# which stats are correct?

#### Do you have a question? Post it now! No Registration Necessary.  Now with pictures!

•  Subject
• Author
• Posted on

I have discrepancies in the number of people reported on my website
www.exampleproblems.com.

I paid I-Web-Marketing for 100,000 visitors and they report this many
sent for Apr 17-23:

Apr 17th 2006        497
Apr 18th 2006       580
Apr 19th 2006       529
Apr 20th 2006       535
Apr 21st 2006       506
Apr 22nd 2006       558
Apr 23rd 2006       541

Apr 17th 2006        153
Apr 18th 2006       149
Apr 19th 2006       150
Apr 20th 2006       157
Apr 21st 2006       154
Apr 22nd 2006       117
Apr 23rd 2006       148

Webalizer says I had this many visits:

Apr 17th 2006        1406
Apr 18th 2006       1339
Apr 19th 2006       1385
Apr 20th 2006       1700
Apr 21st 2006       1463
Apr 22nd 2006       1339
Apr 23rd 2006       1901

Who should I believe?

-Thanks

Todd Smith
http://www.exampleproblems.com
Graduate level mathematics examples and solutions!

## Re: which stats are correct?

Doh eh? You live and learn.

It is best to spend money on content rather than traffic. Web Marketing
companies can cause you serious trouble as many of them use decidedly
underhand method to deliver you "traffic".

As for which is right - well, they could all be right as they could all be
measuring very different parameters. Always treat any traffic measuring unit
as a relative measurement, not an absolute measure. Never try to equate any
measurement as "actual live people viewing your page".

The main thing with traffic measurement is that you consitantly use the same
tools, because as with all relative measurements, you are reading trends and
not actual values. Comparing last years webalizer stats to this years
awstats is going to skew even the ability to provide a relative measurment.

## Re: which stats are correct?

On 24 Apr 2006 00:06:30 -0700, elliptic1@gmail.com wrote:

No one. trust no one. believe nothing. Remember, there are lies,
damned lies and statistics.

for pages. But this information will not reveal those requests which
were satisfied by proxies along the way.

Unique visitors is a worthless concept. People do not visit a web
site, a computer and browser combination make requests. This
combination might be at a busy cyber-cafe frequented by numerous human
individuals. Or it may be a domestic computer with a single paranoid

Matt

--
Veritas Vincti
http://www.probertencyclopaedia.com

## Re: which stats are correct?

On 24 Apr 2006 00:06:30 -0700, elliptic1@gmail.com put finger to
keyboard and typed:

[all except one value for each system removed for brevity]

All of them, because they're measuring different things.

To give a bit more detail on that, here's how they count and why they
differ:

1. Webalizer is using your raw log files to count hits and then
extrapolate into visits. The way it extrapolates can be configured,
but basically it's taking any request for an HTML page as a page view
and then further extrapolating to assume that any sequence of views
from the same IP address within a specific timeframe constitutes a
visit. This tends to inflate the number of visits, for three main
reasons:

a) It counts a single visit by an AOL user as multiple visits,
due to the way that AOL's network is configured.
b) It includes visits from spiders and robots, and if you're
getting a lot of spider traffic then that will generate a lot
of visits in the stats.
c) Artificially limiting the timeframe for requests to count as
a visit means that someone who exceeds it counts as two or more
visits.

2. Google Analytics is using Javascript embedded in the footer of your
page. It tracks visits by a variety of means, primarily cookies, but -
as it knows which links people actually click on - it doesn't have any
artificial timeframe constraints. It also excludes robots and spiders,
as they don't trigger the mechanism. Google tends to underestimate
slightly, due to two main factors:

a) Being Javascript and cookie based, it completely ignores any
visitors who have these disabled.
b) As the code is in the footer of your page, it won't count
anything until the page has completely loaded - meaning that
counted. This can be significant if you have a data-heavy site
and lots of dial-up users.

3. I-Web-Marketing are counting any clicks from their site onto yours,
whether or not these clickthroughs have any value to you. It tends to
inflate the numbers sent, for one main reason:

a) You're paying them for visitors, so it's in their interests to
tell you that they've sent you lots of them.

As to which set of stats is the most useful, that depends on what you
want. Those provided by I-Web-Marketing are only useful for knowing
whether they've delivered on what you're paying for - they tell you
nothing at all about actual usage of your site. Webalizer (and other
logfile-based packages) will be the most accurate as regards the
underlying figures, and the most valuable from a sysadmin perspective,
but the results need careful interpretation to be useful from a
marketing point of view. Google Analytics (and other Javascript-based
systems) will be somewhat less accurate than Webalizer, but the
figures produced will be considerably more meaningful as regards
actual human visitor behaviour.

Personally, I use a combination of Google Analytics and Webalizer, and
don't bother with anything else. Between them, these two give me
pretty much all the information I need, and the fact that they don't
agree on the number of visitors is irrelevent - I'm using them for
different purposes.

Mark
--