Speed Index Distribution

WebPageTest General Discussion
1

In https://sites.google.com/a/webpagetest.org/docs/using-webpagetest/metrics/speed-index (https://8d4a14bf-a-42404934-s-sites.googlegroups.com/a/webpagetest.org/docs/using-webpagetest/metrics/speed-index/si-cable.png?attachauth=ANoY7coSMaR3F34REW_ZwPzNbGpHuPnh9r5Fhp3DfbdZy6ELHbl283s8S7vfMWk-tf_leHxAVJlbFNaTNlJonibbkd-MNJ1n44PHT9-STj_znc1eHSoelEneeBs_X3f8nT_D_NTO7Ag4VdaqwNI6To9PVa-GSGQsY8skc6VDcMGS9wAhZGjD8iV9fCpJoqsFFJ6huVTBTiQw9wsthyKk76_cF88phucGAas-tyPawIWNoBGR3f3NWooGahFyb964HfuqAYffoTCV&attredirects=0), there is speed index distribution figure.

I get the csv data for pages from httparchive. But I find the figure I plotted out is somewhat quite different to it for May 15,2013,

But in 2014,2015, the data is also different to it. The value varies significantly
(in 2013, the max value is 1800, but in 2015, max value becomes 80000).

Can you explain to me anything I did wrong?

2

1800 sounds suspiciously (and impossibly) low for a max. The 2013 data can’t possibly be right if that is what you are seeing. It might be worthwhile to spot-check a few of the individual tests to see if it was a reporting issue or a measurement problem.

3

Thanks for your reply, I’m a bit confused about it as I used same date data from your document figure.

One more thing is what’s the reasonable range for maximum of speed index, as 2014 and 2015 data are around 80,000, which is 4 times larger than the figure in the document

4

I don’t know that I’d ever use an absolute maximum for anything interesting. A single outlier could easily skew things. The SpeedIndex is effectively reported in milliseconds and is the average point in time when the viewport pixels are displayed.

80,000 Would imply an 80 second max in the data set which is plausible if some site was sufficiently slow (or just timing out/failing to respond.

1.8 seconds (the 2013) number is WAY too fast and I’d be surprised if the average was even close to that.