“Data is the new oil.”
“Data is KING.”
“Forward-looking organizations make data-driven decisions.”
We hear stuff like this every day. Indeed, very few organizations in this day and age would argue against the importance of data. The data analytics market is booming and is forecasted to grow to $40.6 billion by 2023, and that’s excluding data visualization and discovery tools.
Data is the gold rush of the 21st century, so it comes as no surprise that many organizations dive right into it without thinking through and setting up protocols for data management, but winging it instead. Having worked with dozens of teams on hundreds of data projects, we started noticing data mismanagement patterns. Here are just a few common mistakes companies make, plus practical advice on how to avoid them.
1. Allowing technological choices to dictate data management policies
Buying off-the-shelf data management and analytics solutions presents two challenges: 1) there are too many choices, making it difficult to identify the right one, and 2) having to adjust your needs to the capacities of a particular solution. Furthermore, many purchasing decisions are made based on factors such as price and functionality. Data management processes tend to become an afterthought. As a result, instead of finding a solution that matches your data management protocols, you adjust to the existing protocols of the solution you purchase.
Create a list of all your data management requirements. How many levels of access would you want to set up? How do you protect data online? What is the process of removing data from the system? What other solutions do you want your system to integrate with? Do you need to compartmentalize your data? If you are based in Europe or work with European clients, how do you ensure compliance with the GDPR? Create your own data protocol and then find a solution that ticks off all of your requirements. Or hire a data specialist to help you with the process.
2. Focusing on data volume instead of data quality
Bad data is like mold. You may not notice it at first, but it will grow and infect your entire data set. We see so many organizations greedily aggregating data, failing to set up quality control and filtering systems. It’s also a mistake to assume that only bigger companies with access to large volumes of data that make this mistake. Small organizations can just as easily fall into the trap if they don’t develop a proper data culture from the start. Insights derived from bad data are corrupted. According to research, dirty data has a direct impact on the bottom line of 88% of companies, resulting in 12% revenue losses for an average company.
If you are a small organization and your data size is still manageable – create data filters from the start. Run regular audits and clean incoming data. If you are a larger company or if you already have considerable amounts of data, hire a data professional to audit and clean it as well as to create data input rules. The longer you wait to address the issue, the less useful your data becomes.
3. Failing to future-proof data
Say you purchased a data management and analytics solution. Up until now your data was kept in a spreadsheet, so, naturally, you import it. For your immediate project you don’t really need all the data in the spreadsheet, so you only import the columns that you need at this very moment. A couple of months later you have another project, so you go back to the spreadsheet and import other bits of information. And on and on it goes. If we had a penny for every time we witnessed a similar scenario…
This results either in data duplicates or numerous workstations where different bits of data from the same data set are stored.
We will give you the same advice that your mom gave you when you were a kid. Never put off till tomorrow what you can do today. Profile your data today to make it future-proof. For example, say you are adding a customer to your system and that’s the first piece of data you have. Information on the customer includes his name, gender, and email address. You create three corresponding tags in your system. What you should do instead is think of as many possible usages of your data in the future as possible and create a list of tags that you may find useful in the future. For example, is this a returning or a new customer? Since when was this person a customer? What’s the size of the purchase? What was the purchase? How many times did this person visit your website before making a purchase? This will make your data more productive in the future.
4. Storing data until the end of time
Sure, identifying trends and being able to predict patterns require actually accumulating data over a period of time. However, many organizations take it to mean that data should be stored forever. In fact, 63% of organizations report an average of 20% annual data growth. The system speed starts to slow down, storage size turns into a constant issue, and the idea of ever moving to a different system becomes inconceivable. There has to be a balance between storing old data and keeping a data management system lean.
- Archive your data. This way you do not delete it forever, but remove from the system and store in a much larger size.
- Update it on a regular basis. Emails to a particular customer have been bouncing back for the last month? The email address is probably no longer valid. Remove the person from your system.
- Let it go. Do you have data that the system shows hasn’t been accessed or used in 5+ years? It’s probably time to part ways.
5. Being extreme with data access
Some companies are extremely strict with their data management access, while others give it out like candy on Halloween. Both are a big mistake. The former leads to data silos, significantly limiting the capabilities of data. The latter leads to serious threats to security and private data being exposed.
- Set up time-limited access. If you have many people within your organization who require access to data, it’s easy to forget to disable that access when needed.
- Create multiple levels of data access. The ‘admin’ level of access should be granted very carefully and to a very few employees.
- Never set up access with employees’ personal email addresses. This should be self-explanatory.
- Create a lego-type of data system. Create a system that makes it easy to compartmentalize data and give access to only certain parts of data.
This is the first part of the common data management mistakes that we’ve witnessed over the years. Make sure to check back in for part 2! In the meantime, share your experience with us in the comments section below.