The thing that surprised me most about why a mentor and mentee work well together: skills are only part of it.
Before running our cross-company mentoring pilot, I assumed the most important thing for someone looking for a mentor was their skills. What they'd achieved. Which company they worked at. If you're setting up a mentoring program, this is probably your assumption too — it was mine, and it's the default way most people approach how to select mentors and mentees.
I based that on my own experience and on what mentees told me directly. Then I ran the pilot, and some pairs that looked great on paper didn't click.
Skills were clearly only part of the equation.
Why skills-based matching isn't enough
There was something else people looked for in a mentor, mostly subconsciously, that made them actually value the support they got. And it wasn't the same thing for every mentee.
So I spent months studying what makes a mentoring pair click, not just what makes them look compatible on a form. A few patterns came out clearly.
What actually predicts a good mentor-mentee match
Background similarity matters — for some people. Some mentees valued a mentor who shared their background: same industry, similar education path, a comparable career arc. For this group, "someone who's been where I am" was the whole point.
Personal characteristics matter more than background — for others. For a different set of mentees, shared history barely registered. What mattered was how the mentor showed up: their communication style, their vibe, whether they felt easy to be honest with.
Values are a stronger driver than most programs account for. Research points at values, specifically achievement and ambition, as some of the strongest drivers of affinity between a mentor and mentee. Two people who are both equally driven recognize it in each other fast, even across very different industries or career stages.
None of these replace skills. They sit alongside them. A mentee still needs a mentor with relevant experience to draw on. But experience alone doesn't predict whether the relationship will hold up past the first two sessions.
Building this into a real matching process
I worked these data points — background fit, personal characteristics, and values alignment — into a new matching algorithm for PeopleTopics' next cohort. Instead of matching primarily on title, seniority, or industry overlap, it weighs the dimensions that actually predicted a strong connection in the pilot.
This came directly out of our first cross-company pilot: 18 participants, 9 pairs, an average recommend score of 8.4 out of 10, and the three pairs that didn't click. We reported that last part honestly, and it's exactly what led to this.
What this means if you're setting up a mentoring program
If you're an HR or L&D lead building a mentoring program: don't stop at matching people by title or years of experience. Ask what each person is actually looking for in a mentor relationship — shared background, personality fit, or shared drive — and build your matching process around the answers.
You won't get every pair right. We didn't, and we run this for a living. But the difference between a program where most pairs click and one that quietly fizzles is usually decided here, at selection, before a single session happens.
And if you'd rather not build the matching machinery yourself, that's exactly what our cross-company cohorts exist for. The next one starts in October.
Illustrations by Storyset