In almost every organization, some thought has been given to introducing data-driven work. After all, no manager likes to manage on assumptions and feelings, with a few exceptions. Choices are best supported by facts and figures. In recent years, Dutch organizations have therefore invested heavily in infrastructure and services to enable data-driven working. However, recent research by consulting firm Gartner shows that only 22% of the surveyed organizations know how to get concrete financial added value from the use of data, analytics and the deployment of AI.
- Many organizations fail with data-driven work because the strategic goal is missing.
- Success starts with asking the right questions, not technology or data volume.
- Small, targeted applications deliver more than expensive data lakes with no focus.
- Employees and data quality are crucial; governance and training make the difference.
As a data strategist at E-mergo, Louis de Roo speaks to many companies and knows where the pain point is: "Management wants to collect all available data right away and then expects the benefits to surface automatically. The focus is often primarily on improving reporting." With nice dashboards and reports, management thinks it can get a better grip on the money flows within the company and thus achieve efficiency gains. This is disappointing. "Apart from financial service providers for whom finance is the core process, in most organizations the administration is not the process by which they can distinguish themselves from the competition. You do that by increasing your own competitiveness or by achieving more margin. Ideally, of course, you do both."
In search of strategic added value
To reach that point, management must go back to the drawing board. With good questions and in-depth discussions, a clear picture emerges of the activities with which the organization actually adds something to what is available in the market. At Bol.com or Amazon, for example, it is clear that the size of the customer base allows these platforms to advise customers on a follow-up purchase with each purchase. "Those who bought this, often ordered that too," it says under all shopping baskets. Not every store or manufacturing company has that capability, but company data can show, for example, the minimum quantity of products delivered each month. This can be used to negotiate more advantageous purchases in contract negotiations with suppliers.
There is no universal rule; each company must get clear internally which insights from data can truly deliver strategic value. Then the data sources must be sought for creating those insights. "Start small; it is not necessary to create an entire data lake to achieve demonstrable results," de Roo argues. This sometimes results in embarrassing situations when it turns out that the investments previously made in cloud storage and applications were actually for naught. "Some organizations do indeed decide to phase it out, and I don't yet know of a case where management later regretted it. Still, many organizations keep their expensive datalake on hoping to benefit from it someday," is his experience.
It is not necessary to build a whole data lake to achieve demonstrable results."
Data quality must fit the purpose
Far more important than the amount of data collected is attention to the quality of the data. It should be kept in mind that 100% correct data does not exist. De Roo: "What matters is data quality that is good enough for the purpose. It makes a difference whether you need to create a construction drawing for a nuclear reactor based on data or whether you want to discover a trend in the sale of sports socks." In the former case, the data must be correct to many decimal places; in the latter, it does not matter if it contains a few input errors. However, there must be a clear picture of the bandwidth within which each data stream should be allowed to vary. Establishing those bandwidths is a key focus for management early in the journey to successfully implementing data-driven work. "That's part of the data governance policy. There should also be signaling when unexpected values occur so you can respond to that if necessary." According to de Roo, there are enough statistical tools available to implement that data quality policy automatically.
Recognize role of employees
Technology has a clear role in data-driven work, but human creativity is leading the way. This is important not only at the beginning of the journey in answering the questions of what you want to find out by using data and why. However, data-driven working is an ongoing process to which everyone in the organization can contribute. Employees often see new possibilities and opportunities. That's why management should pay close attention to training on how to use the tools and create the opportunity to experiment, de Roo believes. "It's about democratizing the data. Familiarize employees with their own dashboards and show them how to add new features. That should also be part of onboarding new employees and periodic in-service training." Democratizing data also means giving employees a responsibility in working with the data. "That too is part of data governance; that everyone knows how you handle personal data and other sensitive information. Especially now that more and more AI tools are becoming available with which you can easily perform very large operations. Then you can also do great damage. That awareness should always be addressed in training sessions."
It's about democratizing data. Familiarize employees with their own dashboards and show them how to add new functions."
Good support secures trust in data
Good support for the data environment is also a prerequisite for the success of data-driven work. Responsibilities for managing the data and updating the software, replacing certificates and applying patches must be clearly assigned. "If you don't think carefully about how you secure that, you dig a trap. You then know that things will go wrong at some point, and if it turns out that the data has been incorrect for some time, the trust in the data within the organization is immediately gone. The more dependent people are on a dashboard, the shorter it takes them to go looking for a workaround."
Despite good support for the data environment, crucial components can fail temporarily or for long periods of time, for example due to cybercrime or geopolitical developments. That's why it's important to properly assess the dependency of the systems . De Roo: "Actually, you then go back to the beginning: determining what the key process within the organization is. On that basis, you create a minimum viable product of a new system, first on paper. That means that from a data perspective you actually create a process that you can also do with pen and paper. In that case, life becomes extremely uncomfortable and unpleasant, but once you have a clear understanding of what the basic process is, what questions absolutely have to be answered in it and what decisions play a role in it, you also know how to do it without digital tools. You hope, of course, that it will never be necessary."
Not another data platform! How to give data a place in the organization. The apt title of Louis de Roo's lecture. Would you like to know more about this and steamline data in your organization? Then come on Wednesday from 12:45 - 13:15 to lecture room 6. Be there and order your free tickets here.
