Improving Data Quality

It's well worth it to invest the time and effort required to maintain high-quality data at your district. Not only will integrations go more smoothly when your data set is complete and organized, but you'll also reap the long term benefits of sound, data-driven decision making.

Here are some tips from our team about how to get your data into tip-top shape.

Know Where Your Core Data Lives

The first step to improving data quality is to know where your data lives. This means understanding where your data is stored, how it's collected, and who has access to it. This is especially important given that most districts are working with dozens or hundreds of different vendors.

Establish a Single Source of Truth

It's very common that we encounter districts whose core data (e.g. a list of staff members) is split across many systems. This can lead to a lot of confusion, errors, and general headaches when you're trying to establish integrations. It's worth the effort to establish a single source of truth for your core data. This could be your SIS, HRIS system, or another system that you trust to be accurate and up-to-date.

Use Globally Unique Identifiers When Possible

Identify your students, staff, classes, courses, and other entities with globally unique identifiers. This will make it much easier to integrate your data with other systems and ensure that your data is consistent across all systems. When an entity is identified by a globally unique identifier, you can be sure that you're always referring to the same entity, no matter where you are in your data systems.

Ensure The Vendors Rely on Your Unique Identifiers

When vendors report data back to you, make sure they're using your globally unique identifiers. This will make it much easier to integrate their data with your core data and ensure that your data is consistent across all systems.

Things That Don't Really Work

• Asking vendors for a data export after you stop using their product. This is a common request, but it's not a good way to maintain data quality. The data you get back is often incomplete, and it's not in a format that's easy to use. Plus, the only company who (probably) knows how to interpret the data is the one who collected it in the first place (that you're no longer a customer of). Instead, make sure you have a plan for how you'll get your data out of a vendor's system before you start using it and select vendors who are willing to integrate with your data systems.
• Custom SIS tables or datastores. These are often created with the best of intentions, but they can quickly become outdated and difficult to maintain. SIS and LMS systems are pretty powerful these days, and if you are looking to store data that doesn't fit into the existing schema, it's worth considering whether you're using the right system for the job (or if it's the right job at all).
• Manually modifying downstream systems to save time. If you are working with a vendor who is integrated with your SIS, it can be tempting to make changes directly in the vendor's system to save time (e.g. because you don't want to wait for the changes to propagate). This is typically a bad idea. It's much better to make changes in your SIS and let the integration handle the rest. This will ensure that your data is consistent across all systems and that you don't accidentally break the integration.
• Prefixing globally unique identifiers (sometimes). This one is a tough one. Globally unique identifiers can be annoying to work with if they're not prefixed with something that makes them human-readable. However, if you prefix them, you will lock yourself into a particular format and may create a headache for yourself in the future. For example, if you generate teacher IDs in the format T-1234 and staff IDs in the format S-1234, you may introduce an issue when a teacher becomes a staff member (or vice versa).