eValid: FT-150 Top Page Download Time Survey

How does WebSite quality track with company size and reputation?
How well do big companies WebSites compare? Can we learn something
by studying the technical properties of these big company's

In an earlier in-depth study we detailed some partial answers to
this question; see:

But what about a web surfer's very, very initial experience? What
you might call the "speed of first encounter" experience. What
happens when someone goes to the very topmost page of a WebSite?
For just these pages alone, how do the FT-150 sites compare?

Experimental Setup

To find out we used eValid's record/play capability to record total
page download times.

The playback script downloads the top page of the each of the 150+
sites with eValid set up to run with an empty cache, exactly the way
you would download the page the very first time you navigate to that
page. The full page timings are reported in the Performance Log.

We ran the test script on fast DSL connections. From our office
here in San Francisco, we ran the test 10 times and averaged the
data. We ran the same test from Eastern Canada, from France, and
from South Africa, all with similar speed web connections. each
page, To minimize the effect of web latency all the test downloads
were run in quick succession. As a result, the performance data
differences we observed are due to primarily to variations between
the particular WebSite's server capabilities and not on web latency
or "last mile" factors.


Here is a sampling of the results we found. Some were expected;
some were amazing.

o The achieved download rate varied by nearly 16:1. Because the
rates never exceeded 40% of available DSL capacity, and the
tests were run so quickly on after the other, we know that the
wide variation we observed is due primarily to server

o The top page size on the 150+ WebSites varied from 17 KB to over
575 KB, a 32:1 ratio. Amazingly, some of the larger pages
actually were downloaded and rendered quicker because the
servers were faster so much faster.

o The average top page took ~5.9 seconds to download. This is
well over the "the 3-second click-away rule" -- the guideline
that is generally accepted as a worst-case response criteria.
Amazingly, some pages took as long as 65 seconds.

The conclusion is that the biggest companies often don't pay as much
attention as you'd think to maximizing the "speed of first
encounter" -- probably to their detriment. We think they could do a
lot better.

You can see the complete results -- including the full set of download
timing data -- at:

Please let us hear from you if you have any questions or comments!

eValid Division
Software Research, Inc.
1663 Mission Street, Suite 400
San Francisco, CA 94103 USA

Phone: +1 (415) 861-2800
FAX: +1 (415) 861-9801
Email: info@e-valid.com