Data Integrity at the RISD Museum: The Long and Winding Road

Guest Post from Shannon Knight, RISD Museum

Visitors provide data to museums and other institutions all the time: when they make a donation, when they become a member, when they visit or attend an event. And we, as museum staff, use that information to make decisions that (hopefully) result in better services for those visitors. But that’s only possible if the information we’ve gathered is consistent and accurate.

Back in 2011, the staff at the RISD Museum knew there was a problem with the integrity of our data, not to mention the accessibility. In order to use the latest visitor data in any way, several weeks of work were required since that data was scattered all over the place. Ticket sales and event registration were handled by one system. Online purchases took place via another system. Memberships and donations were handled by yet a third system that wasn’t designed for either memberships or donations and was not directly accessible by Museum staff. Youth program registrations were kept in a variety of Excel spreadsheets. Contact information could be found in Word documents and on pieces of paper tacked to bulletin boards. Any time we wanted to utilize data – whether to make a decision or to create a mailing list – we had to scramble to pull all the information together. And, with the inconsistencies of the disparate systems, there were always questions as to the accuracy of the data.

At this point, it became clear that we needed a new enterprise system that was designed specifically for a cultural institution and that understood our business practices and needs. Enter Blackbaud’s Altru! All data related to our visitors – whether membership, attendance, or donations – would be kept in one, easy-to-access place. It was perfect for what we wanted to do. Out of the box, it was a clean, beautiful system, just waiting to be populated by clean, beautiful data. But, since we already had roughly 23,000 constituents, we decided to migrate the data from all of our old systems into the new one. And that data was not at all clean or beautiful.

The biggest lesson learned during our migration process was a variation on an old adage: garbage out, garbage in, garbage out. When we pulled data from the old system, we decided to take everything, all the way back to 1972. This included not only the major donor who had a lifetime of giving, but also the student who gave $5 during a phone-a-thon in the late 80’s and was never heard from again. We spent roughly six months mapping the data to the new system and fixing as many problems as we could before Blackbaud engineers imported it into Altru. While the migration process itself went extremely smoothly, Bad Data followed us into the new system. Some were things we simply missed, such as phone numbers with incorrect area codes or addresses with duplicate Line 1 and Line 2. The other types of issues we ran into came from not fully understanding how the new system worked. For example, our membership program in Altru only allows one membership per person; however, our migrated data created a separate membership for each level that the individual had ever purchased over the course of their lifetime.

In retrospect, it may have been better to only migrate data that was directly relevant to our organization today. So the major donor mentioned before would have been migrated, but not the student we only had contact with once in the late 80s. This would have minimized the old data that needed to be cleaned up.  Hindsight, of course, is 20/20 and there are several things we’ve done since migrating to identify and correct Bad Data:

• 
  
• Run the Duplicate Constituent report monthly. Records that match on 100% of the criteria are automatically merged every week.
  
• Request a National Change of Address (NCOA) report from our mailing house for every mailing that is sent.
  
• Identify bounced emails from email appeals and other email communications.
  
• Create queries to identify inconsistent constituent information, such as female constituents with “Mr.” as the title, or records with an unknown gender but a common name such as Thomas or Janet.
  

We also created a Data Integrity Team with representatives from all groups in the Museum that either create data or have information needs. By meeting regularly, we can identify what type of questions we want to answer and make sure we have the reports and queries to support those questions. To make sure those reports and queries are accurate, we revise the data gathering processes as needed and document data standards so that everyone responsible for entering data knows how to do so correctly.

With all these efforts, you would think that our data clean-up is finished. And it would be – if new data wasn’t coming into the system every day. We plan to always have new visitors to the Museum and will continue to interact with existing constituents. So data integrity will always be an issue. And data clean-up will never, ever be done.