IntroCave revenue was down ~20% compared to February. A big part of that is organic traffic -- my main keywords ("intro maker" and its derivatives) are getting more competitive in high value countries and I'm starting to get pushed to the second page.
I'm super heads-down on getting my new render architecture finished, and that had a few bad side effects in March:
I wasn't paying as much attention to the old render servers, which means they were running a little slower than usual. A render node will occasionally get stuck and need to be rebooted. If 2 or 3 get stuck, renders can start to back up.
For the new renderer, I've been tinkering with the idea of dropping screenshots in the customization form and instead seeking to the specific frame of video. That would mean WAY less asset wrangling. To do frame seeking, the server needs to return some extra headers on video files so the html5 video tag can do partial loads and skip to specific frames. There's probably a way to do that in Apache, but I just put it into a PHP controller. That seemed like I a good optimization, so I rolled it out to every video embed. I serve a LOT of video requests, and it turns out routing those through Laravel instead of straight to Apache was too much load. It's a solvable problem for another day -- for now I just reverted it back.
Although revenue was down, it actually felt like a pretty good month. I made a LOT of progress on the new render servers and the end of that project finally feels within reach (which could still mean months at my part time pace). I've been doing a lot of benchmarks while those machines aren't doing production traffic yet, and it looks like I'll be able speed up render times by around 40%.
Faster renders (and higher conversion rates) are just a bonus, though!
My main goals with the new system are stability (less devops babysitting) and simplicity (git-based, easier to deploy content). With stable and scalable render servers in place, I can move on to the next two big things on my roadmap: subscription plans and deciding whether or not to rebrand to Intro Maker (and migrate to the currently-still-parked https://intromaker.com).