I have a guitar that I keep in the closet. She’s about 30 years old and has been through eight moves. She lost a tuning peg at some point and now the strings are corroded and way, way out of tune. But I won’t part with her, since she’s the first guitar I ever bought. I learned basic chords on that guitar and have a feeling she knows Dust in the Wind almost as well as I do. But do I ever play her? No. The body shape isn’t comfortable for me to play and the neck is a bit too wide for my hand. I have other guitars that are in better condition and easier for me to play. But, I move that old guitar each time I pack up. And she could make music again. Dust her off. Replace the stings and tuning peg. Get her in tune. She’s not the best guitar I’ve got. But she’s not bad. And she could probably still play Dust in the Wind, given a chance.
You probably have data that’s in the same shape as my old guitar.
You’ve had it forever and won’t give it up for whatever reason. It might have migrated a few times. It’s dusty and corroded and missing a few parts. And it doesn’t quite fit how you do things these days. Using that data for an analysis or a campaign is like me trying to play that guitar without fixing her up. I could do it, but it would sound just awful!
I attended several sessions at bbcon 2015 about data-driven decision making. Regardless of whether you’re an art museum, a hospital, or some completely different type of organization, data and algorithms have a tendency to outperform human intuition in a wide variety of circumstances. But the one thing every bbcon session had to say is that you can’t make accurate or even reasonable decisions without clean data. Anything you try to do will only be as good as your data is valid. So if your data is faulty, your decisions will be, too.
So what do you do with that old guitar, er, data? It really boils down to four options:
Get rid of it.
If there’s absolutely no reason to keep the data, delete it. Do you have a bunch of records that contain names only with no contact information or any other details? We ran into this at the RISD Museum when we realized we had imported a group of records with names only – sometimes only first names — and there was very little chance we would ever discover anything else about those visitors. If those visitors had other interactions with the Museum, there would be another, more complete record. So, even though it was difficult to do – we’re a museum; we save everything! – we finally agreed that getting rid of those records was the right thing to do.
Keep it in the attic.
If you’re not going to use the data but there’s potentially some value in it, archive it and move it to an offline storage system. This way the data isn’t gone, but it’s not cluttering up your day-to-day life. You can get to it if necessary, although it might take a bit of effort. When the RISD Museum migrated to Altru, we archived our old admissions and event registration data, just in case. We ended up accessing the old data about a year later, when we realized that we needed a few more pieces of information for our end-of-year reporting. But that was it. Two years later, we haven’t touched that data since. For now, it’s out of the way on a rarely used server. But eventually we may want that space for something else. Or the data format may become obsolete and we won’t want to expend the effort to migrate it. At that point, we’ll reevaluate and reconsider Option 1.
Keep it in a closet.
If you think the data is still useful or if you have a plan to one day fix it up (Option 4), keep it in your day-to-day system. The RISD Museum migrated approximately 23,000 records from our previous system, including many constituents we had not interacted with in decades. Given the impenetrability of our old system, we’re not sure we were able to extract everything. That said, the data we were able to migrate still provides a valuable, if admittedly incomplete, historical view. In order to keep that old data from inaccurately influencing current decisions, it’s excluded from queries and analyses. Most of our mailings are targeted to “recent” constituents: those who have visited in the past three years or specifically asked to be contacted. Although we keep the old data close to hand, we also keep it out of the way.
Fix it up.
This is the most time-intensive and expensive option, but the one that will actually let you use the old data with confidence. Figure out what needs to be changed to make the data useful. You can use a tool like Blackbaud’s AddressFinder or request a National Change of Address (NCOA) report to get the most recent contact information. Send an email to all of your constituents to obtain current contact information and to identify bounced emails. Create queries to identify inconsistent data or common misspellings. And keep in mind that these tasks aren’t just for your old data. They’re part of the never-ending housekeeping chores that will keep your incoming data as clean as possible.
You may end up using any combination of these options. Unlike an old guitar, you don’t have to pick just a single course of action. Once you have more confidence in your data, you can have more confidence in the decisions you make based on that data. As it says in the Blackbaud Knowledgebase, “When you have clean and accurate data, you can spend more time on your organization’s purpose and goals.” And that’s music to anyone’s ears!