Our vision with Herefish is to automate busy work across the recruiting cycle, and Semantic Job Match is another step toward fulfilling that vision.
Semantic Match enables recruiters to consistently engage with their ATS talent and match the most relevant jobs to candidates. It uses Machine Learning from hundreds of millions of Bullhorn candidates, jobs, submissions and placements to score and match open jobs to candidates based on historical success.
Semantic match is available to Herefish customers with Full Licences. It is not available for those with a Blueprints Edition.
Currently, Semantic Match can be used to add dynamic jobs to your messages. This can be leveraged for many use cases, some of the most popular are outlined below:
When nurturing your talent pool - include some "Jobs you might be interested in." -- Start with the blueprint titled "Re-engage Passive Candidates" to set this up.
Automatically respond to all online applicants and keep them engaged by suggesting other roles they could be a good fit for. -- Learn how to set that up here or start with the blueprint titled "Engage Applicants" to set this up.
When a job has closed, send all the candidates who did not get the job a message they didn't get the role, but to check out some of your other openings. The same could apply as soon as someone is rejected for a job. -- Learn how to set that up here or start with the blueprint titled "Notify Candidates When Job Closes" to set this up.
When a candidate's assignment is ending, accelerate redeployment by suggesting roles they might be interested in. -- Start with the blueprint titled "Redeploy Candidates via..." to set this up.
Enabling Semantic Match
NOTE📝 - The following URL structure for your website's jobs page may be required to enable Semantic Job Matching.
https://example.com/jobs/%ATS_JOB_ID%/. The %ATS_JOB_ID% would be representative of the Job ID as listed in your ATS. This would mean that your job board MUST be connected to your ATS.
Contact Herefish Support to ask about additional options for configuring Semantic Match if you're unable to change your website's job page URL structure as stated above. Begin a chat message with Herefish Support (or email email@example.com) and type: "More options for Semantic Match URL structure" and we'll respond with the steps to send to your webdeveloper to make this work for your company.
Once the above is addressed, you have set up Semantic match inside of Herefish as it will apply your rules for all suggested job matches throughout the system. To do that, you will take the following steps:
Go to Settings > Website Configuration > Scroll to Semantic Job Match > Click Edit
Once you've clicked edit, you will open the setting screen. From there you will make 4 selections and your settings will be complete:
Included Jobs (required) - This will be a dynamic list of jobs that Herefish will analyze to determine the best match for the candidate. We'd suggest at least including criteria for jobs that are open AND posted to your website. You can add any additional criteria you'd like to further segment your main list of jobs.
Job URL Structure (required) - When a job is included in a message, we will include the link to that job with a specific generated URL structure, this section is how we will determine that link. If the beginning of your website URL does not reflect one of the following URL structures:
Then please contact Herefish Support using the chat icon for more information regarding this setting.
This also means that your ATS HAS to be configured to post jobs on your company's website directly through your ATS.
Maximum Distance (optional) - This allows you to only match jobs that are near the candidate. Even if you don't have a maximum distance entered, Herefish will still give preference to jobs that are closer to the candidate.
Using Semantic Match
As outlined above, most of the usage of the semantic matching will happen automatically as part of a workflow. However, especially as you are just starting out, we'd recommend you review some of the matches using your data. To do that, you can take the following steps:
Click on "contacts" to run a candidate search OR search for a known candidate in the search to get to their detail page. -- For this, we suggest using a real candidate, rather than a test contact. We look at many data points that won't be accurate for someone who is just there as a test candidate.
Scroll down to the "Semantic Jobs Match" section of the candidate's profile.
You will see the top 5 roles based on our Semantic Matching Model. If you were to merge the top jobs into an email, these are the ones we would include for that candidate.
Some key points to remember, each time you view the candidate's page or we send a message, we determine the best jobs at that time, based on the jobs that match your list when you set up Semantic Matching. This means that the matches are always changing depending on changes to the candidate or open jobs that match your rules.
You'll see the top job not only has a high match score, it also matches the skill and industry. The other jobs all have a very low score and even though they match other fields, they are still ranked as "Poor."
Can Semantic Match find the best candidates for newly opened roles?
This is the first release of Semantic Job Match, helping recruiters match jobs to candidates. It is our intention to extend the functionality to match candidates to jobs along with additional capabilities in future releases. This current functionality is an important step in that progression, it allows our matching model to get more real-life feedback based on how candidates interact with the jobs they are sent and the success rate for jobs that we've matched.
What data does Herefish consider when matching?
Semantic Match allows you to Match Jobs to Candidates that are similar in Title, Location, Category, Skills, Specialties or Industries. This is optimized for English-speaking countries at this time.
Semantic Match Potential Data Points
Job Title/Candidate Position Title
Fields that you may need to activate
If you create a list and do not see Candidate Skills, Speciality, or Industries as an available list search criteria in Herefish, then we are not syncing those fields. If you are using this and would like to enable them in Herefish, submit a quick chat to the support team below to get this synced!
When is Semantic Match available?
Semantic Job Match is available to all non-Blueprint Herefish customers today. We are eager to hear your feedback to continue to improve the match quality and experience.
I configured the settings for Semantic Matching, should this work immediately?
Not quite. If your current URL structure for your job board is entered and configured correctly, you can reach out to Herefish Support and the team can enable this setting for you.
If I already have Herefish job syncing set up, can I still use Semantic Job matching?
Yes! You'd still need to configure Semantic Matching using the steps outlined above, but if you begin using Semantic Matching, we no longer consider your previous job syncing settings. Contact Herefish Support using the chat icon for further questions about this.
What if I want to use location, but don't have a zip code for my candidates?
When identifying job/candidate locations, we use zip code OR the city/state. If they have either of those two entered properly, we will be able to identify their location. For reference, roughly 85% of candidates from all of our customers have a location entered.
What if there aren't good matches? Will a message ever show no jobs?
Regardless of the data quality, it will find the number of jobs you've included in your message to a candidates (Unless you attempt to add 3 Herefish job links in an email, but only have 1 job in your system).
The cleaner the data, the more potential matches that job will have. For example, updating the candidates job title with their most recent placement title, is likely more relevant than the title they had when they were sourced.
What are the limitations?
The current version requires the %ATS_JOB_ID% to be present in the url structure as example.com/jobs/%ATS_JOB_ID%. We'll be changing this in future releases, but today this is a requirement.
Currently, Semantic Job Match only matches ‘Jobs to Candidate’, we can’t use it yet for matching ‘Candidates to Jobs’.
Other things that you discover during the early preview period! We want to hear your feedback about what you'd like to see!
Does Semantic Job Match 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.
What does this mean for our search partners in our Bullhorn Partner Marketplace? Does this replace them?
Whilst streamlining the job matching process, Semantic Job Match is not intended to encroach on the offerings of our search partners in the Bullhorn Marketplace.
Our marketplace partners offer deep search and match functionality and are still a key part of our ecosystem. Semantic Job Match aims to solve the most common problems our customers have; matching jobs to candidates and vice-versa, with no intention of overtaking the search capability of partners like TextKernal or Daxtra.