Tim Mitchell
Follow Tim Mitchell on Twitter  Like Tim Mitchell on Facebook  Subscribe on YouTube  Connect on LinkedIn  Subscribe to the Data Geek Newsletter

When Best Practices Aren’t Best

bestpracticesMost everyone is familiar with the term best practices. This phrase describes an industry-specific design or methodology that can be applied to most problems in a given domain. Data professionals, carpenters, medical practitioners, retail professionals, hair stylists, and most any other craftsperson will learn through experience a set of rules that usually govern how to do a thing. That holds true here as well in my little corner of cyber real estate, where I’ve been compiling a set of ETL best practices over the last few years.

Best practices serve a number of functions. They help to guide those new to the trade by defining useful solutions to common challenges. They reduce the need to “reinvent the wheel” by establishing tried and trusted methods. Best practices can also streamline the documentation and knowledge transfer of a project by using (mostly) standardized approaches which are often well-defined in the public domain. Finally, the use of industry-accepted best practices can make for a more supportable product, reducing the amount of specialized training required when hiring experienced professionals.

When Best Practices Aren’t Best

I have to admit that the term best practices feels increasingly awkward to use. In my early days as a technical professional, I was far more liberal with absolute terms such as always and never, and would sometimes describe a methodology as “the” way to solve a problem. The more I learn about the business of data, however, the more I’ve come to learn that there are very few absolutes. For most every best practice, there are myriad exceptions. There is no all-governing list of methods that one should (or should not) use to solve technical or business problems.

The most practical use of best practices is to understand why each one is accepted as such. While it is useful for professionals and craftspeople to have a set of best practices for common problems, being successful is far more than possessing a canned list of how to do a few things. When building anything from custom cabinets to custom data warehouses, best practices will only carry one so far. Before blindly following an industry-accepted best practice, be sure you understand the benefits and risks that come along with it.

Best Practices for Best Practices

With the above caveats in mind, here are a few, ahem, best practices for applying best practices.

  • Keep in mind that best practices are tribal knowledge. They are rarely codified, so they’re not set in stone.
  • Consider the source. The phrase best practice is thrown around liberally, and one person’s best practice might be another’s nightmare.
  • There’s always an exception to that best practice. Yes, even for that everybody-knows-it’s-true standard.
  • Don’t be afraid to question best practices, or to define your own.
  • The only immutable best practice is that you should always use the gray matter between your ears. Best practices != a checklist of things that must always be done, so use your intuition and experience to figure out the best course for the factors at hand.

Having a good set of best practices can make you a more effective professional, but don’t rely solely on those loosely-defined rules when making decisions.

About the Author

Tim Mitchell
Tim Mitchell is a business intelligence and SSIS consultant who specializes in getting rid of data pain points. Need help with data warehousing, ETL, reporting, or SSIS training? Contact Tim here: TimMitchell.net/contact

1 Comment on "When Best Practices Aren’t Best"

  1. I really prefer the term recommended practices, though it doesn’t have any traction.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: