Documentation contents

Provider scoring

Updated

What each number on a listing is, where it comes from, and what it does not mean.

A provider listing carries several numbers. They come from different places and mean different things, and the quickest way to misread a listing is to treat them as one score with several views.

The numbers

FigureSourceWhat it is
Overall ratingReviewsThe mean of published parent reviews. Absent, not zero, when there are none.
Performance, Support, Value, ReliabilityReviewsThe four things reviewers score separately. See the pages in this section.
UptimeMonitoringMeasured reachability over a stated window, from a stated number of places.
Average responseMonitoringMedian round-trip time across measured targets.
IncidentsMonitoringPeriods the consensus recorded as degraded or out.
Review countReviewsParent reviews. Replies are not counted.

What a score is not

Not a ranking position. Listing order depends on what a reader asked for.

Not purchasable. Nothing on this page can be changed by a commercial arrangement.

Not a prediction. A provider's history is a record of what happened, not a forecast. A flawless ninety days is not a promise about the next ninety.

Reading a listing honestly

Three questions worth asking of any figure here:

  1. How much is it made of? Four reviews and four hundred are different kinds of number, and the count is always shown.
  2. What was measured, and from where? Uptime carries its window, its country count and its worker count for exactly this reason.
  3. Which network? Monitoring figures name the autonomous system they describe. A reseller and its upstream are different networks behind one brand.

The individual scores

Each of the four review dimensions has its own page in this section, describing what reviewers are actually being asked and what the resulting number can and cannot support.

Was this page helpful?

Your answer tells us which pages need rewriting.