After 5 1/2 years my time with Zindus, an open-source add-on for Mozilla Thunderbird is drawing to an end.
I got a lot of satisfaction from writing and supporting software that tens of thousands of people use every day. I would have loved to have found a way to make it sustainable. I couldn’t. Time to call quits. I have other things to work on.
I’m taking this opportunity to reflect on my experience developing and maintaining the add-on, beginning with the activity that’s taken the most time - email.
Email? Really? Not software development?
Nope. Email. Once the software has been written, that’s that. What’s left is bug fixes and support. Years and years of it.
4,265 support emails have landed in my Inbox between 1 Dec 2007 and 30 June 2013.
Two emails per day on average. More immediately following a release, less in between.
One thing I’m pleased about is that support volume isn’t correlated with the number of active daily users.
Active daily users has been flat the last 2 1/2 years, with new users taking the place of those who leave. Growth stalled because an add-on with a more descriptive name appeared on addons.mozilla.org.
What is support?
Ordered by volume, support is mainly about:
- how do I do XYZ? / can you add feature XYZ?
- minor corruptions of the thunderbird addressbook
- bugs in the zindus add-on
- bugs in thunderbird + bugs in the google api
The best support requests are from people who’ve experienced a problem that relates to a bug in the add-on. If I think someone has encountered a bug in the add-on I’m eager to find and fix it. Most support requests don’t relate to bugs in the add-on.
I want to say “thanks” to all those people who’ve worked with me in debugging the add-on. Help from users is critical to the functioning of the whole mozilla ecosystem.
I’m also going to hazard a national stereotype and offer special thanks to German users, who in my experience give the best bug reports. Polite, clear, and accompanied by logfiles.
The chart above understates the total volume of zindus-related email. Including mailing list traffic at mozilla and Google, bugzilla, and myriad other sources the email volume easily doubles.
The add-on has around 80,000 active daily users.
I’ve had contact with 2,807 unique email senders, around 3.5% of the zindus user base.
Lets see whether the User-Agent header tells any kind of story.
Thunderbird users want support for a Thunderbird add-on. No surprise there.
Number of replies
Most support requests are satisfied with the first email response. A few involve some back and forth.
The longest email thread ran to 22 replies.
Time to first response
“Time to first response” is a standard customer support metric.
The first three years were certainly better than the last two and a half. It’s really hard to fit good customer support into small chunks of free time.
Overall, support was super responsive. But a few people got a very slow response.
And some people got no response at all.
Actual response rates are probably better than this but the bottom line is: my passion for providing great customer support began to wane. To those folk who got no response - sorry!
Wait, isn’t there a better way?
There are definitely ways to handle this ton of email better. Filtering. Self-service. Auto-responders.
There were a few periods where customer support software would have been really handy. When Google’s APIs have a bug, it affects a lot of people in a short space of time. Everyone thinks the add-on is at fault. The flood of support requests can all be serviced with the same reply.
But I would ask myself the question “do I really want to be running a commercial-scale help desk?”. The answer to that was always “no”.
This is a natural point to begin exploring motivation and the absence of monetisation options in mozilla’s developer ecosystem but I’ll leave that for a separate post.
In the landscape of open-source software, the Zindus add-on is a tough customer support challenge for a number of reasons:
- desktop sofware
- the users are regular people, not other developers
- Thunderbird is old and clunky in parts
- sync has the job of reconciling different systems, and is exposed to problems from all sides
I don’t plan to run screaming into the night. I’ll continue to support the current release of the add-on, but won’t update it unless any security or stability issues emerge.
The add-on is 100% open-source and on github. If anyone is interested in taking over the maintenance, drop me a line. After reading this post, I suspect not :-)
But with that said, developing and maintaining software that lots of people use every day is a great experience. Open-source is a great route to that. I’d recommend it to anyone.