Feb 16 2011

Believe Everything You Analyze? Think Again (Google Analytics)

There are many complaints that have cropped up regarding Google Analytics, which can be expected considering that it is the most widely used application of its kind. Being on top isn’t going to come without a few shots, although it must be noted that the capabilities of a free program will always be limited in some respect.

Instead of taking each figure at face value, below are reasons why you should look closely at what you see from Google Analytics.

1. Conversion rates and metrics are provided in percentages, and not real values. While having access to representative percentage figures is useful, it cannot match what can be garnered from the actual hits and hard data. Additionally, the Google Analytics conversion rate includes bounces, which can eschew data and make it harder to know if a pitch is working.

2. The figure provided for the number of visits to your site can be misleading. Instead of accounting for the number of people that actually came to the site and looked around, it is the number of people who just simply stopped by.

3. The exit rate includes bounces as well. Knowing the exit rate, like the conversion rate, can be tremendously helpful in reviewing the effectiveness of your strategies to attract visitors and keep them engaged. However, by including bounces in the exit rate it muddles the data of those entering and those staying.

4. The tracking capability is limited in some areas. It limits tracking in the areas of keywords, search engine hits, ad groups, and campaigns. Also, tracking the visits of browsers that don’t run JavaScript is out of the question, which misses the many visits from people on mobile phones.

5. Its Site Overlay report is nearly inoperable. The purpose of the Site Overlay report is to better assist you in understanding how visitors are reacting to the navigation and content offerings of your site. However, it unfortunately includes stats created by duplicate links, which will put results way off target.

6. Ad filtering programs can block the tracking code of Google Analytics. If enforced, by such extensions like Ad Block and No Script from FireFox, this means some traffic data goes uncollected.

7. Google Analytics cookies are blocked or deleted. If cookies are not enabled, then Google Analytics will not collect data from that user, making their visit non-existent. Thus, site owners are at the mercy of the user’s demand of privacy and protection.

8. Margins of error in sampling can be big in small data segments because of limits instituted to random samples in reports.

CONCLUSION

Looks like lack of supportive online customer service is not the only issue with Google Analytics.

The whole concept of analytics is to investigate quality data to formulate informative decisions and a reliable game plan to meet goals. Inaccurate data can cause major unnecessary mistakes, which is why it’s important to double check data from multiple analytics sources. What inaccuracies have you found in your data, and what did you do to resolve them?

4 Responses to “Believe Everything You Analyze? Think Again (Google Analytics)”

  1. Leslie Strickland says:
    February 17, 2011 at 8:32 pm

    With my Google Analytics I’ve found that paid search traffic does have quite a discrepancy. Usually when I pull this information, I manually talk the visits from my paid search engines out of the total search engine visits number. I have also found that bounces must be included in the average time spent on site meaning my averages are nearly cut in half compared to another source. There I just had to adjust my expectations for the average – and prepare the client for the difference in the metric.

    I agree that there are plenty of areas in which Google Analytics could improve their accuracy. However, there are plenty of issues with other analytics applications. Omniture for instance has the same issues with javascript tracking and cookie tracking as GA, no? And on top of that Omniture:

    1- Lacks the simplicity of set up of GA
    2- Does not bring meaningful metrics together
    3- No longer gives an average time on site – just a distribution.
    4- Takes much longer to download information
    5- Has discrepancies between its 1rst time visitor vs total visitor data

    I’ll stop there. I see why you need to point out Google’s issue. It is essential that analysts are aware of exactly how metrics are pulled and of where discrepancies may lie. However, I’ve found that, for smaller sites (0-60,000 visit sites, no ecommerce), Google Analytics is much more intuitive, presents the data so it is more readily available for use and doesn’t cost exorbitant amounts of money. I’ll take GA for now.

  2. Mike Belasco says:
    February 18, 2011 at 2:24 pm

    Most of these things are true, but I don’t look at these as issues or bugs, it is just the way the program is built. Any analytics platform has to make decisions about what to count as a bounce, how to calculate conversion rates etc. In terms of the JS, your log based analytics platforms has plenty of issues as well.

    I think the moral of the story is to really understand whatever analytics platform you use.

  3. Rebecca L. says:
    February 18, 2011 at 2:38 pm

    Some of the issues you mention are easy to get around with Advanced Segments. Don’t want bounces to be included in your conversion or exit rates? Build a segment that requires a page depth of 2 or greater.

    For the most part, I’m looking at GA not for precise numbers but for general trends. Install five different analytics tracking programs on the same site and they’ll record five different sets of numbers, but the trends should all be about the same and the numbers should all be in the same ballpark. There’s always going to be a margin of error no matter what you do.

  4. Justin Freid says:
    February 20, 2011 at 4:49 am

    It is important to remember that Google Analytics is there to help you analyze trends and understand how people get to your site and what they do when they get there.