Classic Matching

This article covers Bullhorn Automation's Classic Matching, the default matching algorithm used to power the Match Candidates Step.

Classic Matching (also referred to as "AI Auto MatchClosed") leverages AI to match up jobs and candidates. It evaluates the similarity between job postings and candidates by analyzing factors such as Title, Location, Category, Skills, Specialties, and Industries.

Bullhorn Automation's Machine Learning model is able to determine that a Job posting “Java Developer” is semantically closer to a Candidate with a position title of “Software Engineer”, than to a Candidate with a position title of “Real Estate Developer.”

Classic Matching is available for Bullhorn Automation Enterprise edition.

How Classic Matching Works

Matching Candidates to Jobs

Classic Matching identifies the most suitable candidates for open job positions by following a structured process:

  1. Match on Zip: Candidates are initially matched based on their zip code to ensure proximity to the job location.

    • If Filter on Distance is unchecked in the Match Candidates settings, this step will be skipped.

  2. Match on State + City: If a zip code match is not found, the system expands the search to include matches based on the candidate’s state and city.

    • If the Candidate does not have a Zip Code or a City and State listed they are filtered out.

    • If Filter on Distance is unchecked in the Match Candidates settings, this step will be skipped.

  3. Filter out already matched Candidates: Candidates who have already been matched to the job in question are filtered out.

  4. Filter out Negative Feedback: Candidates with previous negative feedback related to similar roles are filtered out to ensure quality matches.

For candidates who meet the initial criteria, Bullhorn Automation then employs AI Sourcing to refine the list further by focusing on job titles. It also takes into account skills, specialties, and industries to conduct a more detailed analysis. If a candidate has a relevant job title, they are prioritized for matching. However, if a candidate lacks a job title, they are not matched to jobs, as the model relies on job title data to make the best matches.

Results are also filtered based off of the maximum distance and minimum score configured. Results are presented in order of score (highest match) or if there was no score, then by radius (closest to furthest).

The distance (zip, city, state) logic will only apply if the Match Candidates Step is configured to Filter on Distance. If Filter on Distance is not enabled, matches will not be filtered or ordered based on candidates' locations.

Best Practice for Optimal Performance

To ensure optimal performance of Bullhorn Automation's matching features, it's essential to use the following standard fields exactly as intended. Custom fields are not compatible with Classic Matching, so adherence to these guidelines is crucial for achieving optimized results.

Required Standard Fields:

  • Job Title: The Title field on the job record.
  • Candidate Position Title: This should correspond to the occupation field in the ATS.
  • Location: Optimized for locations in the US, Canada, UK, Australia, and New Zealand.

Fields You May Need to Activate:

  • Skills
  • Specialties
  • Industries
  • Category

Using these fields correctly will maximize the effectiveness of Classic Matching and enhance the relevance of your matches.

FAQ

Does Classic Matching allow for selection ‘bias’?

To avoid any concerns regarding algorithmic bias, at no point of the process do we rely on any kind of personal candidate information such as age, gender, race, location or other factors.

Classic Matching is audited to ensure it is not biased. For questions regarding the audit please contact Compliance@bullhorn.com.

How does providing match feedback impact future matching logic? 

If a candidate/job match receives negative feedback, that feedback is stored in Bullhorn Automation and the matching is filtered out of all future matches so that the same result will not be delivered again. Neutral feedback undergoes manual review and does not automatically adjust the matching logic.

How much should I automate?

You may choose to start the candidate matching process by simply alerting recruiters about a new job and including the top matches.

Once you are confident in the matches, you can automate alerts to candidates about these roles. You can also use surveys to ask pre-screening questions and send calendar links to book time. The process would look like this:

  1. A new job is entered (this can be done automatically via VMS Sync).
  2. The best candidates are matched to that job.
  3. Matched candidates are sent a message.
  4. If the candidates are interested, they click a button to answer a few pre-screening questions.
  5. If they pass the pre-screen, a message can be sent directly from the recruiter with a calendar link.
  6. The recruiter is now talking to a candidate who is a good fit, interested, screened and scheduled.