Every company has and uses a predictive hiring system—whether they know it or not. Let's elaborate on a concrete definition of what a predictive hiring system is...
Definition of a predictive hiring system
A system or process that uses characteristics of a candidate to guess or predict the desired future outcomes of the candidate in a given role at a company."
Sorry, that was a lot. Let's break this definition down.
A system or process refers to repeatability. How do we repeatedly hire people and use their characteristics to predict the desired future outcomes?
Characteristics of a candidate are anything that we can know about a candidate before we hire them. These characteristics are the building blocks for the predictions we make about how well a candidate will do.
Some common characteristics include:
- Total years of work experience
- Certifications
- College GPA
- College major
- College alma mater
- Interpersonal skills
- Communication ability
Some less popular characteristics that are starting to gain traction include:
- Personality types
- Skills tests
- How well someone handles stress
- Work samples
To Guess or Predict, these characteristics are gathered by various means and used by the hiring decision makers to predict who will be the best candidate for the job. Some characteristics are given more weight then others, and some characteristics are ignored altogether.
The desired future outcomes could be anything that the employer is looking to get from the employee. Since each role is different, and the world of work is changing, desired outcomes are also flexible.
Some typical outcomes include:
- Job performance: How well will this person do the job?
- Tenure: How long will this person stay at the job?
- Satisfaction: How much will this person like the job?
- Culture fit/add: How well will this person contribute to the company culture?
In a given role or company: Since each company and role are different, the predictions we make are always custom to that specific position regardless of other generalized inputs.
Starting the (not so predictive) predictive hiring process
Now let's look at how this all works together in a common hiring situation.
Beth is a recruiter at Acme, INC. She needs to hire an entry-level sales rep in the next 30 days for one of the teams she supports. From here, she has a standard process to fill the position that might look something like this.
- Gather requirements from the hiring manager
- Source candidates
- Initial candidate screening
- 1st On-site interviews
- 2nd On-site interviews
- Offer
This is the system or process that Beth uses to find and hire the "best" candidate for the job.
To start, she has to understand what the "best" candidate looks like, so she talks with Jane (the hiring manager) and asks: "So what are you looking for in this sales rep position?"
Jane replies with a list of characteristics that Jane believes will lead to the "best" candidate. The list looks something like this:
"I need someone with ...
- A bachelors degree from a good school
- Smart
- Can handle rejection
- Great communicator
- Hard worker
- 1-2 years of sales experience in B2B"
Beth writes these characteristics down and starts sourcing candidates based on the characteristics she received from Jane. As candidates start applying, Beth screens their resumes against the list of characteristics and accepts candidates to interview that "look good on paper."
The candidates are brought in and interviewed by Jane and others who ask additional questions to assess characteristics of candidates. They might say "Tell me about a time you were in a stressful work situation and how did you handle it?" to gauge how well this person can handle stress.
Once candidates are all interviewed, all the data is gathered in one place, often a conference room table, where the hiring manager and team review resumes, interview notes, and interviewer feedback to determine who is the "best fit" for the job.
The Prediction Happens
With all the data gathered on the final candidates and the stakeholders in the room, everyone has a chance to voice his or her prediction about the candidate and what data was used to make that prediction.
"I think Sally is the right person for this job. She has 2 years of sales experience, has been in some stressful situation at work, and I think she would work well on the team."
"My concern with Sally is that while she has sales experience it's B2C sales and we need B2b. I think Greg is our guy. He did great in the interview, has 1 year of B2B experience, and can start immediately."
This goes on until Jane, the hiring manager, takes in all the data from the team and then ultimately makes her own prediction about who will be the best fit for the role.
Problems
There are several common problems with the manual approach laid out in this example. Here are a few...
- No/unclearly defined outcomes: It's impossible to determine what will predict the outcomes of a successful employee if success is not well defined. If you don't know what you want, you certainly can't get it.
- It assumes the hiring manager knows the characteristics that cause the desired outcomes: Even assuming you have clearly defined the desired outcomes, the hiring manager is merely starting with a hypothesis of the attributes that lead to them.
- Decisions are typically made relative to cohort reference points: Hiring decisions are often framed in Sally vs. Greg terms. Rarely do companies check previous candidates cohort to ask, "How did people hired last with a similar data profile as Greg perform?"
- The data stays on the conference room table: Data from the hiring process is rarely kept and used later to evaluate whether the hypothesized attributes actually correlate with desired outcomes.
A better way
All companies predictions based on data collected during the hiring process. The more companies leverage data and reduce the bias in what data to gather and use, the more accurate they will be. Here are a few quick tips on how to improve your predictive hiring system...- Articulate the outcomes expected from a "successful hire" for the given role.
- Document a clear hypothesis on what candidate attributes will cause expected outcomes
- Follow a process for how those candidate attributes will be measured
- Collect and record data gathered during the hiring process
- Work with a data scientist/analyst on your team to identify trends in post-hire performance and candidate data gathered during the hiring process (or get Journeyfront to help 😃)