Understanding Web Analysis

Well, okay, that's a bit harsh, but I prefer to set expectations low and then pleasantly surprise you. In fact, web analytics can be really, really cool. But you have to understand what its limitations are and how to use it.

Let me give an example: a good web analytics tool is like a powerful microscope. I'm not talking about one of those cheap microscopes you used in elementary school to look at thread. It's like one of those neat scientific microscopes you see in labs on CSI. Now, as neat as an expensive microscope is, it won't do you any good if you're trying to watch a comet. For that matter, it won't even show you what's across the room.

Web analytics tools are limited. They are consciously designed to measure only certain types of data. This data can be really useful if you know what you're doing. If you don't, it can be dangerous. Perhaps as dangerous as a powerful microscope in the hands of an evil mad scientist.

What it is

What exactly does a web analytics tool measure? To put it simply, it gives you an estimate of what happened on a website.

"An estimate?" you ask. An estimate, I say.

With few exceptions, web analytics tools only capture a sampling of the activity on a website. Generally, the sample is quite high, like 95%, but it is only a sample. A survey, if you will.

"Why only a sample?" you ask. You sure are nosy, I say.

The Web is fluid and anonymous by nature. Web analytics tools try to go against that nature, and with surprising success when you consider that the Web is trying to hide people. To date, analytics tools rely primarily on some combination of log files, cookies and JavaScript. These can be manipulated or masked in one form or another by the visitor. Sometimes certain browsers or mobile devices will handle them in an unexpected way. Sometimes the way the tool calculates an element of a visit won't reflect reality--or at best it will be counterintuitive. Not a single web analytics metric is 100% accurate.

In short, some visits are missed, some are misrepresented, some data is flat wrong. And there often isn't a way to discern which pieces are wrong or missing.

What it isn't

I hope you can see now why web analytics tools cannot be used as accounting tools. That is, unless you enjoy the adventure of never knowing exactly how far off your reported revenue is from reality. I guess I can see how that might be fun.

Web analytics reports also are not a census of your visitors and their demographics. There is very little demographic information that can be gathered without the visitor explicitly giving it to a website.

Finally, web analytics is not a crystal ball. It doesn't read minds. It can't tell you why a visitor did something. It can only tell you what they did.

Purpose

Now that your confidence in web analytics has been blasted, we should review what it can realistically be used for.

Keeping in mind that web analytics tools spot trends and correlations rather than exact figures, they are excellent at helping us to evaluate the impact of both the site itself and the marketing that brings visitors to the site. If you can identify the visitor actions that indicate a successful visit and then tie that to their original traffic source, you can go a long way in evaluating what works and what doesn't.

Web analytics is good at evaluating the roadblocks to conversion. Which pages on the site have a high abandonment rate? Which seem to be most influential? Which marketing sources bring only confused visitors? Which pages send all your visitors to the contact page to write nasty emails?

Now that you've been armed with a healthy dose of cynicism, you can finally begin to use the tools the way they were intended: to give you a big picture of what is happening on your site.