<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
</head>
<body>
<div class="moz-cite-prefix">On 1/8/2020 6:16 AM, Tanstaafl wrote:<br>
</div>
<blockquote type="cite"
cite="mid:2d9db074-4d70-1eb1-74a6-e1fd5016b45d@libertytrek.org">
<pre class="moz-quote-pre" wrap="">On Tue Jan 07 2020 11:57:10 GMT-0500 (Eastern Standard Time), Ryan Sipes
<a class="moz-txt-link-rfc2396E" href="mailto:ryan@thunderbird.net"><ryan@thunderbird.net></a> wrote:
</pre>
<blockquote type="cite">
<pre class="moz-quote-pre" wrap="">You would be surprised Axel, 30% sounds exactly right.
Check this out: <a class="moz-txt-link-freetext" href="https://stats.thunderbird.net/">https://stats.thunderbird.net/</a>
</pre>
</blockquote>
<pre class="moz-quote-pre" wrap="">
Change the date range on that to 1-6-18 to 1-6-20 so it shows a full two
years, and it shows the exact same dip last January...
</pre>
</blockquote>
You can also click "All" and you will see the full set of time we
have data for(Aug 1 2018 and forward). It defaults to 1y because
that's the most useful cyclical period we have. Month-to-month and
week-to-week have a lot of variation. There are 3 yearly, temporary
slumps: Christmas, Easter, and August which is European Vacation
Month.<br>
<br>
Our user base is heavily business oriented so any time that people
are not working, we see dips. This is also why older versions cycle
up in % on the % adi graph during the week, and down on weekends.
Business users are slightly more likely to be using older versions,
and they are a lower proportion of users on weekends.<br>
<br>
P.S. You can click version numbers in the legend at the bottom of
the graph on the % page to make lower % items easier to see.<br>
<br>
In the future, I suggest looking at global stats(or at least <a
moz-do-not-send="true"
href="https://addons.thunderbird.net/en-US/thunderbird/addon/lightning/statistics/">Lightning</a>,
which accounts for ~85% of users) to compare to before panicking.
It's always important to remember that any major shift over a short
time period is far more likely to be a data artifact than a real
effect.<br>
</body>
</html>