IntroMaker May Recap
$5216 in gross profit (+11%)
- MRR: $1161->$1137
- Subscribers: 49->48
- Trials: 21->25
TLDR of below:
- I had a huge low quality traffic spike
- I added render queue priority based on country / LTV
- I now reject low CVR countries if the queue is full
- Render servers are much more stable now
Avg traffic went from ~3k/day to ~7k/day, but really it tripled half way through the month.
Lots of direct searches, apparently went viral or ranked for the first time in Indonesia somehow.
Preview render demand shot from ~2k/day to ~10k/day and killed sales for a few days.
Why does more demand kill sales?
Indonesian sales round to 0, and CVR goes down the longer a video preview takes to render.
(Not picking on Indonesia, I've had this problem before with India and other SE Asian countries. 80% of revenue comes from US/UK/CA/AUS/NZ)
At ~2k renders/day, queue time was pretty low. I usually have a free render node ready to pick up a new renderer.
At ~10k renders/day, every render node is fully utilized 24/7 and each new render has to wait a long time in the queue.
I probably max out closer to 3-4k.
My render servers are still manually deployed/updated/managed and cost ~$100/mo apiece.
Option 1: add 5 more servers ($500/mo) and make operations more annoying.
(Obvious) Option 2: add render queue priority (something that's been on my TODO list for years anyway).
I've resisted paying for geocoding services in the past when all I want is a country code, so I did some searching and found https://api.country.is
Free, fast, awesome! Love it.
I set up a render priority with some rough values and a bonus for being signed in.
If there are any renders waiting for a renderer already, submissions from [ID, IN, VN, TH] skip the render preview and go straight to the checkout page.
In a normal month I'll do 50-60k renders. I deleted ~20k while all this was blowing up.
In May I rendered ~80k and rejected another 55k.
The peak has come down a bit and I've settled down closer to 4-5k submissions and back to ~2k/day.
Render time overall is still too long.
People in the 40/50 buckets (unknown countries, most of Europe) have the worst queue time right now at ~5-7 minutes.
The queue is usually full when they submit and US traffic jumps in front of them.
Ironically, the lowest priority group have fantastic queue metrics, as those renders only make it to the queue if there's a free renderer to pick them up straight away.
The new renderer I'm testing seems faster/more stable so far, but still has a few kinks/features to work out before I deploy it.
That combined with the priority queue/rejection system should make it much easier to keep things online going forward!
I went for a 3-day weekend over Memorial day and didn't pull out my laptop at all.
My goal is to have all this fully deployed/automated before heading to Japan/Taiwan for 3 weeks at the end of this month.
Original Thread
I've been working on moving random posts from Twitter + Indie Hackers. This one comes from Twitter!
Intro Maker May Stats
— SimianLogic (@SimianLogic) June 3, 2023
$5216 in gross profit (+11%)
MRR: $1161->$1137
Subscribers: 49->48
Trials: 21->25