OnlineOrNot Diaries 14

I'm writing this from a train between Florence and Rome, in Italy - I had a bit of time to kill, and wanted to reflect on the last two weeks.

In case you missed it, last diary I wrote about how I wasted a fortnight yet again trying to use a distributed monolith to monitor my users websites instead of serverless. It turned out ensuring that one customer's checks don't influence another's is a difficult problem to solve, and I'd rather pay the ~100% premium to not need to worry about that on AWS Lambda.

Features people actually want

So I started this fortnight eager to ship things my customers actually asked for. Someone on twitter suggested adding on-call integrations to OnlineOrNot. I had been meaning to do so since March 2021, a whole 26 days after I released the first public version of OnlineOrNot, and have had several requests to add on-call integrations (albeit as "nice to haves" than "blockers").

The reason I never got around to it is really dumb: the docs for the incumbents were god awful, and I couldn't bring myself to figure out the mapping between "services", "teams", and OnlineOrNot's checks.

It took seeing new players in the on-call market (shout-out to spike.sh) with really clear docs before I realized it wasn't actually that difficult - I could start with a webhook URL to POST messages from OnlineOrNot to the on-call service.

So over two mornings I integrated four on-call services into OnlineOrNot: PagerDuty, Opsgenie, Grafana OnCall, and Spike (and wrote docs on my side for how to configure each integration).

Landing pages, again

I added even more information to my main landing page this week.

We'll see if this is a good idea, but for the longest time my main selling point was "modern uptime monitoring" (what does that even mean?!) - I eventually changed that to something people actually care about (ease of use), and gave my conversion rate a noticeable boost. Now that I have on-call integrations, I realized the main selling point is alerts, and the business processes that go into action when an alert is fired.

So now OnlineOrNot's landing page follows a sort of chronology by business processes: alerting your team -> status pages for keeping customers updated

What feels extremely natural to follow those features is a sort of incident management system, but I'm drawing the line at monitoring and status pages - it already feels like a sufficiently large feature set to maintain and iterate on as a solo-founder.

Giving away free tools

This week I also challenged myself to build another free tool in under 60 minutes.

I originally launched OnlineOrNot as a public webpage that could check your webpage from around the world: you enter a URL, I run the check, and tell you what OnlineOrNot saw in us-east-1, us-west-1, eu-central-1, ap-southeast-2, and ap-northeast-1.

Having built Do I need a CDN? a few weeks back, and having a decent number of folks signed up for OnlineOrNot off that page, I wanted to bring the original tool back.

I present you: Website Down Checker.

The name is terrible, but it's visually a lot more appealing than the original OnlineOrNot tool!