What features does Prometheus based monitoring lack wrt. to APM tools?

I think many people here use Prometheus/Grafana for infra and application monitoring esp. in Kubernetes now. It is almost the default OSS solution.

What features you find that Prometheus/Grafana lacks compared to full-fledged APM like DataDog - which makes you consider trying a paid APM?

Some features which I can think of are reporting - like generating PDF reports to send it to team or management. Another is a visualization of overall infra - in terms of server maps and microservices in them, and graph of calls between them.

What are other such features you find lacking in Prometheus/Grafana compared to paid solutions?