A mentorship pilot, rebuilt from feedback

GitHub’s Adacats employee resource group ran a pilot career-advancement program within engineering in mid-2019, pairing members who wanted to grow with more senior staff. After six months of monthly check-ins and surveys, the organizers had a clear picture of what wasn’t working—and used that feedback to redesign the initiative before rolling it out company-wide in May 2020.

What the pilot got wrong

The initial structure was deliberately light: pairs were told to meet regularly and make a growth plan, but were given little else. The gaps showed up quickly in feedback from the 10 pilot pairs:

  • Expectations ran hot. Mentees and their managers often disagreed on realistic timelines for promotion or skill improvement, and mentees frequently expected faster advancement than their mentors or managers did.
  • People wanted more structure. The lack of documentation left participants unsure what a productive session or growth plan should look like.
  • Managers were left out. A mentee's manager didn't know what the pair was working on, how much performance information should be shared with the mentor, or what part they were expected to play.

Iterating toward a clearer program

The Adacats response addressed each failure mode directly:

  • A name change shifted the focus. The program was rebranded as the Adacats Mentorship Program to emphasize holistic career growth instead of quick promotions, which the organizers describe as only one form of advancement.
  • Documentation was added. New materials included a detailed growth-plan template for mentees, role expectations, mentoring resources focused on underrepresented groups, and public guides to GitHub’s promotion process.
  • Managers got a playbook. A dedicated manager guide explained the program's purpose and how managers could support it. An initial three-way meeting (mentor, mentee, and manager) set shared expectations about what information would flow between them, and six-month check-ins with the same trio kept everyone aligned on progress.

The organizers say the structure was always meant as a scaffold, not a script. Kickoff meetings remind pairs that the documents are guidelines, not mandates, and that participants are free to adopt the tools that work best for their particular growth goals.

Lessons for anyone starting a program

Looking back, the Adacats team distills the experience into several takeaways:

  • Ship to learn. Iterating based on actively solicited feedback let a volunteer-driven group build the program incrementally rather than attempting a full-fledged rollout upfront.
  • Bring managers into the loop early. Leaving them out at the start caused confusion and avoidable pushback.
  • Promote the program where decisions happen. Presenting at manager and staff meetings increased sign-ups and gave potential participants and their managers a chance to ask questions directly.
  • Marginalized groups need specific kinds of help. Mentors can support mentees by reviewing their self-reviews, and including sponsorship training is critical, since people from underrepresented groups tend to be over-coached and under-sponsored.
  • Don't underestimate the appeal of junior mentors. The program saw a shortage of mentors because some potential volunteers felt they weren't senior enough. In practice, mentees valued mentors who were closer to their own experience level.

The program had concrete results beyond positive feedback: participants were promoted, moved to more suitable teams, or gained skills and opportunities they otherwise wouldn't have had.