LogicMonitor's Documentation and support a top-tier. We had occasion to use them to solve some edge-case situations, and they were willing to understand what we wanted to do and work through their back-end channels to get answers that the documentation couldn't provide. The API is complete, we could not find anything we needed to do but could not.
Detailed performance data was available in reasonable accuracy ( I don't recall exactly what it was, but we found it adequate). It was easy and impressively fast to drill into any of the data available.
Interaction with the graphs, grouped by Dashboards, is highly customizable. There is enough automation to make the task tolerable and produce nice results.
The datapoint collection is super flexible and I especially appreciated the ability to write scripts to collect data. It can take a little manipulation to get into the format LM wants, but there are good examples and goo documentation to guide you through the process. We were able to track and create alerts that fired when there were problems (very few false-positives); I think that's a testimony to the flexibility. I was able to write, from scratch, accurate kafka monitoring by integrating a metrics collection tool for kafka called burrow with LogicMonitor with a high degree of success in detecting issues early (such as producer/consumer deviations).
It integrates well with services such as PageDuty, which we used for some aspects of our product. Other services used the adequate internal alerting system.
Overall, the product was found to be accessible by Operations and Engineering alike; and while every tool that pages someone at 3am gets cast a little shade, LogicMonitor was a great solution for us. Review collected by and hosted on G2.com.