While not difficult to build a basic version, at-scale incident alert & escalation systems are a mix of all the things that engineers hate building for (timezones, calendars, telephones, human-injected overrides, contact management, and 3rd party notification / acknowledgement tracking) all wrapped into a system that's being watched and scrutinized by everyone from compliance, legal and eng all the way to the CEO when they're at their most emotional during and after an incident.

IME for most companies it only takes a homegrown system failing once during a major outage for the budget to open up.

Scheduling, Groups, Escalation. Yea you could probably do something with IFTTT but I'd rather just pay a little bit and not deal with the hastle.

Their moat is the same as Jira's. Despite being expensive, and easily replaced by a bevy of alternatives, management are familiar with the name and purchase it.

It's annoying to build. Their moat is they did the annoying work of finding the team that's willing to put up with building it.

You need to have your own infrastructure monitored by completely different infrastructure, otherwise they will fail at the same time. Instead of having your AWS shop maintain a separate Rackspace-coloed team, etc. it is much simpler to outsource

I'm guessing you aren't a software engineer on a team that shares oncall responsibility, in an org with other engineering teams and dozens of mircoservices.

for a single person, sort of. for a team, with multiple people, multiple on-call schedules, schedule overrides, etc, its so much nicer.

It's a little more than IFTTT, but not much.