Ultimate control can be a difficult thing to give up, particularly for data leaders still wary of their organisation’s lack of data maturity. Doing so is essential for anyone looking to ensure that responsibility for the analytical process is democratised throughout the organisation; one of the pillars of data-maturity.
Inevitably there will be occasions in which people and processes slip through the cracks, but this should be expected within any realistic journey to data maturity. By giving in to instinct and rushing in to pick up the pieces, for example by cleaning and remediating low-quality data, data teams can’t expect to breed the literacy and accountability needed to ensure a sustainable flow of high-quality data from the business. As one member in the pharmaceutical sector put it: “Controlling everything doesn’t breed a data-driven approach.”
The problems posed by relinquishing control are heavily dependent on the size and data maturity of an organisation. One DataIQ Leader present at the roundtable discussed how their relatively data-mature organisation’s federated model – wherein analysts sit within lines of business, rather than in a central hub – made it essential to implement a framework to outline accountability, as well as standardisation to prevent silos.
The member said: “While encouraging innovation at the edge, we’ve standardised things such as discipline and created report templates to breed a common data culture wherever possible. We’ve also launched a data academy, which brings professionals together from across the business to encourage standardised ways of working.”
Innovating at the edge means taking advantage of non-business as usual opportunities to test new analytical techniques. This can only really happen in organisations with a robust data culture and clearly defined parameters for the use of, and accountability for, data.
Venturing down the path to data maturity requires clearing barriers such as data silos, poor communication and poor levels of general data literacy. In fact, DataIQ research reveals that poor data literacy and data silos are the most prevalent threats to the effective consumption of data within organisations.
One member from the retail sector discussed how their organisation’s lack of data maturity means they aren’t quite confident enough to relinquish control just yet. The member said: “Because of our immaturity in the context of data, we can’t just let everyone crack on and do their own thing. Silos persist, there are different levels of expertise across the business and a general lack of communication. If we relinquish control now, it’ll be very hard to get everyone pulling in the same direction. This all comes down to culture.”
While context will change as organisations move up the maturity ladder, data teams at all levels of maturity need to remain mindful of the importance of data culture. If that culture is neglected, silos and bad practice can quickly reappear, even within organisations with advanced levels of data maturity.
Outlining the analytical process
The analytical process can often be a closed book to those sitting outside of data departments. This is influenced by two key factors: The data literacy of stakeholders sitting outside of the data department, and the ability of the data department to communicate with those stakeholders.
While data literacy rates are slowly creeping up, in part thanks to literacy bootcamps and programmes that became increasingly popular during the pandemic, DataIQ research shows that only 36% of organisations have a developed or very high level of data literacy. Meanwhile, 46% of organisations are investing in soft skills development for the data department. This is a virtuous cycle.
When data departments can, and want, to communicate the analytical process to the broader business, they can also clearly demonstrate where responsibility for data ownership in the context of that process lies. In turn, by taking a looking “under the hood” business stakeholders gain a broader understanding of processes, thus heightening their data literacy.
One member organisation has taken to creating a seven-step guide for analytical product management. The guide demonstrates not only the steps involve in any process, but the role of the business in each of those steps. “The steps probably took me about an hour to create, but it was about a year before it really took off,” said the member behind the process. “Key to this was convincing the board of the process’ value and begin advocating for it.”
Such processes are valuable tools for data teams looking to cultivate a greater sense of responsibility for data beyond the data team. By articulating the process in a digestible manner and ensuring buy-in, the process can become a powerful tool for signposting where, and why, projects have halted or been delayed, and where the responsibility for overcoming that delay rests.
Accelerating through repetition
At the more advanced end of the data maturity spectrum, organisations are considering how best to encourage their analysts to create models that fit to a framework and can be reused and repurposed.
In the past analytics professionals had been inclined to build their own unique models. This was partly down to a lack of trust, with analysts being reluctant to risk relying on the work of others. Value also played a key role, with many analysts believing it easier to demonstrate their own value and expertise if they had built a model from scratch.
This paradigm can be a problem for organisations, particularly when models are bound by the data they use, thus limiting reusability. In a time when demonstrable value is key to the uptake and acceptance of data, replicable model building is vital for any truly data mature organisation.
The mantra of “build once, use often” is increasingly in vogue, and leaders are working on methods of encouraging their engineers and scientists to adhere to it. One member is looking at developing a compensation framework, whereby analysts are rewarded financially according to the amount their model is reused.
This type of issue may sound a million miles away for organisations still trying to embed a basic degree of data usage into the business, but by encouraging a “build once, use often” culture at the onset of the journey to data maturity, organisations can accelerate more quickly up the maturity chain as they build a backlog of replicable templates and models.
Key takeaways
- Resist the urge to take control. Data maturity advances when the responsibility for data is democratised throughout the business. Shifting responsibility from the data department requires data leaders resisting the urge to take control when people and processes slip through the cracks.
- Walk the business through the analytical process. By outlining the analytical process in a digestible format, data teams can highlight where responsibility and accountability for the stages of that process lies.
- Build once, use often. Rewarding analysts that build reusable models is a valuable tool for organisations looking to quickly accelerate up the maturity chain.