OrangeSoda Internet Marketing Blog

If this is your first time visiting our internet marketing blog, you may want to subscribe to my RSS feed. Thanks for visiting!

Like SEO, web analytics is not as clearcut as many would like. Unfortunately, there is no standard that all web analytic companies follow. And here’s the even worse news: without a “standard” there can never be reconciliation of web analytic data.

Why not?

Because every program defines a visitor, a bounce, and a click, differently. So when it comes to web analytics this holds true: “The trend is king when analyzing web analytics data.”

There are general guidelines that may help make sure you’re comparing data from two different site tracking information. Here are some important questions to ask.

1. Are both analytics tracking codes implemented correctly on every page of the sites?
2. Are the reporting date ranges the same?
3. Is the same page of the website being reported?
4. Is the bounce rate an isolated metric? Are there also massive differences in the number of visitors to the site, or other statistics?

And the final kicker…

How does the web analytics program define a bounce?

What is the difference in the definition of a bounce for each program? One program might define a bounce as someone leaving after 5 seconds, another 10 seconds, etc. This could result in a big difference. And there are many other variables that can create large discrepancies, in fact, some programs allow the web developer to define their own parameters of a “bounce.” Because of these reasons and the fact that this information is usually propriety, it is not possible to reconcile data from different analytic programs.

If we look at the trend then the differences don’t have to be a problem, but can be complementary. However, those that like to deal with absolutes, this is a hard pill to swallow. It certainly doesn’t make our clients happy and some threaten to cancel unless we can reconcile web stats. It’s not something we’re going to take on but we can guide our clients through to try to make sense of the information.

The bottom line?

SEO and PPC is an art form that are always chasing a moving target.

This post was written by Clint Eagar who previously worked for web analytics company Omniture. I asked him to write it after some complaints about numbers on our tracking not matching the client’s tracking system.

This is a common question - how do you reconcile the differences in web stats. It seems like every analytics program comes up with a different number, which can be frustrating for businesses of all sizes. It’s also a problem with paid search because the numbers you see from your PPC analytics may not match up with the numbers in Google.

OrangeSoda recently hired Clint Eagar who worked for web analytics firm Omniture. I asked him to write a post to try to demystify the discrepancies that are common between different web analytics tools.

I get a lot of questions about why there is a variance between how different web analytics packages report traffic results.

It’s All About Cookies
This has to do with how an analytics vendor uniquely identifies a visitor. Most analytics providers uniquely identify a visitor by a persistent browser cookie. When a visitor comes to a website the analytics code checks to see if the cookie exists. If the cookie does not exist it attempts to place it.

If it cannot place the cookie many analytics providers will ignore the entire visit. A large portion of the discrepancy between analytics providers comes into play when a web site cannot place this cookie. Some vendors will build a unique visitor cookie by combining user-agent and IP address. Some analytics vendors use third party cookies to uniquely identify visitors while others set a first party cookie and some visitors have their browsers configured to not accept third party cookies.

Establish Analytic Metric Definitions
The next thing you need to understand is how each analytics vendor defines a page views, visits and other metrics. One vendor may define a visit as a user session that lasts for at least one minute. Others will count an additional visit if the visitor views a page and then leaves the page idle for more than thirty minutes.

So, for example, say you’re reading a news story at CNN.com and get about half way through the article then you head out to lunch for thirty minutes and then come back to your open browser, finish the article and then click to read a new article. This will count as two visits – not one. Some analytics vendors will count this as only one visit. How does your provider track a visit?

How is a unique visitor defined? Is it a daily unique visitor (a visitor that is unique to the site today)? Is it weekly (a visitor that is unique to the site this week)? Is it monthly, etc? I think you get my point.

Tracking Code Execution
Other obstacles to having perfect harmony between analytics vendors could be loading time of site and the location of the tracking code JavaScript, does it load before page content or after. Did the visitor close the browser or click back button before the JavaScript had time to execute?

Web Analytics Is About Trends
Trend is king when analyzing web analytics data. More important than squabbling over a ten percent difference in how Google Analytics or Omniture reports a visitor you should instead be questioning: How many visits to do I have this week compared to last? How are different referring domains driving conversions over time?

Ultimately the differences between analytics vendors is just noise and you should never (did I say never?) attempt reconciliation.