Top 10 Do’s and Don’ts that Every Data Practitioner Should Follow

Austin Haygood Supply Chain Leader
Read Time: 5 minutes apprx.

Stock market price displayBelow are useful guidelines for how to build and maintain a data-centered process that can help drive analytical capability for you and your organization.


Businesses across all industries are pushing to grow their data and analytical capabilities. In the pursuit of a data-driven business, common mistakes can be made. To help you better navigate that journey, below are some tips I’ve learned along the way.


1.  DO determine your data strategy

 Before you do anything else, ask yourself these questions:

  • What is the solution or application you are trying to build?
  • Who will use the data?
  • How will you get the required data?

While keeping the above in mind, remember that it is helpful to have an adaptable approach and dataset. It’s very likely that over time your scope will change and that other projects will require new data.


2.  DON’T go big with your data before you are ready

 I get it, BIG DATA sounds great. But functional data works WAY better. Start small to help you build the process and identify exactly what you need. Once you can prove the concept, scaling it will be much easier.


3.  DO build collaborative capability

Most major analytical projects require input and approval from multiple groups and cross-functional stakeholders. So, why wouldn’t you want to build your data practices across that same organization?

In practice, it’s possible that certain data elements aren’t available in a central system or you may not know where to find them. Making friends along the way who want to improve your organization’s data practices with you is extremely valuable.


4.  DON’T be confined by a specific software or tool

hammerscrewRemember, these software solutions and ‘tools’ are supposed to be aids, not constraints. If you are running into issues, maybe you haven’t identified the right tool.  A hammer might be able to drive a screw through wood, but is that really the best tool for the job?

In fact, many of the tools today are more capable and easier to use than ever. Your software solutions need to be breaking down barriers for you, not building them.


5.  DO validate Validate VALIDATE!

Be your own biggest skeptic when it comes to your data and analysis. This is one of the most important disciplines you can have. There is no faster way to lose the credibility and the confidence of your managers than presenting bad data or the resulting invalid analysis.


6.  DON’T let bad data or records go unresolved

This means removing duplicates, understanding why you have nulls, standardizing your data formats, and maintaining your key fields. Consistent pruning of your data will ensure its effectiveness and accuracy while at the same time keeping it up to date.


7.  DO automate where it makes sense

Are you pulling data once a year? Month? Week? Day? Hour? Are you building a report that gets updated routinely? In my experience, the intention to automate is usually there, but it just seems to perpetually be just one task away.

Automating your data practices can be extremely valuable because it not only frees up your time, but it also enables more dynamic and real-time analytics. If you could spend a week designing a process that can automate what takes you 2-3 days to build every month, why wouldn’t you? Plus, it could allow you to take that monthly reporting down to the week, the day, or even up to the hour. Imagine how impactful that could be to your organization if you could see problems in your operations the moment they begin to happen. Now we are talking analytics!


8.  DON’T try to do everything in a single step 

For many reasons this is not a good idea and will inevitably leave you and others frustrated.

  • Remember you are building a process that you want to repeat, to modify, and possibly teach to others.
  • As you design the process, troubleshooting will be much easier if you only have to troubleshoot a single step versus a web of embedded logic and actions.
  • This step will significantly help with documentation. This might not seem helpful to you, but it will significantly help others, and likely your future self.


9.  DO think about how you want to share your data (and the resulting insights) 

guy computerYour data are only as powerful as the insights you are able identify and successfully communicate. Your audience is more likely to understand these insights if you provide a tangible medium like visualizations and metrics. Data visualization software can help you communicate your ideas better and also help you in your pursuit of automation. (Remember, lucky #7!)



10.  DON’T be afraid to learn new things!

This might be the most important item of all. The space of data handling/manipulation/blending/munging/etc. is constantly evolving and growing. This can be both exciting and a bit scary. That’s part of the reason why it’s so difficult to put just a single label on it.

The data businesses are currently capturing are only getting larger, more detailed, and more complex than ever before. As a result, there are innovations in software, strategies, and concepts to best complement this growing complexity. Find forums and partners to help you understand how others tackle these challenges and to find new solutions that could help you and your organization stay ahead of the curve.


This is a first article in a series of posts on how businesses are using their data. Next week, I’ll talk about the next big wave of software that is either currently changing or is about to change the way your team manages data and performs analysis. Stay tuned!