Data cleaning is a skill that recruiters and hiring managers are learning these days out of necessity, but it’s far from universal. Thankfully, the recruiting analytics experts at Datapeople have put together a data cleaning checklist for recruiting operations teams.
“Data management in applicant tracking systems, or ATSs, has a big impact on what hiring data tells recruiting operations teams,” says Datapeople Content Marketing Manager Charlie Smith. “It’s important that everyone using an ATS understands data management best practices and sticks to them rigorously.”
Datapeople’s data cleaning checklist includes nine steps. The first four have to do with how companies approach their hiring data:
First, teach data management basics, the company says. Whoever is using an ATS needs some training on the basics. They should understand what a literal creature an ATS is (e.g., it doesn’t know that ‘LinkedIn’ and ‘LI’ are the same thing). Second, attach candidates to jobs. Including complete information for every candidate provides a full pipeline picture of a job. Third, incentivize data cleaning. Hiring teams need data to understand why a recruiting effort didn’t yield a hire. Four, use thoughtful reasoning. Hiring teams have to look at the right metric, according to Datapeople. For example, there’s no point in using percentage of qualified candidates to unqualified candidates because number of qualified candidates is the metric that matters.
The next five steps in Datapeople’s checklist deal with the data directly:
First, collect end-to-data on every job. End-to-end data is crucial to data cleaning because it’s the only way to get a full view of every hiring effort, according to Datapeople. Second, use enough data to be significant. Rather than comparing individual jobs or single-digit numbers against each other, use comparison groups of at least 10 or more jobs. Third, separate data into thoughtful buckets. By segmenting data into thoughtful buckets such as location, seniority level, and job type, hiring teams can see what’s happening on the ground in each situation. Four, remove outlying pieces of data. Some jobs behave differently than others because of their unique nature. For example, evergreen jobs, internal hires, internships, and new-grad hires. Five, use median calculations over mean calculations. Applicant pool data can contain lots of outliers, Datapeople says. Given that a single outlier can muddy the picture, it’s important to use median calculations instead of mean calculations.
According to Datapeople, data cleaning isn’t particularly difficult, although it does take some basic data management skills and constant vigilance. But it’s important because unclean data can show hiring teams an inaccurate picture of their candidate pipeline.
90 Broad St Suite 803
Disclaimer: The views, suggestions, and opinions expressed here are the sole responsibility of the experts. No Fortune Bulletin journalist was involved in the writing and production of this article.