There is a distinction I keep drawing in these reviews between training a vendor writes to sell a product and training a vendor writes because it needed the material itself. PagerDuty's incident response documentation is emphatically the second kind. It started as the internal runbook the company used to train its own engineers, it was open sourced, and it sits at response.pagerduty.com with no login, no email capture and no product pitch. That provenance shows in every paragraph.
It covers what an incident commander actually does, which is to coordinate rather than to fix. It covers the difference between the commander, the scribe, the subject matter experts and the communications lead, and why one person trying to do all four is how a thirty minute incident becomes a three hour one. It covers severity definitions and why arguing about severity during an incident is a symptom of not having agreed them beforehand. It covers how to run a postmortem that produces action items somebody owns, rather than a document that gets filed.
It covers the specific mechanics of handing over an ongoing incident at a shift change, which is a genuinely hard problem that almost nobody trains for. This is the accumulated operational knowledge of a company whose entire business is other people's worst days, and it is free. The Incident Management Framework training built on top of it is deliberately platform agnostic, and PagerDuty deserves real credit for that choice. You can take the course, learn the framework, and go implement it with Opsgenie or with a Slack channel and a shared document.
A vendor that teaches a discipline rather than a product is doing something unusual and should be rewarded for it. The certification story is pleasantly loose. PagerDuty has repeatedly run free certification events, including ones tied to charitable causes, and the exams turn up free often enough that paying for one seems unnecessary. The Certified Incident Responder credential tests the framework rather than the button positions, which is the right emphasis.
Now the criticisms, and they are real. The product-specific courses, the PD101 and PD201 material on configuring services, escalation policies and event rules, are noticeably more ordinary. Competent, unmemorable, the sort of thing you work through because your employer asked. The gulf in quality between the framework material and the configuration material is wide enough to be jarring.
My bigger complaint is alert fatigue. This is the central pathology of the entire on-call industry. Teams instrument everything, everything alerts, engineers are woken four times a week for things that did not need a human, and within a year they have learned to swipe away pages without reading them, at which point the alerting system is actively harmful because it provides false assurance. Every experienced SRE knows this.
The training acknowledges it and offers sensible-sounding advice about tuning thresholds and using alert grouping. What it does not do is confront the uncomfortable version: that alert volume is usually a symptom of unreliable systems and unclear ownership, that noisy alerts persist because deleting an alert feels risky and nobody gets promoted for it, and that a paging tool makes it trivially easy to add alerts and socially difficult to remove them. A company that charges per user for a paging platform is not the natural author of the lesson that you should page humans much less often. The same applies to on-call sustainability.
There is content on schedules and fair rotations, and it is decent as far as it goes, but the genuinely hard questions, about whether a team of four can sustainably hold a twenty four seven rota at all, and about what compensation for out of hours work should look like, sit outside what the material will address. Four point two. The open sourced response documentation alone would earn a good rating. The product training drags it down slightly, and the honest limits of vendor-authored material show exactly where you would expect.