How to Match Mentors and Mentees: Criteria & Methods

Rules, self-selection and hybrids all work as matching methods. What matters more is clear criteria, a process people see as fair and matching people quickly.

The three matching methods

Admin decidesMentee decides

1. Admin rulesSoftware scores matches, admin approves
2. HybridAlgorithm proposes, a person confirms
3. Self-selectionMentees browse and request

Admin matching by rules. You define criteria, software scores every possible match, and you review and approve. It is fair, fast and easy to defend, and it scales past a few dozen people. It needs decent profile data.

Mentee self-selection. Mentees browse mentor profiles and request a match. It fits communities where hierarchy would feel wrong. Cap requests per mentor, nudge mentees who stall, and keep a fallback for those who find no one.

Hybrid. The algorithm proposes a shortlist and the mentee picks, or the mentor confirms. Giving people a say has support: in a study of formal programs, perceived input into the mentoring process was consistently related to better mentoring and mentorship quality (Allen, Eby & Lentz, Journal of Applied Psychology, 2006).

Choose per program. An accelerator can run both: lead mentors assigned by rules, plus open office hours any founder can book.

Which criteria to use

In rough order:

1. Hard constraints. Shared language, workable time zones, no direct reporting line (for employee programs). Make these required rules that remove a match entirely, not just lower its score.
2. Goal and experience. The mentee’s goal (“move into product management”) should match the mentor’s actual experience, not their job title. Ask about goals in structured fields on the application form, not only free text.
3. Background, same or different. Same department transfers skills; different departments open networks and candor. Pick per program goal and set it as a same/different rule.
4. Career-stage gap. A reasonable default is two or more levels, or around five years, of extra experience.
5. Shared interests, lightly weighted. They help the first conversation but do not carry six months.

Keep it to two or three weighted criteria plus one or two required rules. Each extra factor dilutes the ones that matter.

How rule scoring works in Mentornity

The scoring is simple enough to explain to anyone who questions a match. Each rule compares a mentor field with a mentee field (same, different, similar, equal, greater) and carries a weight. A match’s score is:

matched-rule weight ÷ total rule weight × 100

With four equally weighted rules and three matched, the match scores 75%. A higher weight makes a rule count for more.

1. 1Define rulesFields, comparison and weight
2. 2ExcludeMatches failing a required rule
3. 3ScoreEvery remaining match
4. 4AssignBest matches within each person's capacity
5. 5ReviewAdmin approves one by one or all

Matches scoring 0% are never assigned automatically. In the admin panel you click Calculate on Matching Algorithm, review the proposed matches, and approve them one at a time or all at once. Given the same data, the result is always the same. Every match action is recorded on Matching Actions, so you can answer “why these two?” with the rules rather than a hunch.

What a fair process looks like

• Transparency. Publish the criteria at kickoff. Visible rules prevent speculation about how matches were made.
• Consent. Someone, mentor, mentee or admin, explicitly confirms each match.
• A no-fault exit. A rematch window at week three or four, with no reasons required. Forcing a match that is not working to continue can lose both people.

Common matching mistakes

• Matching on job titles. “Director” covers very different jobs. Match on stated experience and goals.
• Overloading star mentors. Giving the best five mentors eight mentees each burns out the people the program depends on. Set caps.
• Treating matching as one-off. Late joiners, rematches and new cohorts arrive soon. Save your rules and rerun scoring when needed; this is where software helps.
• Silent leftovers. Tell unmatched people what is happening. Offer group mentoring or the next cohort, and fix mentor supply next round. If a required rule excluded them, check whether it really needs to be required.

Matching is one step in a larger sequence; see how to start a mentoring program. To try the matching engine, create a free program for up to 10 users, define two rules and click Calculate.