The success of any mentor or protégé relationship depends on a range of factors - from commitment and goals to personal chemistry. However, no mentoring relationship will stand much chance if the partners are poorly matched.
Matching is especially critical in structured e-mentoring programs such as MentorNet's because participants don't meet in person, and they typically don't share affinities such as organizational associations, peer groups, or geography.
To bring the right mentors together with the right protégés, MentorNet's online matching system uses a complex, customized algorithm that takes each participant's interests, background, and preferences into account. The challenge, says Stephanie Fox, MentorNet's director of technology, is to use the right profile data and preferences and to accurately weight those criteria. If the process is too strict in aligning participants' backgrounds and preferences, too few will match; if the parameters are too wide or key factors aren't taken into account, poor matches will result.
Student career goals, mentor's role get priority
Where MentorNet's goal is preprofessional development for students, closely aligning a mentor's position in industry or academia with a protégé's area of study and career goals is extremely important, so those criteria are weighted most heavily. MentorNet has also found through evaluations that it's often better to match students with mentors who are older, have a higher degree, and have more years of experience, so the algorithm attempts matches based on those demographics, says Fox.
To start the matching process, participants complete online profiles of basic information such as area of study or work, background, specialty, and so on, and rank their interest in such issues as balancing work and family, school decisions, résumé/job advice, ethnicity, sexual orientation, and the like. Protégés may state preferences for certain mentor characteristics: gender, ethnicity, professional background, employer or field, alma mater, and geographic location. Mentors' protégé preference choices include gender, education level, school, and citizenship.
This process not only provides the matching criteria that MentorNet finds work best, it also helps set expectations and reassures participants by providing insight into how the process works. Researchers have found that setting expectations and informing participants about the matching process gains their buy-in and confidence.
When a protégé requests a match she has several options. She may ask MentorNet to make the match for her, in which case MentorNet's algorithm and systems determine the top match based on the profiles and preferences of mentor and protégé. Or she may opt to choose from as many as five mentors that the system determines are the best matches. In this case, the protégé may review their anonymous profiles before making a selection. If none seems right, she may adjust her profile and run the matching process again, or wait up to two weeks to see if any newly enrolled mentors will be a better match. If protégés don't choose within two weeks, MentorNet makes their matches for them.
"So far, we've gotten great results with that," says Fox. "I feel we've done a very successful job at providing the best matches and the most matches."
Refining the matching system
MentorNet has continually refined its online matching system since launching seven years ago, based on program goals and annual program evaluations. For example, MentorNet knows that "forcing" matches contributed to a high failure rate, and where evaluations showed that participants felt that they didn't get good matches, MentorNet adjusted the matching algorithms to fine-tune weighting of characteristics that seemed to be particularly influential in making good matches. Protégés wanted a choice, and researchers have also found that choice is, indeed, a factor: "The least successful matches are typically those where the mentor and the mentee feel imposed upon each other," writes David Clutterbuck in his book Everyone Needs a Mentor.
Relying on choice alone didn't work, however: when left to their own devices, almost 50 percent of MentorNet protégés never chose a mentor. There are many reasons this occurred: MentorNet learned through its 2003-04 that while some students simply didn't like the choices they were provided, others were intimidated by the process, concerned they would make a "wrong" choice, felt unworthy to be paired with someone they perceived to be extremely accomplished, or were just too busy or tired to make the choice.
MentorNet found the right balance in its current system last March in creating systems that allow students to continue to choose a mentor if they want to do so or ask MentorNet to make the match for them. The result: only about 25 percent of protégés don't get matched, and in the vast majority of cases, the reason is a lack of appropriate mentors in the selection pool.
"If there's a match for [protégés] out there, we'll find it," says Fox. "It's always exciting to hear from mentors and students about how their matches worked out - that it was a terrific match and went really well."