Revenue Breakdown
Revenue was down ~8% month-over-month.
Sales were pretty flat month over month. Ad revenue was down about as much as expected (I removed AdSense on mobile as part of a page speed pass).
Content + Render Updatespe
I bumped my A/B test of the old vs new renderer up to 80%-new-renderer and migrated one of my old servers over to the new setup. I've been slowly migrating old video templates over to the new render pipeline while also adding new content each month, but there were still 16 videos that needed to be rebuilt. I also took stock of all the old templates that had been disabled over the last couple of years (mostly for performance reasons). In a normal month, I'll add around 10-20 new videos, but May turned into kind of a mega-month:
As soon as those videos were ready, I killed the last "old" render server and migrated it to the new system.
The two older machines will eventually be sunset and replaced by a single machine or two with faster CPUs, but they were 50-60% "ready to go" for the new pipeline and this was the easiest way to throw the switch.
I got a large traffic bump towards the end of the month, but most of it was low quality (lots of keywords with "free" -- "free gaming intro" / "free intro maker" / etc). Free-seeking traffic doesn't tend to convert well, but it was a very useful test of the new render infrastructure.
I was averaging around 1900 preview renders per day in April, and that number shot up to ~2900 in May (with a peak day of close to 3900). For the most part, the new machines held up well. I did a lot of performance testing before throwing the switch, but that was at the lower ~2k/day volumes.
End of Month Updates
The end of the month is usually my busiest time. I need to finish getting new videos out, cut together a YouTube reel (posted on https://youtube.com/introcave), write a blog post, write/test a newsletter, and then queue all the various social links. This was doubly stressful last month because the active A/B test meant processing all the videos twice.
The April content update was pretty easy.
Making it easier to release content was the biggest driver for rebuilding the backend, but this was really the first month where I got to feel that.
It felt good!