Matching is a key component within Eightfold and is used in almost every part of the product platform. This article gives an overview on matching: what is Calibration, types of Calibration, Calibration workflows, and an explanation of match score analysis for hired candidates for different types of calibration.

Introduction

Calibration is the process of defining criteria for hiring, allowing Eightfold’s AI to surface leads and match applicants to open positions. It produces a match score for each candidate and narrows down the pipeline of leads.

Calibration is relevant within the context of a specific position. Instead of listing multiple requirements for a role, Eightfold Calibration helps define the requirements in terms of the following key attributes:

• Titles
• Skills
• Ideal Candidates
• Work Experience Range
• Location

Calibration status types

There are three types of calibration statuses:

• Complete: Calibration is considered Complete when a user clicks Confirm Calibration after providing all the relevant information such as titles, skills, and more. Calibration can also be confirmed without providing all the key components.
• Needs Confirmation: This status applies to positions where the user mostly completes the calibration and does not click Confirm Calibration. The reason is to let these calibrations learn from candidates in similar positions. We update the calibration based on actual user requirements through hired candidates, not by defining them in the calibration.
• Incomplete: This is the third type of calibration status which is associated with incomplete calibrations. This means that the calibration is missing one or more of the following:
  • Three or more titles
  • Three or more skills
  • Three or more ideal candidates
  • Minimum or maximum work experience
  • Location

Calibration methods

Manual Calibration

The calibration is created by a user (recruiters, hiring managers, and others) where the user defines all or some of the key calibration components and clicks Confirm Calibration.

Auto Calibration

The calibration is created by parsing available position information and continuously updating it using Eightfold’s ML models. As the name suggests, users do not need to intervene in this calibration method.

Calibration completeness

Calibration completeness is a measure of sufficiency of the details added to a position that fetches you the best available leads for the position. Higher the completeness score, the better the leads pipeline will be.

The conditions for the calibration completeness of a role are listed below:

• at least three ideal candidates
• at least three preferred skills
• at least three alternate job titles
• at least one of {maximum_years of experience, minimum_years of experience} is present