How to use predictive hiring analytics

Imagine being able to predict if a candidate you’re recruiting would excel in a role before you extend an offer. 

You’d know ahead of time whether they’d take to the job and if they’re likely to stick around. If you could use data to know who would be a high performer and a culture fit, think of how much time (and money) you’d save on recruiting. And when you streamline the recruitment process, not only do you save, but you also offer a better candidate experience that makes you the easy choice for a competitive applicant.

This scenario isn’t a pipe dream. Predictive hiring analytics makes this reality accessible to hiring managers and recruiters who put it to use, but it also comes with certain risks, like a propensity for biased decision-making. Here, we’ll provide practical guidance on using hiring analytics to improve recruiting decisions while managing risk.

What are predictive hiring analytics?

Predictive hiring analytics use current and historical data to forecast a candidate’s chances of thriving in a particular role and company culture. Whereas resume screening looks at what’s on paper, and intuition-based interviews let hiring managers “feel” for fit, predictive hiring analytics gathers inputs from recruitment data and feeds it into a machine learning model that predicts applicant success, performance, retention, and hiring needs. 

Instead of looking at one piece of the puzzle, predictive hiring analytics factors in all relevant data to see the bigger picture, allowing hirers to make smarter hiring decisions that maximize retention.

How predictive hiring analytics work

Predictive hiring analytics start with gathering current and historical data, like interview ratings, performance, tenure, etc. A predictive model processes that data and uses it to deduce who would be a great fit based on who at the company already excels. While this data is useful, it’s essential for a human hiring manager or recruiter to vet the output before making the final decision.

Model inputs and analysis

Predictive hiring analytics runs on data. Some of those inputs include:

• Assessments: A candidate’s score on a trial exercise. This is a common and insightful way to see how an applicant would approach their work.
• Interview ratings: Interviewers’ notes and overall ratings of a candidate’s interview. This input factors into the candidate’s off-page impression on potential future colleagues.
• Performance: Current employees’ accomplishments, including target attainment and quarterly review scores. These are used as benchmarks to measure future performance.
• Tenure: How many months or years an employee stays at the company, whether in their original role or one they were promoted to.
• Retention: How well the company as a whole retains talent. The retention rate is the standard that a candidate’s potential is measured against. 
• Standardized ATS records: All the relevant information the predictive model uses to build a candidate profile. This includes the applicant’s resume, interview notes, employment history, etc.

Scoring and continuous refinement

Using the data above, the model calculates a candidate’s potential for success based on criteria you trained the model to prioritize. Over time, the model becomes better at identifying right-fit candidates because you evaluate the outputs and refine the criteria based on what you see working. 

Benefits of predictive hiring

Predictive hiring analytics leads to more effective hiring processes, like faster time-to-fill, better quality applicants, fairer hiring, and lower hiring costs.

Hiring outcomes and efficiency

• Improved quality of hire: Hiring managers can identify and onboard better candidates by pattern-matching their qualities against the traits of current high-performers.
• Faster time to hire: Recruiters can rank and prioritize candidates based on how they score, then send out offers more quickly.
• Lower cost per hire: By spending less time hiring, recruiters also reduce hiring costs like job ads, signing bonuses, and recruiter comp.
• Stronger retention: Hiring the right person for a role improves the chances of them sticking around long-term, as they are a good fit for the work and culture.

Decision quality and candidate impact

• More consistent evidence-based decisions: Since data is at the core of every candidate selection, decisions borne from a fair predictive model are more likely to follow a consistent pattern.
• Better candidate experience: Data helps recruiting teams make faster decisions, improving the candidate experience, since applicants wait less time for a response.
• Improved role fit: Predictive hiring makes it more likely that the person chosen for the role will be a fit for both the job and the company culture, as recruiters aren’t hiring on gut but signals tied to on-the-job performance.

Risks and challenges to consider

Predictive hiring analytics empowers data-driven hiring practices, but it’s only effective if the underlying data is accurate, complete, and consistently collected and interpreted. If it isn’t, your team runs the risk of introducing bias in your hiring processes, which can lead to higher churn and more workforce disruption.

Bias, fairness, and compliance

Say your model was trained to identify ideal candidates based on the demographic qualities of your current workforce. If your talent pool is skewed towards a certain demographic, then the model may ignore perfectly suitable applicants because they don’t match that pattern. 

Discriminatory scoring, which includes ranking candidates who attended certain schools or graduated in a time period, will inevitably deliver a narrower pool of candidates and undo the core benefits of predictive hiring analytics.

Limits of automation

There are nuances that even well-programmed hiring models can miss or disregard. A human is always needed to course-correct and refine the outputs based on their own insights and intelligence. Especially in the beginning and at a regular pace over time, you should continuously audit and validate the model’s outputs to ensure they’re making accurate suggestions. Constant review helps you keep your hiring models effective even as your criteria change.

