incubator-ooo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Rob Weir <robw...@apache.org>
Subject Re: Some stats and observations on OpenOffice upgrades
Date Wed, 26 Sep 2012 14:34:20 GMT
On Wed, Sep 26, 2012 at 8:44 AM, Ariel Constenla-Haile
<arielch@apache.org> wrote:
> On Tue, Sep 25, 2012 at 03:06:12PM -0400, Rob Weir wrote:
>> I've been looking at the upgrade numbers, the downloads that are
>> triggered from upgrade notifications in the OpenOffice client.
>> Although we are not tracking how many times such notifications pop up
>> in the OpenOffice client we do know from Google Analytics how many
>> users click the link to get more information on the update, and how
>> many of these users actually download the upgrade.
>>
>> The trends have been pretty steady, a slight peak when a release is
>> initially made, but a lingering steady state of upgrade requests even
>> several weeks later.
>>
>> For example, let's look at the status for a single day, last
>> Wednesday, Sept. 19th.
>>
>> On that date we had 164,752 total downloads of AOO.  Of those
>> downloads, it looks like 54% of them come from upgrading users.  The
>> remainder are either from new users, or existing users that went to
>> the website directly rather than from an upgrade notification.  (No
>> easy way of distinguishing these two).
>>
>> The interesting thing is the breakdown by OpenOffice client version.
>>
>> For the upgrade installs on Sept 19th we see:
>>
>> 31% of upgrades were from AOO 3.4.0
>>
>> 52% of upgrades were from OOo 3.3.0
>>
>> 15% of upgrades were from OOo 3.2.1
>>
>> 3% of upgrades were from OOo 3.2.0
>>
>> Note the OOo 3.3.0 numbers.  Nearly 4 months after AOO 3.4 was
>> released we are still getting large numbers of OOo 3.3.0 users
>> receiving and responding to upgrade notifications, nearly 20,000/day.
>>
>> I'm not sure how to explain this.  Upgrade notifications should
>> surface once a week.
>>
>> Maybe:
>
>
> With free software, many people often wait for micro releases, which
> usually are bug fixes releases.
>
>


If there were a large pre-existing number of OOo 3.3.0 users, and they
only slowly upgraded, then I'd expect to see regular OOo 3.3 -> AOO
3.4 upgrades, but that this would be slowly declining over time.  This
is because the number of upgrades would be proportionate to the number
of OOo 3.3. users:


upgradesPerDay = NumOOoUsers * constantUpgradeRate

And over time NumOOoUsers would slowly diminish.

But we're not seeing that.  Except for the spike when AOO 3.4 first
came out, the rate of OOo 3.3 upgrades has remained nearly constant.

To me this suggests either:

1) We're in a steady state configuration, where there is a new source
of OOo 3.3.0 users coming into the system, 40,000 or so per day.  This
could be explained if sites like openoffice.fm are attracting that
many users and offering them old versions of OpenOffice.  If we get
160K/day downloads in general, it is possible that sites that
advertise on Google, Bing, Twitter, etc., would get at least 40,000.
It is plausible that this could be a source of ongoing installs of old
versions of OpenOffice.

or

2) The number of OOo 3.3 users is so large, and constantUpgradeRate is
so slow, that the reduction in the pool of existing OOo 3.3 users is
not obvious in the charts yet.  Legacy estimates where that there were
100 million OOo 3.3 users.  We're getting close to 20 million for AOO
3.4.  So that would account for only 20%.  It looks like LO has less
than that.  So that would leave the majority of OOo 3.3.0 users
unaccounted for, potentially still on OOo 3.3.

In any case, when we do a user survey it might be good to put in some
questions targeting OOo 3.3.0 users who have not upgraded.  I've heard
suggestions in this thread that Base quality might be an issue.
Another might be language support, since OOo 3.3.0 supported many more
languages than AOO 3.4.1 does.

-Rob


> Regards
> --
> Ariel Constenla-Haile
> La Plata, Argentina

Mime
View raw message