Job description best practices based on hiring data
Feb 09, 2017
Before landing at Textio I spent several years studying physics. There’s a curious education path in physics where you essentially take the same three classes over and over again adding a new layer of complexity each time. While a second-year physics major can tell you whether or not an outfielder will catch a deep hit to center field a moment after it leaves the bat, that same student probably can’t tell you why someone can drop a basketball off a dam and have it “fly away” from them to skip along a pond. The rules that applied on the baseball field were different, or simply didn’t apply at the edge of that dam. This was crystalized perfectly for me when, in my junior year thermodynamics class, my professor succinctly stated “remember everyone, energy is always conserved except when it isn’t.”
“Remember everyone, energy is always conserved except when it isn’t.”
This quote stuck with me as I transitioned into tech. When building machine learning models, experimentation systems, or analyses, I constantly found a familiar theme popping up “things work well, except when they don’t.”
This is especially true when it comes to job listings. When we built “Opportunities” last year, we started using Textio’s predictive engine to find the hidden meaning behind your words and suggest alternative ways to communicate that are proven to attract stronger candidates to your job. One thing the data has taught me: Opportunities come and go as the context around phrases changes. Basically, some phrases that are Opportunities work well… well, until they don’t.
When we launched Opportunities, we had 15 million job posts with hiring outcomes, now we have over 70 million job posts with hiring outcomes, and our database is growing every week. With all this new data, we can find new and deeper patterns where words or phrases are proven to find larger quantities of more diverse candidates faster.
Context matters
Today, we’re updating the predictive engine to help find more patterns that are context specific for job types around the globe. Textio now includes context-specific phrases for the administrative roles, customer service, project management, research, design, personal services, and product management industries. This builds on the context specific phrases we launched last year for the engineering, IT, retail, healthcare, human resources, and finance industries.
Textio now has industry-specific phrases for:
- Administration
- Customer service
- Project management
- Research
- Design
- Personal services
- Product management
An interesting way this plays out: if a job listing for IT professionals mentions “establish” they will be turned off, but if you switch that to “develop”, they are more likely to apply. However, the reverse is true in personal services and product management, they would rather “establish” something than “develop” it.
Another interesting example we came across — our data tells us that, in general, current job listings can be improved by swapping out “great attitude” for “positive outlook”. However, for Design jobs we see that “great attitude” benefits from a change to “great work ethic”. In Administrative job listings, “attitude” can be improved by being similarly concrete with “work ethic.” Finally, in Finance, “attitude” surprisingly can be improved by shifting to “personality”.
These are just a few of the millions of patterns that the predictive engine has been updated to help find. With these new context-specific outcomes in store, Textio will become even better at helping you hire people faster.
