On this page
Quick answer
RubixKube's Analytics page turns reliability work into numbers a leadership team can read: the hours of investigation Rubix did instead of your engineers, mean time to understand, coverage, noise removal, speed and accuracy. Pick a period and environments, and download a four-page PDF report to share.
Reliability work is hard to show. Nobody notices the outage that took ten minutes instead of two hours. Analytics makes that visible, with definitions you can defend in a review.
What the report shows
Analytics
| Number | What it means |
|---|---|
| Time saved | Hours of investigation Rubix did, and roughly what that is worth |
| Mean time to understand | Median time from an incident being confirmed to the report that found the cause |
| Coverage | How many incidents got an investigation |
| Noise removal | How many alerts did not need their own investigation, thanks to grouping and triage |
| Speed | How much faster than an hour of manual investigation |
| Accuracy | How often a cause was found with good confidence |
Numbers you can defend
- Time saved is a stated estimate. It counts one engineer-hour per investigated incident at $50 an hour, and the page says so. Use your own rate in the room if it is higher.
- Each investigation counts once, per incident or per group. Failed runs never count. Inconclusive ones count as coverage, not accuracy.
- Accuracy measures that a cause was found, with confidence of 0.7 or more or a determined or likely verdict. Pair it with your team's Right, Partly right and Wrong ratings to show how often it was correct.
- Mean time to understand is the metric most teams never measure: the time to know what broke and why, before anyone starts fixing. Read what MTTU is.
Charts behind the numbers
New incidents per day, the open backlog, how serious they were, what keeps going wrong and which monitors fire most, where incidents happen by environment, what is new and what keeps coming back.
Frequently asked questions
How is time saved calculated?
One engineer-hour per investigated incident, at $50 an hour. It is a fixed estimate, shown as one on the page.
What is mean time to understand?
The median time from an incident being confirmed to the report that found its cause. It measures how long it takes to know what broke and why.
Does accuracy mean the RCAs were right?
It means a cause was found with good confidence. Your team's ratings show how often it was right.
Can I share the report with people who do not use RubixKube?
Yes. Download it as a four-page PDF.