Congratulations! You’ve created a new group to bring people together IRL.
People out there share your passion, but how will they find your group to make that connection happen? Good news: Meetup is here to help!
Shayak Banerjee, senior engineer on the Machine Learning team at Meetup, shares some insights about how we use algorithms to help connect your new group with the members who are most likely to be interested in joining.
It all starts when you create your Meetup group
When you create your group, you lay the foundation for finding members by choosing your group name, description, location, and topics. Think about what potential members are looking for in a Meetup group as you customize your group page to be clear, authentic, and welcoming.
Next: your Meetup reviewer
Once your new group is created, a community team member at Meetup HQ reviews the details (location, topics, name, and description). They’ll ensure everything aligns with our community guidelines, and may update your topics or filters to help you find your community.
If you’re bringing together people with specific characteristics (e.g., Moms of Orange County or Kayaking for People in their 30s and 40s), the reviewer will set those filters in order to find them.
The Meetup platform has tens of millions of users and processes close to 500+ new groups every day. Meetup balances the needs of organizers (publicizing new groups to encourage membership) with the needs of members (sending relevant emails). So how does Meetup actually determine which members to alert about your new group?
When you have this much data to sift through, it’s time to call in the machines. Over time, our computers get better at the task by using techniques known as “machine learning.”
Clear and detailed information (like location, topics, title, and description) means better output from the machines. Here are some ways the computers can build on the information you enter to find your potential members:
Location location location
Since Meetup groups involve meeting in-person, location is a strong component of our algorithms.
We use zip code, city, state, and country for groups and members to make a good match. We also look at other groups in the same zip code as you, their members, and where they come from.
Expanding the list of topics
Choosing topics is the best way for members to show interest in future groups. Our algorithms match members and groups that have selected the same topic. However, there are thousands of topics and an organizer can only pick 15 to define a group.
Don’t stress—Meetup automatically expands the list of topics to best find members interested in your group. Let’s say you picked snorkeling as a topic for your group. Based on topics selected by other Meetup groups and members, the algorithm will spot an overlap between snorkeling and scuba diving. Our emails will then reach out to members interested in both snorkeling AND scuba diving, telling them about your new community!
Meetup considers member-entered information like age and gender when promoting groups with a designated demographic, like Kayaking for People in their 30s and 40s.
However, not everyone specifies their age or gender, so Meetup may use clues like other groups they are part of, and the age/gender focus they have. The Machine Learning team is constantly refining and improving this process.
Active members (those who RSVP, join groups, send messages, etc.) are more likely to consider joining new groups. As a result, new group announcements go to active, engaged members.
Let’s do it!
Your Meetup matters. We’re excited to introduce your new group to potential members in your community and help you find your people!
Shayak Banerjee, Senior Machine Learning Engineer at Meetup.