Skip to main content
4.2 Matching candidates*

Understanding the Candidate and Order Matching Process in Globus AI

Pål Torgersen avatar
Written by Pål Torgersen
Updated over a year ago

After an order is created, a number of candidates will be suggested. The candidates are suggested based on the configured rules of your environment, information in the order, and information on the candidates. This is discussed during the implementation with your customer success representative.

Example of rule weights and available rules:

Rule

Rule details

Data

Weight (max 100%)

Type weights

Exclusion

Work history
(Worked there before, works there now and will work there in the future)

Customer or Department/project level

Work history from ATS or generated in Globus

Customer / Project 25%

Dynamic

No

Availability

-

Availability requests + assignments booked

20%

Dynamic

Yes (if fully unavailable)

Time preference

Candidate can report in availability requests

In Globus

5%

Flat

No (optional)

Role

-

From ATS attribute or set in Globus

15%

Flat

Yes

Competence/skills

-

From ATS attribute or set in Globus

15%

Dynamic (% of skills added in order)

Yes, if mandatory skills added to the order

Location

-

Candidates address from ATS or set in Globus

5%

Flat

No

Preferred location

-

From ATS (if available)

10%

Flat

No

Activity

How active is the candidate

From work history in ATS and generated in Globus

5%

Dynamic

No

Customer blocking

-

From ATS (if available)

-

-

Yes

Dynamic means that the percentage given to each candidate will vary. Ex, for work history, Globus will calculate the total number of work history on a project for all candidates and split the percentage of hours worked based on the percentage of the total amount of hours. Meaning, that a candidate who has worked 16 hours for a specific customer/project will get a much lower percentage score than a candidate that has worked 600 hours.

Flat means all candidates will get the same percentage score if they match the rule.

Your environment is also configured to match only active candidates or active and lead candidates. It is also possible to set a minimum percentage score needed to be recommended and a maximum number of recommended candidates.

Each recommended candidate will receive a percentage score that shows how much the candidate is fit for the role so you can easily compare and select between candidates. By clicking the percentage score you can get a detailed list of how the total score was calculated.

Modify candidate recommendations

If you are not satisfied with the recommendations or want to further limit the recommended list of candidates, there are two options available. You can:

  • Editing the order and add more skills, roles etc. to improve the level of detail (there is also the option to add default skills and role on the projects card under customer)

  • Or remove/select more suitable candidates using the "Adjust recommendations"-option on the top right of the list of candidates.

Here it is possible to add additional roles to include and filter on any data (1) similar to what is done in candidate segments. The filters set will override the recommender. Use the reset button (2) to revert to matching on the original details in the order.

All individual filters selected (Rolem Tag, Work etc.) are connected by AND, meaning the candidate must meet all criteria set for each filter. But within each filter, it is possible to choose ANY.


So an example could be:

Role - ANY of these - "Nurse" OR "Healthcare worker"
AND
Skills & Competencies - ANY of these - "System X" OR "Procedure Y" OR "Procedure X"

So a candidate with Role = "Nurse" and with one of the skills "System X", "Procedure Y" or "Procedure X" will match

For a "hard" filter you can utilize "all of these", then the candidate must meet all criteria set.

The recommender will become more accurate over time, but if you wish to make any changes to the recommender rules, contact our Support Team at support@globus.ai to make these adjustments.

Did this answer your question?