Diversity Mentoring Program: DEI and ERG Mentoring Guide

A diversity mentoring program (also called an inclusive, DEI or ERG mentoring program) is structured mentoring designed to widen access to mentors, networks and opportunities for people who have had less of it. It combines open eligibility, fair and opt-in matching, protected personal data and honest measurement of who takes part and how they progress.

This guide is for HR, talent and inclusion teams and for employee resource group (ERG) leads who want to build or improve such a program. It covers the main program models, matching without stereotyping, handling sensitive data, measurement and the pitfalls that quietly undermine these programs. For other formats, browse all our mentoring guides.

What does an inclusive mentoring program try to fix?

In most organizations mentoring already happens informally. Senior people tend to mentor those who remind them of themselves, who sit near them or who were introduced by someone they trust. Nobody designs it that way, but the result is that access to advice and advocacy follows existing networks.

An inclusive mentoring program makes access deliberate. It does three things:

• Opens the door. Anyone who fits the program goal can apply, through a published process, not a tap on the shoulder.
• Widens the network. Pairs and circles cross teams, levels, locations and backgrounds, so participants meet people they would not have met otherwise.
• Makes opportunity visible. Mentors and sponsors learn who is ready for a stretch assignment, and participants learn how decisions are actually made.

Keep the goal concrete. “Increase access to senior mentors for early-career employees in regional offices” or “give ERG members a structured path to leadership conversations” will guide design decisions. A broad statement of values will not.

Which program models work for DEI and ERG mentoring?

There is no single model. Most organizations combine two or three of the formats below.

ERG-led circles are often the easiest start, because the ERG already has members, trust and topics. Give the ERG a clear structure, one or two experienced mentors per circle and a session plan. Our group mentoring guide covers circle size, rhythm and a sample agenda.

Cross-difference 1:1 mentoring pairs people across lines that rarely meet: a junior analyst in a regional office with a senior manager at headquarters, an engineer with a commercial leader. The point is a wider network and a different perspective, not a match by demographic category.

Reverse mentoring turns the usual direction around. It helps leaders hear directly how policies and everyday habits land for colleagues with different experiences. See the full setup in our reverse mentoring guide.

Sponsorship tracks add advocacy. A mentor advises; a sponsor uses their influence to open a door. Many diversity programs find that mentoring alone builds confidence but not always access to roles, so they pair a mentoring phase with a sponsorship commitment. The difference is explained in mentoring vs sponsorship. If your focus is gender, our women’s mentoring program guide goes deeper on that design.

How do you design the program step by step?

1. Write one goal sentence that names the group you want to reach and the outcome you expect, such as access, readiness for promotion or retention through the first two years.
2. Set open eligibility. Describe who the program is for in terms of goals and situation (“first-time managers”, “employees in their first year”, “ERG members preparing for leadership roles”). Check whether your local rules allow programs restricted to specific groups, and prefer goal-based wording where they do not.
3. Recruit mentors widely. Ask leaders from every part of the organization, not only volunteers who already champion inclusion. Brief them on listening, confidentiality and avoiding assumptions.
4. Choose formats. For example: ERG circles for peer support, 1:1 cross-difference pairs for individual goals, and a small sponsorship cohort for people close to a promotion.
5. Design fair matching (see the next section).
6. Set the cycle. Six to twelve months with meetings every two to four weeks is a common rhythm. Fix start, midpoint and end dates.
7. Plan topics. Give early sessions a structure: goals, career story, navigating the organization, visibility, a stretch opportunity.
8. Decide what you will measure and how you will protect the data before anyone applies.

How do you match fairly without stereotyping?

Matching is where good intentions can go wrong. Pairing people only because they share a demographic label assumes what they need. Pairing them only across labels can feel like being used as a teaching resource. The answer is to let participants choose.

