We use it at my company. I'm not intimately familiar with all the features, but here's what I can tell you:

-It's easy to set up and manage different on-call schedules

-It's easy for someone to manage their own alert settings (e.g. I can add a bunch of different ways to be contacted, like "text me, then call me after 5 minutes if I haven't ACK'd")

-We've integrated it with our internal metrics collections systems so it's really easy to say, "page someone when ${complex_condition} happens"

-IIRC it's not particularly expensive and totally better than the overhead of setting up and managing your own alerting systems.

Exactly. Its a human intervention/alerting/escalation broker. No ops team (or team who does ops) should be without it.

Pagerduty puts alot of focus on reliability. They have redundant telephony providers and are hosted in multiple cloud provider data centers. You don't want your alerting service to go down with your service.

When AWS goes down, Twilio, hipchat, and many other services can go down with it.

Others have already mentioned most of what I like, but I'm also a fan of the muting. If an alert has already fired, additional instances of the same problem don't result in a new alert. No more opening your mail client to find 5000 emails with the same stupid "An error happened" subject line.

Integration was dead simple as well. I couldn't use the official client (because reasons) and it was only a few lines of Python to fire alerts from TIBCO Hawk.

Curious - was it to alert on events from BusinessWorks apps or EMS servers?

Escalation, shift scheduling.