5 Rules for Managing Your Database Gremlin

This past weekend, an oldie but goodie was on the television:  Gremlins.  Remember the three rules?  No bright light, don’t get him wet, and never feed him after midnight.

When it comes to CRMs, AMSs, or your basic Excel workbook, it seems that everyone wants to break the third rule, feeding the database gremlin.  That puts you as a user of the data in a position to not follow the first two:  working dawn to dusk, not seeing the light of day, and feeling like you need a shower after the mental workout that is trying to make sense of all that data.  Okay, maybe that’s an extreme example, but nevertheless.  Data has become unruly in many of our client organizations.

When trying to make sense of the data, the issue is less about the lack of internal expertise or skills and more so the overwhelming amount of information being collected.  Where do you start?  Where are the variables which can provide unique and valuable insight into your customer or member base, and where do you tend to waste a lot of time?

Not to be outdone by the Gremlin rulebook, here are five of our rules for managing your unruly database and getting a fresh look at your customer or member information:

  1. Connect the dots. If you are in an organization that relies on multiple data sources, you may be at risk of having tremendously valuable information…that is pretty much useless to everyone else.  One of the primary fail points in data mining is that there is not a unique customer or member ID that ties together multiple sources.  You may also have a “grandfathered” set of IDs from a previous system, that are not matched with the new IDs.  While we may not recommend purging the unlabeled data, the overall usefulness and power of the information will be significantly less.  Going forward, make sure that any data source contains a consistent unique identifier that can tie back to a master data file.
  2. The glass 80% full. Depending on the size of your database, you will need a comprehensive and representative set of data points to get conclusive insights from it.  As a rule of thumb, if you are missing or questioning the accuracy of more than 20% of the data for a particular variable, there is enough risk to warrant avoiding that data.  Hide the data until you can update or complete it, or eliminate it altogether.
  3. Check your expiration dates. For most dynamic data points (moment-in-time relationship measures, transactional trends, competitive assessments, variable individual/business descriptors), making decisions on information collected more than 3 years ago should be considered out-of-date.  Throw it out!  You can make the argument to question data even more recent than that, if your market or customer/member population is changing at a faster pace.
  4. What you don’t know shouldn’t hurt you. If no one in your organization knows how and why a data point was captured, or how it should be interpreted, don’t risk assuming you know what it might be telling you.  Don’t be a superhero – consider hiding or purging the data so that you aren’t tempted to make something out of what may amount to nothing.
  5. Adopt (and enforce) a labeling standard. If you find yourself asking questions related to #4 far too often, you may want to consider developing a set of labeling standards.  Don’t go overboard – if Twitter would warn you on your character limit, you’ve gone way too far.  Think about the major categories of information that you collect (account management interactions, revenue transactions, customer service contacts) and start naming “buckets” of information.  Collect feedback from your database “owners” across the organization.  What is going to work best for the greatest number of people.  It may not be perfect and not everyone will follow it immediately.  Keep in mind, if you develop the standards, you may be the one tasked with enforcing them, at least early on.

Taking these five steps will help you pare down and better manage your database gremlin and focus you more on the insights than the work it takes to get there.  Stay tuned for our next post – What You Should Be Tracking in Your CRM/AMS (But Might Not Be).  Now that you’ve pared down your database, it’s time to start adding and enhancing it to make it even more powerful.

Posted in Blog, Insights, LRC Blog.