Configure your Pairing Algorithm:
In this next section, you will be able to set up your pairing algorithm. Here you can see the criteria that our algorithm will take into consideration when trying to pair users together. Keep in mind, that you can always change these rules later prior to launching pairing for your program.
To edit any of these rules, click on them. To add a new rule, click on the + Add Rule button under the respective type of rule you'd like to create.
There are two primary types of rules that you might want to consider creating:
Pairing Requirement Rules are what we typically consider "hard" rules that must be met in order for a pairing recommendation to be made by our algorithm. In other words, users must meet these requirement rules to be paired with another user. If they donβt meet these requirements with other users, they will not be allowed to be paired together. For example, you might want both members of a pairing to be from the same division within your company.
-
-
These rules can enforce two different types of conditions: either a response Matches or Does not Match.
For example, you could have a rule that states a Mentee's Response To What is your current role? Matches Mentor's Response To What is your current role?
This means that if the mentee and mentor's answer to "What is your current role?" is the same answer (i.e. matches each other), the algorithm will pair the users together, based on this rule alone.
-
There are two specific requirement rules that you should consider before inputting any of your custom rules.
- Timezone differences: you might want to consider requiring that pairings be within a certain amount of hours away from each other depending on their time zones. To implement this, you can click on the Timezone Difference rule, check the box, and select the number of hours of difference you want to enforce. For example, if you select 4 hours, the mentee and mentor can only pair if they have less than or equal to 4 hours of timezone difference.
- Reporting Lines (employeeID and supervisorID Data Required): Checkmark the box under this rule to enable a restriction if you do not want direct reports paired with their manager, manager's manager, or anyone who reports to the same manager. Be sure to click the Save button once you've made your selection.
Best Practices:
- Try and limit the number of Requirement Rules. The more there are, the more strict the pairing opportunities for your users. Reducing the number of these rules will empower your mentees and mentors more in the pairing decision process by opening up the number of potential pairings for them.
- Consider whether one of your Requirement Rules could be a Recommendation Rule with a higher weight instead if the rule does not need to be strictly met for all of your pairings.
- Explore using a custom weighting scale to increase the level of granularity of your recommendation rules outside of the standard 1, 3, and 5 priority scale.
Pairing Recommendation Rules on the other hand are typically considered "soft" rules. These rules are not mandatory for users to be paired together, but they instead work to improve the match strength between users. Be sure to prioritize what is most important for your program. For example, you might want the pairing to share similar interests, but it is not mandatory that they do. With these rules you can give each a different level of priority/weighting; the higher the priority, the more influence the rule has when pairing users together. Be sure to prioritize what is most important for your program, as this will be the basis of how we present top pairing recommendations. These numbers assist with tipping the scale so that there are fewer chances of a tie where 2 or more mentors share the same match percentage for a mentee. You may also create a custom weight if needed.
Once you have adjusted each of your Pairing Rules accordingly, be sure to click the blue Next button in the bottom right corner.
To continue building your Evergreen program, head over to Step 5 - Setting up Program Access
Have more questions? Submit a request here and let us know how we can help!
Comments
0 comments
Please sign in to leave a comment.