July 21, 2015
We recently co-hosted a webinar with Kevin Zeigler of TechnoServe about data quality. After the webinar, we realised that there was a lot of learnings to unpack around the subject of data quality. We will be publishing a series of blog posts on this topic over the coming weeks.
One of the major issues of running a data-driven organisation is that your data can be problematic. If it is riddled with errors, you not only lose a lot of accuracy in your results and findings, you also lose credibility within your organisation. With data becoming more important for many organisations and increasingly so in the development sector — the UNDP recently blogged about data being the key to successfully implementing the Sustainable Development Goals — we thought it’d be useful to take a closer look at data quality. After all, if people are going to successfully use data to take on some of the biggest challenges and problems in the world, they’re going to need to keep their data clean.
By using TaroWorks, you also begin to use Salesforce.com. All the data your field agents collect is stored and sorted on your Salesforce instance ready to be sorted and analysed using Reports and Dashboards. (For some great tips on using Dashboards, check out this presentation by Illuméxico). Usually, Reports are used to sort data to find positives. How many participants did we have? Did we get more than 50% women? How many families have we reached? The results of such queries are normally things you want to share and you can set them to run automatically. What people don’t usually look at with Reports are mistakes. Data errors often follow patterns. This is very useful for two reasons: 1) You can figure out how to fix the underlying problem. (This will be the subject of an upcoming blog post on data quality). 2) You can find the same types of error extremely quickly using Reports. Once you identify what looks like problematic data you can very easily clean it up. Anybody with a Salesforce login can be sent the report so you can send the reports to the project manager responsible for that specific program or product to double check the results. If the errors are confirmed as duplicates, you can delete them. Otherwise, you can correct the details.
You know when your field staff are collecting data. If you create a report that finds all data points with timestamps outside of the possible collection dates, you can isolate a lot of bad data very quickly.
There are a number of applications for this technique. TechnoServe looks at timestamps and National ID numbers of participants in training programs, but you can adapt this method to suit your specific use case. There are two basic things you will need to look for:
It’s fast! The major benefit of this technique is that it takes a lot of the time pressure off your M&E team. Finding data errors in a large dataset can be extremely time consuming. These reports, once set up, are automated — you can schedule them to run once a week or once a month depending on your use case. Even better, once the results come in the problematic data can be highlighted to other team members who are in the field and, therefore, better able to correct errors quickly. Negative reports are an excellent weapon for combating data quality issues. Don’t spend hours combing through data to find errors, just set up some reports!
Sign up to receive emails with TaroWorks news, industry trends and best practices.