CM escalations are one type of request that the Community Management team handles. These are issues that Moderators send to the Community Managers when an issue needs to be escalated to someone on staff. Common reasons for escalations include moderators dealing with a complicated voting invalidation or sockpuppet investigation cases who need staff to dig further. Another common reason for an escalation would be if a moderator notifies the CM team about an issue via chat and it is going to take us more than a few minutes to investigate and resolve; we’ll ask that an escalation be submitted.

On average, we receive over 120 escalations each month. When we’re short-staffed or pulled into time-sensitive situations, the escalations can pile up, leaving us with a major backlog. This time last year, the team had to work on a ticket smash because we had a backlog of over 600 CM escalations.

We weren’t at that level of a backlog this summer, but escalations were piling up. I’m sharing how the team avoided the major backlog we had last year and our plan for mitigating backlogs in the future.

Out with the old approach

At the beginning of the year, we shared that the Community Team now had three subteams and areas of focus: Community Operations, Curator Support, and Trust & Safety. Even with the split, there are certain tasks that are a shared responsibility across all teams. One of those shared responsibilities is Community Emergency Duty (CED). Each week, the Community Managers rotate who is on duty to handle issues that get flagged as emergencies. These issues include things like removing underage user accounts from sites and reaching out to users when it is reported that they have said something that indicates they may be in danger of self-harm. These issues get specialized attention so that we can quickly get appropriate resources to those in need.

During the first half of this year, the Community Manager who was on CED was also responsible for handling the non-emergency CM escalations. By the end of Q2, we realized that wasn’t the best process. If the person on CED had a lot on their plate during that week, they wouldn’t be able to get to many of the CM escalation notices that weren’t true emergencies, leaving the next CM on duty dealing with a backlog. We built better internal notices to warn the team if the queue was getting to 50 or more open escalations and would try to set aside time for all of us to crush some outstanding escalations out; however, with the team being short-staffed at the end of Q2, that solution wasn’t proving to be successful either.

Trying something new

At the beginning of Q3, we decided that the Community Operations team would handle triaging CM escalations. The majority of issues that get sent to us are for the Ops team to handle anyway. During triage, if something falls under Curator Support or Trust & Safety’s purview, we are able to forward those issues to those team members.

We were all excited about this new process; however, getting it off the ground proved to be more complicated than we thought it would be. We were still dealing with the backlog from Q2, in the process of onboarding and training new team members, and due to vacations and other summer commitments still a bit short-staffed in July and August. We were getting through escalations, but if we could only get through 20 escalations a week and another 20-30 were coming in every week, this meant we were only breaking even or, in some weeks, still falling behind. We were hovering at about 60-90 open escalations a week at the peak of this summer.

This is when the team pulled together and focused on a solution to avert the backlog of last fall. Catija and JNat spent time training our new Ops members Slate and Charlotte on the ins-and-outs of CM escalations. Even though Cesar is on the Trust & Safety side of the team, he graciously volunteered to also help with training since he has had a lot of experience handling these issues in the past. During the first half of September, the five of them spent time each week smashing out the backlog of escalations and working together on more complex ones.

This resulted in the CM escalation queue hitting zero on Friday, September 24th. A zero inbox on CM escalations is not something that I’ve seen in the 7+ months I’ve been on the team. Now, a zero inbox won’t last forever. Escalations will continue to come in, but we’re now in a sustainable place where we can triage and resolve them quicker. If we get to a point where the backlog is growing and this new triage process is proving not to be sustainable, we’ll re-evaluate and adjust things.

To give you an idea of how long these escalations can take here are some stats on CM escalation resolutions this year:

Between January and now the CM team has resolved 1105 escalations. Here is the breakout of how long it took to resolve these escalations

225 escalations were resolved in less than 24 hours

116 escalations were resolved in more than a day but less than three

155 escalations were resolved in more than three days but under a week

223 escalations were resolved between 7-14 days

259 escalations were resolved between 15-30 days

77 escalations took more than a month to resolve

I think it’s important that we share this story with you. Process may not be the most exciting thing to read about, but we’re committed to improving how quickly we can respond and handle issues that you bring to our attention. My goal is to come back to you in early 2022 to share the stats under our new process which will hopefully show a faster time to resolution. Also, every time we talk to community members, they express an interest in how things work “behind the curtain”. I also want to thank Catija, Cesar, JNat, Slate, and Charlotte for their hard work in crushing this backlog. They made it their priority because they care so deeply about improving everyone’s experience - including yours.