Last time I wrote about the transition from using a shared email account to Zendesk. Like most of the things we’ve tried it’s a technical solution. This time I want to write about something far trickier: people.

Stack Exchange sites are a little unusual since the community runs them. That’s not just a nice way to say that a few users sit on the Star Chamber. Our community has authority to do just about everything through a system of reputation, privileges, elected or appointed moderators, review queues, and flagging. As a result, hardly any of the problems that crop up on our sites make it to [email protected]. The major exceptions involve account problems that require more access to personal information than we give to our volunteer moderators.

Since I had experience as a moderator before being hired, I was about as well prepared to handle Stack Exchange customer support as anyone can be. Even so, I needed a few weeks of training in order to take on a solo shift. The few first shifts consisted of a ride along with an experienced Community Manager using screen sharing on Google Hangouts. Like pair programming, the mentor starts by driving while asking the trainee what they would do on each ticket.

Later, the roles switch so that the more senior of the pair observes as the new CM handle tickets. Eventually, the new employee is ready to fly solo. But they aren’t completely on their own. We have an internal chat room dedicated to helping each other with tough Zendesk tickets. If any of us get stuck, we drop a link in chat and the rest of the team chimes in with suggestions.

Our approach to customer support takes inspiration from Joel Spolsky’s “Seven steps to remarkable customer service”. We aren’t doing customer support over the phone for a company that sells software, however, so we apply some of the principles a bit differently:

1. Fix everything two ways

For the my first year or so on the job, we did day-long [email protected] shifts. Typically, the day would start by forgetting that it was my day already. “Didn’t I just finish my shift a day or two ago? No? It’s been a week already? Well, I guess I better get started after I wrap up that thing from yesterday.”

When I finally did start my shift, it would be nearly time for dinner and my family would be ready for work time to end and Daddy time to start. Around midnight, I’d give up and promise to finish the next day. (This is probably why it never seemed like it had been more than a day or two since I’d handled support tickets.)

As you can imagine, I was so busy struggling to fix everything once, I didn’t have time to think about underlying problems.

So a few months ago, we bit the bullet and tried week-long shifts. While we still are swamped with tickets, the change has given us time to take a big picture view every now and then.

Last week, for instance, I noticed a strange ticket from an Uber customer reporting problem with a particular driver. Oddly, the person identified themselves as an employee of Disney working just across town from me. Digging through our access logs, I discovered the user came to our site via the Uber API support page. No wonder the Disney employee was so confused!

Now that I knew what the underlying problem was, I set about fixing it:

  1. I contacted the Uber API support email to let them know about the tag name and give them our standard advice about using Stack Overflow for product support and

  2. I created a tag synonym from [uber] to [uber-api].

The next person who follows the Uber API link will see some questions that clearly have nothing to do with Uber drivers. With any luck, they’ll know they are in the wrong place and find the correct support channel more efficiently.

2. Suggest the account recovery page

We don’t do hardware support, but we do have lots of problems with people forgetting how to log into our sites:

  • OpenID can be one level of indirection too far for some folks.
  • Our login system is pretty good, but it could probably be better.
  • OpenID providers go down or go out of business on occasion.
  • Since we log people in automatically, there’s often no need to remember your credentials until you get a new computer or device.

So people can get locked out of there account without any idea how to get back in.

In those cases, an account recovery email can be a lifesaver. It provides detailed and individualized instructions about how to use your own set of credentials.

But those instructions can also be useful in other situations:

  • When a user will writes in to ask why they can’t comment on their question any more or where their reputation has gone, it might be because they are logged out.
  • People occasionally try to use their Gmail address to sign into Stack Exchange OpenID when they used Google OpenID or vice versa.
  • Sometimes people forget they even have an account with us.

In cases like these, I’ll send them the link to the account recovery page and ask them to follow the directions.

I was planning on writing out the other 5 steps, but it’s my [email protected] week and this has turned into procrastination time.

Next time: more of Joel’s Seven Steps.