What to do to speak the same language as your colleagues from the company? What does it mean that a company is data driven? We present 5 principles that will allow you to get answers to these questions and make your company a data-driven company.
A data-driven company is a company that builds its business decisions on the significant use of facts collected in its own resources, i.e. primarily in databases. By analyzing in detail the trends and real reactions to any problems and threats, companies operating in this way are able to work much more effectively.
1. Make your data and the resulting facts a priority
Every company uses industry-specific business terms that appear at many levels. Regardless of whether we are talking about global concepts or focusing on phrases related to a specific activity or department, it is important to speak a common language at each of those stages. To many people this approach may seem natural, but practice shows that this is not always the case. Every detail is important in communication. Sometimes there are also conflict situations in which “my right is better than yours”. This may sound familiar to all of us, not only in the context of cinematography. Therefore, the first condition for mutual understanding is to base the company on data and the resulting facts.
An important issue is the detailed use of available information by all employees, allowing it to be viewed as globally as possible, and not only locally, from the perspective of specific individuals and for the purposes of their narrow objectives. Since we have knowledge, let us share it in order to develop common conclusions. Regardless of whether we are talking about the IT, PR or even the management board.
2. Create a common vocabulary of business terms across your organization
One of the foundations for proper data management is a common language. The available information must be shared and interpreted according to the same patterns. In order to achieve the best possible results, we must avoid a situation in which, at different levels, we define the same business concept differently. This is the basic problem of companies that begin their adventure with the broadly understood Data Governance.
Examples where the same concepts are interpreted differently can be listed endlessly. This even applies to such fundamental data as basic financial indicators. We will look at this problem in the telecommunications industry. From the definition of the customer, going through the service and ending up with the household – all these issues can be analyzed differently by the customer retention, marketing or finance departments. In telecommunications, the customer service department may interpret the user as the person on behalf of which the contract is signed, while the marketing department counts each telephone number separately under the contract itself. Where the truth lies? This is the issue that needs to be jointly established, accepted and thus, as we said – a vocabulary of business terms must be developed. By using it, we can be sure that each employee interprets the household in the same way, understands invoices and other seemingly obvious aspects in the same way.
3. Take care of data quality in the same way as you take care of the documentation
Confirming business concepts is, of course, just the beginning of the road. Once we know how to interpret individual data and adjust it to the approved formulas, we will need to ensure that our analyzes are based on good quality data. Data Quality is another point that is inseparable from data management. Without the certainty that the data is complete, consistent and accurate, there will be a high degree of risk of creating incorrect reports and drawing incorrect conclusions. That is why quality testing is so important, as it will allow us to react quickly in the event of detected irregularities. Already at the moment of loading the data, we can perform basic verifications, and we actually should do it. After all, the mere fact that the process was successful does not necessarily mean that the data we received is of good quality. Sometimes it happens that the data is not sufficient, key attributes are missing or they just differ from the other system.
4. Verify data, because who said that data verification must be boring and arduous?
Ok, but how exactly should we test the quality of the data? First of all, wisely. Creating metrics can be laborious, but we are able to automate everything, which is undoubtedly such a wide topic that it is worth devoting a completely separate article to it. Nevertheless, it is worth looking at the individual errors from the side first and choosing the most popular patterns. In this way, we are able to catch the most common problems that annoying users in one fell swoop. In addition to the basic metrics that verify the quality of data, we can also use the latest technologies and apply AI, among others. It is true that artificial intelligence is definitely a more advanced solution, but it also gives us the opportunity to quickly “track” values that differ from the defined set. Let’s add to that data profiling and, as a result, we get a comprehensive tool that, once prepared, will guarantee a significant increase in quality.
5. Look for data problems where they occur most often, later in the process
Regardless of whether we are talking about data management or its quality, the key is the cooperation from the beginning to the very end of the departments using individual systems in the company. After all, the path that each piece of information travels is so long and bumpy that the more you get into it, the more complicated it becomes, and potentially more risks appear. Let’s give an example to illustrate what we are talking about. The consultant must complete the customer’s address, therefore he/she enters the street name, building number and suite number in the following format: Spokojna street 4 suite 2. It then turns out that due to the lack of an established standard, this information in other systems, relating to the same customer, was assigned to the wrong household, as this format was not handled correctly. Let’s try to determine the cause of the problem. Is it the consultant’s fault? No. Is it the fault of the system? Neither. The cause of the error is the lack of a standard, which could only be noticed during specific analyzes, when many calculations took the provided information for granted. That is why every aspect related to data should be consulted within the various points of common departments, i.e. their intersections and at different levels. By gaining greater certainty as to the information received, we can expect much better work results.
Łukasz Pająk, Senior Programmer / Designer
Łukasz has been working with data since the beginning of his professional career. He is closely associated with the telecommunication industry, where he cares about Data Quality & Data Governance, and at the same time about a good working atmosphere. Privately, a huge fan of new technologies, automotive and unconventional solutions.