February 10, 2014

Career Reflections: My Biggest Data Fail

Peter Campbell of Legal Services Corporation shares his biggest data fail, and what he’d do differently now.

  • If I’m the IT Director, I can’t be the developer.
  • Get organizational commitment.
  • Don’t compromise on a vision for expediency’s sake.

This case study was originally published along with a dozen others in our free e-book, Collected Voices: Data-Informed Nonprofits. You can download the e-book here.

Note: names and dates have been omitted to protect the innocent.

Years ago, I was hired at an organization that had a major database that everyone hated. My research revealed a case study in itself: how not to roll out a data management system. Long story short, they had bought a system designed to support a different business model and then paid integrators to customize it beyond recognition. The lofty goal was to have a system that would replace people talking to each other. And the project was championed by a department that would not have to do the data entry; the department identified to do all of the work clearly didn’t desire the system.

The system suffered from a number of problems. It was designed to be the kitchen sink, with case info, board updates, contact management, calendaring, web content management, and other functions. The backend was terrible: a SQL database with tables named after the tabs in the user interface. The application itself had miserable search functionality, no dupe checking, and little in the way of data quality control. Finally, there were no organizational standards for data entry. Some people regularly updated information; others only went near it when nagged before reporting deadlines. One person’s idea of an update was three to five paragraphs; another’s two words.

I set out to replace it with something better. I believed (and will always believe) that we needed to build a custom application, not buy a commercial one and tweak it. What we did was not the same thing that the commercial systems were designed to track. But I did think wed do better building it with consultants on a high-level platform than doing it by ourselves from scratch, so I proposed that we build a solution on Salesforce. The system had over 150 users, so this would be relatively expensive.

Timing is everything: I made my pitch the same week that financial news indicated that we were diving into a recession. Budgets were cut. Spending was frozen. And I was asked if I could build the system in Access, instead? And this is when I explained to my boss that we should table the project until we had the budget to support it.

Or so I wish. Instead, I dusted off my amateur programming skills and set out to build the system from scratch. I worked with a committee of people who knew the business needs, and I developed about 90% of a system that wasn’t attractive but did what needed to be done reasonably well. The goals for the system were dramatically scaled back to simply what was required.

Then I requested time with the department managers to discuss data stewardship. I explained to the critical VP that my system, like the last one, would only be as good as the data put into it, so we needed to agree on the requirements for an update and the timeliness of the data entry. We needed buy-in that the system was needed and that it would be properly maintained. Sadly, the VP didn’t believe that this was necessary and refused to set aside time in any meeting to address it. Their take was that the new system would be better than the old one, so we should just start using it.

This was where I had failed. My next decision was probably a good one: I abandoned the project. While my system would have been easier to manage (due to the scaled back functionality, a simple, logical database structure and a UI that included auto-complete and dupe-checking), it was going to fail, too, because, as every techie knows, garbage in equals garbage out. I wanted my system to be a success. We went on with the flawed original system, and eventually started talking about a new replacement project, and that might have happened, but I left the company.

Lessons learned:

  1. If I’m the IT Director, I can’t be the developer. There was a lot of fallout from my neglected duties.
  2. Get the organizational commitment to the project and data quality standards confirmed before you start development.
  3. Don’t compromise on a vision for expediency’s sake. There are plenty of times when it’s okay to put in a quick fix for a problem, but major system development should be done right. Timing is everything, and it wasn’t time to put in a data management system at this company.