Inaccurate Traffic Data

Is Google analytics broken?

[gn_wendy]

I am seeing some weird numbers in Google analytics. That is to say, our top keywords are showing ~50% less traffic, and visits are down by about 30%

However the server stat tools on the other hand show the same traffic as the previous week... and the rankings are the same and even slightly improved for several keywords.

Sometimes, Google analytics will update and add ~10% to pageviews and about 4% to visits after 16-18hrs, but I have never seen numbers like this.

Has anyone else had or seen this issue?

[Receptional]

I don't trust the server stat tools - I don't think even Google pretends the numbers are anything buy estimates.

Are you in a Market that would be heavily affected by Thanksgiving?

[gn_wendy]

Thanks for the reply!

No. The website has traffic worldwide, with the USA accounting for tops 15% of traffic.

The analytics data has been updated now, and the numbers are somewhat normal again - that is to say they correspond to the stats (within reason).

I never really analyze data from the previous day, I always give it 24hrs ... but the sheer volume of the variance yesterday surprised me.

[shallow]

I don't trust the server stat tools

Do you mean stats from something like AWStats? The data for my site is quite different from Google Analytics stats.

Which is more accurate?

[makemetop]

Server log analytics and page tracking code based systems like GA are very different beasts and can record very different sets of information. As such, it is rather like comparing apples and oranges. They are both fruit - but the origin of the taste is very different! In my book, javascript based analytics is both more flexible and more accurate.

But then, I am a bit biased!

[craig1972]

To say that "JS based tools are more accurate" is a ridiculous statement.

Get your server logs. Then use a tool to answer the questions you're looking for. Apache (or any web server today) logs every hit someone makes to your site, and if you set it up intelligently -- to reduce all the "noise" of GIF/JPG/CSS files etc being counted as hits -- then use a log analyzer such as Mach5 Analyzer (google for it) to get insights.

Google Analytics is wild and unpredictable in our case. Our web audience is very savvy. Most of them use AdBlock or something similar and block google ads altogether, or have Javascript disabled anyway.

A log is very accurate. But you need the right tool to analyze the logs. AWStats etc are free and have their limitations.

[makemetop]

To say that "JS based tools are more accurate" is a ridiculous statement.

As a blanket statement, I completely agree. But I was trying to point out it was "horses for courses" and completely dependant on what you were trying to analyse.

This can be as simple as defining the phrase "traffic". Server logs absolutely will record all hits, but will the user be able to refine the data properly to get the information they want without having to be be very web-savvy? They may find it much easier to do so via a js based analytics package - and so their data will be more "accurate" for them because of the additional "flexibility".