TrueNet measures Latency from all ~450 panelists every hour of every day.  Thats a lot of data and our monthly report cannot do justice to the information imbedded in the data.  I would like to find a way to better use the data for public consumption by those most affected by Latency.

This is a request for comment on presentation more than data, but the data is real September results.

To gather Latency data we ping the target a number of times and collect the minimum, maximum and average response time of the pings for a test, usually every hour of every day on all connections we measure. 

Current targets include:

 

• Wellington
• Auckland
• Sydney
• Melbourne
• Perth
• Dallas
• UK (Portsmouth)

To analyse I take the median of the tests for a single panelist's connection for tests with common times, e.g. the median of 1am is the median of all tests at 1am for the entire month, usually 30-31 tests.
The medians are then averaged for all probes fitting the criteria we are showing in any plot, e.g. Urban, Auckland, ISP=Spark etc

The attached charts are the result of that analysis, the Min and Max show the spread of results, but the spread is low, I have an idea why, but I have not seen the result of so many latency results together so I thought it a good idea to ask others for opinions before we publish on our website:

I have included the UK pivot table to demonstrate the analysis for those who understand pivot tables