The mentor-shopping advice you’ll find on page one of Google says: check the chemistry, clarify expectations, meet regularly. Forbes even has five C’s for it. All true, none sufficient. Chemistry is how bad mentors get hired: the pleasant conversation is exactly what they’re selling.
So score instead of feeling. Ten criteria, 0 to 10 points each, 100 total. Run every candidate through it, mentor or coach. Run it on me too; I put the links where you can check.
The 10 criteria
1. Seat time. Have they held the role you’re growing into, at a comparable scale? A zero means they read about it; a 10 means they lived it, recently enough that their playbook still matches today’s tooling and org sizes. This is the one criterion that’s mentor-specific, a coach plays a different game and is exempt.
2. Specific situations, on demand. Ask: “Tell me three situations like mine you have personally handled.” 0 = frameworks and adjectives. 10 = three concrete stories with what broke and what fixed it, told in under five minutes. No question I know filters more candidates than this one.
3. Answers, not only questions. If every answer comes back as another question, that’s a 0. A 10 means you leave the intro call with one move you can run tomorrow. Good mentors also coach when the moment calls for it, but they can always produce an answer.
4. Willingness to disagree with you. 0 = they validate everything, you feel great, nothing changes. 10 = the first conversation already contained one thing you didn’t want to hear. You’re not paying for agreement, that you can get at home for free.
5. Pattern volume. How many people like you have they worked with? You don’t want to be the experiment (0); you want your situation to match dozens they’ve seen (10). One warning here: volume without seat time just repeats the criterion-1 failure at scale, because watching 500 people struggle from the outside isn’t the same as operating. My number is 3,400+ sessions with 300+ leaders, and you should demand the equivalent from anyone.
6. Transparent pricing. 0 = “let’s discuss your goals first” and the price appears on a call, sized to your budget. A public number before you ever speak is the 10. Mine is on the pricing page, including the guarantee: a session you rate below 7/10 isn’t billed.
7. A milestone by session 4-6. An open-ended journey with progress measured in feelings is a 0. A named, checkable outcome inside the first two months is a 10. Across my mentees the first big milestone typically lands in 4 to 6 sessions, so a candidate should be able to name what yours would be.
8. An exit design. Ask: “When does this end?” Visible confusion, or “development never ends”, scores 0. A straight answer about making themselves unnecessary scores 10.
9. Checkable references. 0 = anonymous logos and initials. 10 = named people with faces and titles you can look up, ideally people you can reach. Mine are here, with names, and the average session rating behind them is 9.17/10.
10. Knows their own edges. 0 = takes every topic, including the ones that belong to a therapist. 10 = says “that is not my lane” out loud and refers you. Burnout that needs clinical help, legal trouble, personal crises: a great mentor names the boundary before you cross it.
Scoring
| Total | Verdict |
|---|---|
| 80-100 | Hire. This is rarer than it should be. |
| 60-79 | Take a trial session with a concrete situation and re-score. |
| Below 60 | Keep looking. A mediocre mentor costs more than no mentor, because you stop searching. |
Three rules while you score. First, score criteria 2 and 4 live, in the intro call: ignore what they claim, watch whether they push back and get specific. Second, when you’re vetting a coach, drop criterion 1 and score out of 90; the bands become 72 and 54. Third, for mentors no criterion compensates for a criterion 1 below 5.
Where this list comes from
I built the engineering org at Mews through Series C, from 8 teams to 80, and I’ve been full-time mentoring since 2019. The criteria are reverse-engineered from the other direction too: 52% of my mentees arrive in their first leadership role, and the ones who were burned before describe, almost every one of them, the same three failures: no seat time, no milestone, no end.
Since I told you to run it on me: on criterion 1 my score depends on your seat. I held VPE, not public-company CTO, so if you run a 2,000-person org, my seat time is partial and I say so on the intro call. That sentence is what a 10 on criterion 10 sounds like.
If you’re vetting an AI-specific mentor, the trap list is different and I wrote it separately: how to find an AI mentor.
And if you want to run the scorecard live on someone: my intro session is free, bring the checklist. Candidates who suggest that make it easy for you, which is itself a data point.