How to implement predictive analytics in recruitment

Begin using predictive analytics by first identifying goals and cleaning up data. Then, put the model to work in a pilot project and continuously refine its suggestions. Here’s a basic overview of how to get started.

Foundation and technology setup

1. Define recruitment goals: First, know your goals. For example, is it to lower turnover? Speed up time to fill? Reduce hiring costs? You’ll get the most out of data analytics if you’re clear about what you want to get from it. 

2. Clean historical ATS and performance data: Clean up your data so that the predictive model has the cleanest, most up-to-date, and most accurate data available. Dumping bad data into the model will only result in unreliable outputs. Prepare data by standardizing field formats, removing duplicate and outdated records, and deleting fakes/test candidate profiles.

3. Choose integrated predictive tools: Integrate tools that expand your current tech stack’s capabilities (for example, adding sourcing technology to your CRM) to enrich your recruitment workflows so you can work smarter without having to keep toggling between systems.

Model rollout and team adoption

Start small with the rollout so you can work out the kinks before it goes live to the rest of the organization.

• Build role-specific models: An ideal quality in one role might not translate the same for another. Build predictive models that meet the needs of specific roles.
• Validate models with a small pilot: Start by building models for a specific role or department where you have clean data and a clear success measure. Run this model alongside your regular hiring workflow and compare its predictions against results. If the hires that perform well are the same ones the data told you would, you’ve validated the model and can expand its use.
• Train recruiters to use recommendations as decision support: Enable recruiters to assess recommendations and use their judgement to make the final decision based on their experience and set criteria. Train your team on reading the reasoning behind recommendations and the situations in which it’s best to override the model’s suggestions.

How to measure predictive hiring success

Define success metrics and measure hiring outcomes using recruitment KPIs. Here are a few examples.

Core recruiting and business metrics

First, calculate whether your recruitment process is delivering the business outcomes you anticipate: 

• Quality of hire: Measure the retention rate of hires made within the same time period six to 12 months after the offer has been extended.
• Retention: Calculate how long new hires stay with the organization. If they churn within the first 60 to 90 days, it’s a sign to re-evaluate your recruitment process (alongside other factors, like a challenging company culture or unreasonable demands of a role that could be pushing employees away).
• Time to hire: Count the days or weeks from when the candidate first applied to when you extended an offer. A quick time to hire is a strong indicator of an efficient hiring process. Remember, however, an efficient hire isn’t always a quality one, so it’s important to assess fit and retention alongside time to hire.
• Cost per hire: Divide the amount you spent on recruitment by the number of hires you made within a specific time frame. Compare budgeted spend to actual to understand if you’re budgeting the right amount (or if you’re overspending). Low cost per hire is generally a sign of an efficient hiring process, so long as hires stay with the company. Rehiring to fill roles for people who underperform or leave is more expensive than making the right hiring decisions in the first place—even if the process takes a bit longer.
• Offer acceptance rate: Divide the number of offers accepted by the number of offers extended. A low rate points to a problem in your hiring process or job offers, whether that’s uncompetitive pay or a slow candidate experience driving lost interest. A hire offer acceptance rate, on the other hand, points to a positive candidate experience and attractive offers.

Validation, fairness, and model performance

Now, measure how efficient and effective your predictive model is:

• Predictive accuracy: How often is your model right about a candidate? If it’s wrong more often than right, it needs recalibration.
• Adverse impact rates: Are candidates from minority and protected groups being selected at disproportionately lower rates than other groups? If the selection rate is under 80%, federal guidelines consider it adverse impact.
• Calibration: Do the model’s predictions come true? When a model puts forth a probability, measure how closely it aligns with actual outcomes.
• Periodic outcome reviews: How closely are you monitoring the outcomes of your hiring models? Measure KPIs like predictive accuracy and adverse impact rates consistently over time to measure outcome accuracy and improve the model.

Start with better data at the top of the funnel

Predictive models do their best work when there are quality candidates entering the funnel. The better you are at sourcing talent, the better inputs your model has to work with. 

Juicebox is the sourcing layer that delivers strong-fit candidates right to your hiring pipeline. And with independent AI governance (SOC2 Type II, ISO 42001, and public Warden AI bias audit), you can feel confident that the data supports model fairness. 

Start with a free Juicebox search today.

FAQs

What is predictive hiring?

Predictive hiring is what happens when recruiters and hiring managers use past and present recruitment data to make hiring decisions. It’s the process of choosing candidates based on the recommendations of predictive hiring models.

What is predictive hiring analytics?

Predictive hiring analytics uses machine learning and data models based on historical and current recruitment data like retention, assessments, and ATS scoring to forecast and evaluate hiring outcomes.

What is predictive analytics in recruitment?

Predictive analytics in recruitment uses current and historical hiring data, like ATS records, interview ratings, assessment scores, performance, and others, to determine whether an applicant will be successful in a particular role.

Does predictive hiring reduce bias?

Predictive hiring can reduce bias if the underlying data is accurate, clean, and consistent. If the data itself is biased, it is more likely to produce biased hiring outcomes.