Analytics Pro Tip: When These Trends Change – Check Your Analytics Tagging With A HTTP Tool

Author: Gab Goldenberg

This is a guest post by Stacey Armstrong, Web Analytics consultant

Stacey’s Bio: I have seven years experience as a web analytics professional, providing analytics solutions for Microsoft Office, MSN Entertainment, T-Mobile, HomeAway, and Les Schwab. My goal as a web analyst is to identify the story within the data.

When I notice a dramatic change in analytics data I first take a look at the analytics tagging to make sure that the change wasn’t caused by a tracking issue.

I use a HTTP tool to look at the analytics tagging passing with the page load or link click. Fiddler and Charles are two free HTTP tools; HTTP Watch has a free version and paid version.

What I’m looking for with the HTTP tool is that the analytics image gif that passes the analytics variables to the analytics tool is present on the page or link click. There should only be one analytics image gif on page load and another analytics image gif on link clicks (if they are tracked). Different analytics variables are passed for page load and link clicks. Don’t worry about knowing what each variable means, for my example I’m just looking to see if the analytics image gif is present.

In part I, I walk through four common data issues that alert me to an issue with the tagging. In part II, I walk through the basic steps to start using a http tool. I’m using Fiddler for my example. The HTTP tools are similar to each other in how they display information.

Part 1: My top four data issues that can be caused by incorrect or missing analytics tagging:

1) Page views to a page spike compared to previous time periods, e.g. page views double.

Check to see if there are multiple analytics gifs passing on page load, as there should only be one analytics image gif passing on page load.

Often video(s) on the page can cause an issue with analytics tracking and multiple analytics gifs would then be passing with page load or with user interaction with the video.

2) Page views to a page are dramatically lower or zero compared to previous time periods.

Check to see if your analytics gif is passing on the page load.

If it is missing, you’ll see page views drop to zero when the analytics gif was removed from the page. To solve this issue you’ll need to add the analytics code back to the page.

3) Bounce rate is extremely low as in ~ 1% – 2%. *

Check to see if there are multiple analytics gifs passing on page load. If there are multiple analytics gifs passing you’ll need to determine which one should be passing and
remove the other analytics gif. Also, you should check to see if page views to the page have spiked (issue #2), since multiple analytics gifs sometimes inflate page views to the
page. Keep in mind that this doesn’t always happen.

For example, on one web page I found that multiple analytics gifs didn’t inflate page views to the page, but did inflate overall site page views and content group (site section)
page views. In this case, the variables in the second image gif did not cause double counting of the page views to the page.

*(Bounce rate is defined as the single page view visits divided by entry page visits, meaning that the visitor came to you page as their first page in their session and then left without visiting other pages.)

Bounce rate as a measure of success depends on the design of the page and implementation of the web analytics tool. A low bounce rate, under 35%, is considered good.

There are, however, pages or sites that will naturally have higher bounce rates. For example, if your site is one page or has many off-site links that lead the visitor away from the page bounce rates will be higher due to the design.

Readers of blogs also tend to read one page and then leave the blog, so blogs will have higher bounce rates. Whether or not you report on bounce rate as a key performance
indicator (KPI), checking it periodically will help you understand what the average bounce rate should be for your site or page. If you see a change, then you can check to see if there is a tagging issue.

4) Conversions, e.g. clicks to purchase, are low or zero.

Check that the analytics gif that passes on the link click is present and is passing the
correct variables that record a conversion event.

Extra tips: Best practices

5) Log your analytic tracking issues

Keep detailed notes of your analytics tracking issues, the date that they started, and how you solved them.

This can be documented in a bug tracking tool or an Excel spreadsheet. This will save you time when you diagnose future issues and help jog your memory when you explain issues to new members of the team.

6) Test changes to your analytics tagging before the page goes live.

This gives you time to troubleshoot issues and verify that you’re receiving data in your analytics tool. Nobody likes to hear that the first days of a new marketing campaign’s data was lost because there was a tagging issue. Since testing data doesn’t reflect customer behavior, it’s best to have a separate profile for testing so that testing data isn’t included with your regular analytics data.

Part 2: How to use a HTTP tool

I’m using Fiddler for my HTTP tool. The HTTP tool shows you all the items that pass when a page on your site is loaded or a link clicked. You’ll be looking for the analytics image gif that passes the tracking variables to your analytics tool.

1. After installing the HTTP tool open your web page in a browser.

If you already have the web page open, then refresh it while the HTTP tool is running. (It’s a good idea to look at a page with a HTTP in more than one browser (e.g. IE and Firefox)).

2. You will see the analytics image gif (.gif file extension) as well as all the items passed on page load.

3. On the left pane, look for an image gif with your analytics tool name (Omniture, Webtrends, Google Analytics, etc.).

Click on the row in the left pane and highlight it.

4. On the right pane, in the HTTP tool Fiddler, click on the “Inspectors>web forms” tab.

The “web forms” tab will show you all the analytics variables that pass to your analytics tool.

5. One analytics gif will pass on page load with a set of page load variables and one analytics gif will pass on link click (if links are tracked) with a different set of analytics variables.

The analytics variables will be specific to your implementation and analytics tool.

6. If you’re not familiar with your analytics implementation or variables, then your goal is to simply see if the analytics gif is present.

You can copy the analytics code and send that to someone, or better yet ask them to stop by and you can walk them through what you’re seeing in your HTTP tool.

Contact your analytics guru, implementation consultant, or a developer and they can check into the details.

Once you’re familiar with your HTTP tool, you’ll be able to troubleshoot issues quickly. I’ve looked at analytics code in a meeting and been able to show the group an issue that I saw. That made it much faster to fix the issue.

Caveat Emptor: Keep in mind that that the HTTP tool helps you see if analytics code is present and is passing, but it can’t verify that the analytics tool (e.g. Webtrends, Omniture, etc.) received the data.

Analytics tracking is like a radio signal, the analytics code on the page is the radio transmission, your analytics tool is the radio receiving the signal, and you are the listener hearing the song. You’ll have to check that the analytics reports are receiving data to make sure that the whole system is working.

Sidebar Story

Leave a Reply