• Match on goals and experience first. The mentee’s goal (“move from operations into product”, “prepare for my first leadership role”) and the mentor’s real experience are the strongest signals.
• Make self-identification opt-in. If you ask about background at all, make every such question optional, offer “prefer not to say” and explain how the answer is used.
• Ask for preferences, not categories. A question like “Would you prefer a mentor who shares an aspect of your background or experience? If so, which?” lets the mentee decide what matters. Many will say no preference.
• Keep hard constraints. Shared language, workable time zones and no direct reporting line between mentor and mentee.
• Spread mentor load. Mentors from underrepresented groups are often asked to mentor far more people than others. Cap the number of mentees per mentor and recruit more widely instead.
• Keep a human review. Have a coordinator check proposed pairs before announcing them, and offer a no-fault rematch after the first few meetings.

Our guide on how to match mentors and mentees covers matching criteria and methods in detail, and the mentor–mentee matching questionnaire gives you ready-to-use application questions you can adapt, including optional preference fields.

How should you handle sensitive personal data?

Information about ethnicity, religion, health, disability or sexual orientation is treated as sensitive in many legal systems. Under the GDPR in the EU and the UK GDPR, these are “special category data” with stricter conditions for processing. Other countries have their own data protection laws with similar categories.

General good practice, which you should confirm with your data protection officer or legal team:

• Collect only what you need for matching or measurement. If a question does not change a match or a report, drop it.
• Make it optional and explained. Say why you ask, who can see the answer and how long you keep it.
• Separate identity from analysis. Report participation by group only in aggregate, and only where groups are large enough that no individual can be identified.
• Restrict access. Program admins need to know a pair exists and is meeting. They rarely need someone’s self-identification, and they never need conversation content.
• Let people withdraw their information and leave the program without explanation.

This is general guidance, not legal advice. Rules on what you may ask, and on programs aimed at specific groups, differ by country.

How do you measure participation and progression?

Measure what the program can honestly show, in layers:

1. Reach. Who applied, who was accepted, who was matched and how long they waited. Compare the applicant pool with the population you meant to reach. A gap here is the first thing to fix.
2. Participation. Meetings held per pair or circle, sessions completed and drop-outs. A pair that has not met in five weeks needs a nudge now, not a line in the final report.
3. Experience. Short feedback after sessions and at midpoint: is the relationship useful, does the participant feel respected, would they recommend it?
4. Goal progress. Each mentee sets two or three goals at the start and rates progress at midpoint and the end.
5. Longer-term signals. Promotions, internal moves, stretch assignments and retention over 12 to 24 months, compared carefully with similar non-participants where you have enough data. Report them as what happened to participants, not as proof the program caused it.

Agree the report format with leadership before launch. For a full evaluation framework, see our guide to mentoring program evaluation.

What are the common pitfalls?

• Making it a side project of the ERG alone. ERG leads are usually volunteers. Without HR support, budget and senior mentors, circles fade after a few months.
• Matching on labels. Assuming every member of a group wants a mentor from the same group, or that they want to educate colleagues about it.
• Overloading a few mentors. The same small set of visible leaders end up mentoring everyone and burn out.
• Advice without access. Participants grow in confidence but see no new opportunities. A sponsorship element or clear links to talent reviews closes the gap.
• Collecting too much data. Sensitive fields gathered “just in case” create risk and erode trust.
• Announcing outcomes the program cannot prove. Report reach, participation and goal progress honestly. Credibility matters more than a headline.
• Exclusive framing. Programs that feel like they are for some and against others lose support. Describe the program by its goal: widening access to mentoring for people who have had less of it.

How can software help you run it?

If you run a diversity or ERG mentoring program, the work that takes the most time is matching, scheduling, reminders and reporting. Mentornity’s employee mentoring software handles that: rule-based matching on the fields you choose, with weights and mandatory filters, where admins review each proposed pair side by side before approving it; 1:1 and group mentoring in the same program; sessions with set topics; private notes that stay private; and participation reports you can export. It is free for up to 10 users, so you can try it for free with a small pilot group, for example one ERG circle, before a wider launch.