Third party analytics client based still reliable?

Time to go back to log files are primary source of analytics?

[graeme_p]

This thread: [webmasterworld.com...] about ad blocking, incidentally brings up an issue with the reliability of third party analytics.

A lot of "ad blockers" are really privacy blockers (like Ghostery or Privacy Badger), that block anything that tracks users, whether it is an ad network or analytics. If it really means you can fail to count half your visitors, what is the solution? Go back to log analysis as the main way to track visitors?

[lucy24]

Ad blocking isn't the only issue. Another one that has gained a lot of attention in recent years is fake visits, where robots just request the analytics file and never even set foot on your site. It's a more sophisticated version of referer spam. So you've got both false negatives and false positives.

It's a judgement call. (File under: Top Ten Most Useless Answers.) Nothing beats raw logs for identifying exactly who visited and what files they requested. But nothing beats analytics for tracking individual behavior. I have to say that if you're going to use analytics, there are several advantages to using something that lives on your own site, or at least the same server.

[tunadmin]

You can use something like piwik. It can be integrated to your site via JS or you can try to parse log files with piwik like awstats/webilizer stye.

[tangor]

I dump my raw logs into a database and run custom reporting. This is reinventing the wheel that other packages can do, but it is 100% mine and I don't have to worry about it and can run multi-year reports as needed..

[graeme_p]

I did not know piwik does log analysis. Very nice.

I did not know about the fake visits problem. What is the motive for that?

@tangor, I am thinking of doing something similar in one case, but it is not the solution every site. Usually I want something simple for a customer.

[lucy24]

What is the motive for that?

Referer spam is only visible to people who study raw logs (or, I guess, crude analog-type stats). Doing it via analytics lets you reach people who read analytics reports instead. And, as a bonus, if the site uses third-party analytics you can bypass the site's own access controls. (Does GA's server ever block anyone, under any circumstances?)

[keyplyr]

But nothing beats analytics for tracking individual behavior.

Depends on how this "analytics" is done. IMO most (if not all) analytic reports fall short of anything I see as definitive. They seem little more than entertainment to me (I've used most all the popular ones including PIWIK & GA)

Yes I do run raw logs through a customized copy of Analog, but most of my conclusions come from manual look-see techniques & using several scripts to extract specifics. Then I use my brain (which probably needs an update.)

[lucy24]

I don't look at piwik very often, but when I do, I head straight for the Outlinks area. I like to see where, if anywhere, people go after visiting my site. And if they've followed a link, I consider that a gain, because even though they're leaving-- as everyone must, eventually-- they're going where I sent them. Heh, heh.

[keyplyr]

And that JS outlink info, while helpful, may be missing a significant number of your site visitor's actual behavior due to modern browser config.