July 8, 2015
One of the big problems about running data-driven operations is the relative quality of the data collected. People make mistakes when they input data. Sometimes missing information can prove to be very problematic for your database. Maybe the surveys you’re collecting data with are simply too confusing. Every M&E team knows what a hassle spotting problems and cleaning data can be. That’s why we got in touch with Kevin Zeigler, Special Projects Manager at TechnoServe, to give us all some tips on how to solve and, eventually, avoid a lot of these issues. We’ve done some quick highlights for some of the points he made below. Stay tuned for more in depth pieces in the coming weeks.
Kevin introduces us to the idea of ‘negative reports’. Rather than use Salesforce.com Reports to find useful aspects of your data, you can run reports that sweep for obvious mistakes. For instance, you could run a report that looks for data with timestamps from years when you couldn’t possibly have collected it (e.g. 2057 or 1893). This is a quick way of finding a lot of poor data very quickly. This way, you can keep checking problematic areas for a whole array of different surveys and jobs rather than laboriously seeking out individual errors. Even better, you can schedule these reports to run every week to easily remind you when you need to clean up data.
You can set these up in Google or Excel, depending on what best suits your organisation. You should fit these with your Salesforce.com reports by using the Data Import Wizard so everything matches up nicely in a .CSV file. This process also cleans up thousands of records very quickly and will save you a whole lot of time. We will be doing an in-depth article on this process very soon.
It’s one thing to figure out how to fix problems once they’ve occurred, but, by tracking those errors, you can turn this into a great learning opportunity. Maybe particular questions or surveys are problematic and need to be re-designed. Maybe some staff members are less tech savvy and they need some extra training support. Proper training is one of the most vital interventions for improving data quality.
For tips on training you can read TaroWorks’ top ten tips or check out some great advice from some of our users at Trees For The Future and Tetra Tech
This is a great reminder. Cleaning data is only one step of the ‘Data Quality Lifeycle’:
It’s always better to look to solve problems at the field level first. This takes a lot less time than actually cleaning up the data on the back-end. Helping your field staff to submit fewer data errors is usually the best solution. If that isn’t possible, you’ll need to escalate everything up the chain (as shown in the diagram). And there’s a whole heap more useful information to be found in the video. Please let us know if you have any great tips or additional questions so we can post more follow-up guides and tips for you.
Sign up to receive emails with TaroWorks news, industry trends and best practices.