Skip to main content

How to test your recruiting strategy

· 2 min read
David Len
Producer, software developer

"It's so hard to find talent." -- company

"The job market for SWE is cooked." -- Software engineers 2024

Logically, that can't be possible. But nothing in real life is perfectly logical. As engineers we need to diagnose and fix the problems.

In software, we have things like unit test, integrated test, system test.

As it complexity goes higher, we need live simulations, like hiring a game tester or setup Selenium to simulate a login process.

With the same philosophy, a company could hypothetically perform "recruiting test" that is really easy to do.


Recruiting test (no one does)

Step 1: Create a fictional profile

with the ideal resume for the position you're trying to recruit.

Step 2: Let your own recruiter call back your resume.

If it doesn't get past the phone screen, your recruiters are not doing their job.

Proper scientific method

If you have to do it scientifically, create 10 or 2-10% of applications with those profile. Do not announce to your recruiters you have this much "qualified simulated accounts" so the test set is not included in training.

Check how much they missed as a benchmark, say 7/10 to measure you screening false negative rate.

You will solve the tech job shortage + tech unemployment.


"It's going to waste 20 minutes of my life and phone call!"

You waste months of life on a wrong hire. If your CEO picks up the phone just say congrats you done your work.

"Creating and maintaining fictional profiles is so hard!"

Everyone is using AI to generate their profiles already. They are also including as many frameworks (they are not even good at) as they can in resumes to hack the ATS keyword search.

Instead of investing on data science to figure out a better recruiting AI, use cheap data science to test your recruitment hit rate.

"It makes a lot of sense but we don't want to do it"

That's how we get tweets like this.