Tl;dr: In the process of migrating the Support Driven (SD) website, I lost 8 of the podcast audio files from SD’s early days 😭. This is the story of how it happened, the lesson I learned, and where we are now.
ACT I: Done is better than perfect
In February of 2019, we migrated the Support Driven website from an external host back to WordPress.com, where the site had originally been hosted. We made the move for a few reasons. One is that a lot of folks from the support team at Automattic, the parent company of WordPress.com, are part of the Support Driven community, and when it makes sense, we like to go with companies who have members in the community. Another reason is that long-time SD member @justpat had been handling the SD website as a side gig and needed to step down. Without a dedicated website person, we wanted the site to be fairly self-contained and easy for us to maintain. As a former Happiness Engineer with Automattic, I felt comfortable in the WordPress.com ecosystem.
At the time of the website migration, we were a two-person team organizing Support Driven’s first European conference and the Kickstarter that went along with it. We were constantly making decisions on where it was most valuable to spend our time. We’d been talking about moving the site for months, but it kept getting pushed to the bottom of the list.
Finally, we decided okay, if we wait to do this perfectly, then we’ll never get it done. So Scott and I asked, what’s our MVP — what is the bare minimum we need to make sure works on the site?
We determined the most valuable things for the community were the job board and the content (posts and pages). We were looking for substance over style. I knew we might have some broken media due to the way we’d need to do the export and import using xml files, but I asked myself: would the community get more value from me organizing a European conference or auditing 400+blog posts to check for broken images?
I figured the community would rather have the conference. (Foreshadowing: notice how I mentioned broken image files? Not, say, audio files?)
I saved all the exported xml files from our previous host as backups. Once we moved everything over, tested that the job board worked, and kicked the tires on a few other things, we cancelled hosting with the prior company.
ACT II: Oh no, oh no, oh no!
For six months, all was well. We got occasional pings about broken images in old blog posts. Sometimes the image URL needed a little editing in the backend because the image got imported twice and had a -1 appended to it. I’d edit the URL and everything was fine. Occasionally I wouldn’t find an image file, and when I reimported, it still was not there. That caused me some head-scratching moments, but then I’d move on. Spending hours on a broken image from 2014 was not a priority compared to coordinating current resources for the community.
Then, in early August, Scott got a ping in the SD Slack that one of the podcasts episodes wasn’t working. In fact, several of them weren’t working. The old ones.
Support Driven’s origin story.
Scott sent me the message. The recent Recap podcast episodes were fine. I looked at the podcast episodes in question, the ones from 2013-2015, clicked a bunch of little triangle play buttons. They didn’t do anything. I tried not to think about how these were from the very beginning of Support Driven. I kept my cool. Equipped with the site’s backup xml files, I attempted to import the media again.
I got “Media failed to import” messages.
I reached out to support at WordPress.com to see if our xml files could be cleaned up and then magically work (I vaguely remembered this being a thing when I was in support there), and I learned that the xml files were referencing the former host. We contacted the former host to see if they could reactivate the site so we could download the audio files. They could not.
None of us had backups — real backups of the actual audio files — on our computers or in the cloud. The podcast where Support Driven began was gone. I posted a sad message in Basecamp.
ACT III: Getting clear on what’s actually missing
At the time I posted in Basecamp, I hadn’t done a full audit. I was panicked and had only randomly tried different episodes and none of them worked. I thought all 24 of the original Support Driven podcast episodes were gone. I was forlorn and not thinking clearly.
Then @jatorres, who helps SD with plugin updates and other technical stuff when he has a minute, reached out to me on Slack to show me a screenshot of some of the audio files in the media library that I thought were gone. The audio players for the 2013-2015 podcast episodes were broken on the front end because the posts were calling the wrong URLs. But after a careful audit, we realized we still had 28 of 36 podcast audio files.
All 12 of the 2018 Recap podcast episodes with @abby and @dpotter are intact, and 16 of the original Support Driven podcast episodes are also intact. These 28 episodes are available and are now working properly in podcast players and on the site in the Podcast category feed.
We are missing the following 8 episodes from the original Support Driven podcast:
Ep. 12 Kathleen Rouse On Collecting and Sharing Stories Found!
Ep. 15 Christa Collins On Growing From 1 to 180 Part 1of 2 Found!
Ep. 16 Christa Collins On Growing From 1 to 180 Part 2 of 2
Ep. 18 Mat Patterson On Building a Remote Team Found!
Ep. 21: Mikael Blaisdell On Where Customer Service Has Been and Where It’s Going Found!
Ep. 22: Kimberly Bringas On Negotiating On the Job
Ep. 23: Daniel Mooney On Working With Product
Ep. 24: Jim Mackenzie’s Career in Support
I still don’t know why some media files made it over and others did not.
ACT IV: Lesson learned
What gutted me most about losing these files is not just that they are Support Driven’s origin story, but that they are irreplaceable collaborations with community members. No matter how many times I’ve heard “make sure you have a backup,” it didn’t become real to me until I realized that you not only have to have a backup, you have to have the right kind of backup.
So my lesson?
Protect the irreplaceable! With real backups! In multiple places!
This shouldn’t be news. It isn’t news. And yet, here we are. I wish I would have gone for perfect 😭. Where I missed the boat on this wasn’t just with the backups, but with thinking through in more detail what exactly “content” includes. I had completely forgotten that the website is the hub for the podcast: if the podcast is broken on the site, it’s broken everywhere.
To avoid something like this in the future, it’s imperative to do a careful review before taking down or migrating a site:
- Determine what is irreplaceable.
- Download the actual assets that are irreplaceable.
I’ve downloaded the 28 audio files we do have. The mp3s are on my hard drive with backups in a Google drive.
ACT V: Two Pinpricks Of Hope
I got an email last week from an old friend at Automattic. He said, I saw your support request — there may be something we can check for you. I don’t want to get my hopes up (my hopes are up), but there’s a tiny chance the files still exist. Looking for the files will require taking the site down temporarily. Since we are in the final weeks before Leadership Summit, we’re going to hold off on taking that step so we don’t take down the Summit page while people are looking at it for event details.
Another teensy, tiny hope is that maybe someone from the community downloaded one or some of the missing podcast episodes somewhere along the way and will reach out to me on Slack and say, “I’ve got one of the files!” If this is you, please ping Scott in the SD Slack @scott.
Epilogue: Done may be better then perfect, but Good is better than good enough
I’m sorry to the community for losing these, and to the interviewees whose episodes are missing. I will update this post if we are able to recover any of the missing files.