Stats > Browser Stats
This page presents web browser statistics that may be of interest to website designers.
Caution : stats mislead.
Caching distorts raw data;
audiences vary for each site;
methodologies vary for each survey;
surveys miss or omit important details;
surveys mis-identify browsers or other user agents;
some search spiders pose as browsers;
small sample sizes exaggerate fluctuations;
and stats don’t count those who stay away because their browsers are not supported.
[more...]
Caution : the stats may help decide when a browser is so rare that a site need not support it;
the stats may satisfy the curious; but the stats are truly useful for little else.
“Get your facts first, and then you can distort them as much as you please: facts are stubborn,
but statistics are more pliable” — Mark Twain
~
[more stats quotes...]
The table Usage Stats (see above) lists stats from several sources, showing how much stats can vary:
- Source 1 stats are from EWS. Percentages reported by EWS miscount KHTML- and Gecko-based browsers: but a userAgent list is reported from which I can compute KHTML and Gecko stats. Note also that EWS counts hosts rather than page accesses, which are not strictly comparable.
- Source 2 stats are from Net Applications’ Market Share.
- Source 3 stats are from the Pitt County, North Carolina website.
- Source 4 stats are from a USGS server. Users are mainly from the US, many from government and educational institutions. Note: recent IE5.01 percentages are anomalously very high, suggesting that the IE5 percentage is too high and therefore the other percentages too low.
- Source 5 stats are from sites that use a hit counter.
- Source 6 stats are from this domain. Because of its special audience, its stats apply to a narrow segment of the population.
The best stats for a site are the stats gathered for that particular site: and even these are skewed by caching and faulty browser-detection. For example, consider Kerry Watson’s Browser Statistics page: this page uses three different hit counters whose reports should be comparable; but they are not, in part because of faulty browser detection.
Bottom line: use statistics with extreme caution.