How can we help you?

Getting Started with Greenhouse Predicts (Beta)

 

One of the biggest challenges for recruiting teams is accurately predicting when a role will be filled. Using machine learning, Greenhouse Predicts helps forecast candidate offer acceptance and new hire start dates, allowing teams to more easily see around corners, make informed decisions, and effectively communicate timelines. In this article, we'll cover: 

 

How Greenhouse Predicts works

Navigate to an open job and go to the Job Dashboard. Job Admins on the job will see Greenhouse Predicts on the right-hand side of the page.

Screen_Shot_2018-04-17_at_3.52.59_PM.png

Based on the current job’s pipeline, we’ll make an Offer Acceptance Date Prediction and show you the week when we think it’s likely someone will accept an offer.

Screen_Shot_2018-04-17_at_3.54.27_PM.png

We also show a prediction for when we think a new hire will start - this range begins two weeks after the Predicted Offer Acceptance Date and will always be a month long to account for situations out of your control like notice periods, vacations, or relocation.

Screen_Shot_2018-04-17_at_3.54.39_PM.png

Job Admins with the permission ‘Can edit job info’ can click on Set Target Start Date to add or change dates for each opening. 

Screen_Shot_2018-04-17_at_3.54.33_PM.png

We’ll make predictions for one opening at a time. Once an opening is filled, we’ll start predicting for the next one!

Screen_Shot_2018-04-17_at_3.50.09_PM.png

Once you’ve set a Target Start Date, we’ll let you know whether you’re on track to make a hire by your goal. On-track means that the predicted start date falls before or the week of your target start date.

Screen_Shot_2018-04-17_at_3.50.20_PM.png

Heads-up means that your predicted start date falls after the week of your target start date. You might not be on track to make a hire by your goal and should consider adding more quality candidates to your pipeline.

Screen_Shot_2018-04-17_at_3.50.31_PM.png

Head to the Job Status report to see all of your company’s predictions in one place, compare them to actual hire data and check whether you’re on-track to your targets.

Screen_Shot_2018-04-17_at_3.50.42_PM.png

 

How predictions are made

By observing historical data from our customers - millions of applications across hundreds of thousands of jobs - we were able to build a statistical model of job pipelines and candidates moving through them.

We throw a number of factors into an equation to make a prediction. Some of them include:

  • The source of candidates in the job’s pipeline
  • The number of candidates in each milestone or stage
  • Days the job has been open
  • Days since the last hire on the job
  • Candidate offer status
  • Your company’s hiring speed
  • And many other factors!

All of this turns into the predicted offer acceptance date you see on the job dashboard!

Because Greenhouse Predicts was built using machine learning, predictions will continuously improve and become more fine-tuned to your company over time as we get to know your hiring habits.

 

How predictions change over time

Your team can use Greenhouse Predicts to inform and impact your behavior and how your time is spent on different jobs. Some factors that can make your predicted offer acceptance date sooner include: 

  • Building your pipeline through sourcing, job posts, or other efforts;
  • Making sure you have high quality candidates in your pipeline. For a lot of companies, this includes referrals or candidates found through direct sourcing;
  • Moving candidates through the pipeline quickly and reducing your time-to-hire, which includes submitting scorecards sooner and making faster decisions.

Each of these factors is given a certain value and weight in our equation. When any of them change, it’s possible that the prediction might, too.

 

How predictions can help your team

Recruiters:

Recruiters often don’t have insight into the health of their jobs or whether they’re on track to hire someone, making it challenging to allocate budget and time wisely.

Greenhouse Predicts can be used to better balance and manage resources, helping recruiters save effort on roles that are on-track and identify roles that need the most attention. Recruiters can also make more informed and proactive decisions around timing a new search, planning for a departure, or ensuring appropriate coverage on teams.

Hiring Managers:  

Hiring Managers are usually eager to have new hires in seats as soon as possible and projecting start dates is often based on previous time-to-fill data or anecdotal experience, which isn’t always accurate.

Greenhouse Predicts can help manage expectations and guide proactive, realistic conversations about possible start dates while also giving Hiring Managers visibility into the health of their jobs. It can also be used as a tool to show Hiring Managers how changing a job’s requirements or how their actions (like quickly submitting scorecards or making referrals) might impact a start date.

Finance:

It can be challenging for finance teams to accurately forecast salary spend, and if hires start earlier or later than expected, it can have a huge impact on the budget.

Greenhouse Predicts reduces budget uncertainty and evolves the relationship between finance and recruiting by giving more visibility into likely outcomes and guides data-driven conversations that may have been difficult to have previously.  

 

How to think about predictions

The recruiting process can be noisy, and it’s hard to see the big picture -- whether your jobs are on track or not. Greenhouse Predicts helps by serving as a guide and signal.

Since actual start dates may vary widely if your top candidate turns down an offer or accepts but can’t start for 3 months, Greenhouse Predicts may not always be perfectly precise for every role, but should still be a good indicator of “how you’re doing” overall.

And by providing more relevant data, Greenhouse Predicts helps your team make informed decisions, be more proactive, and set better expectations with internal stakeholders.

For more information: