Google AnalyticsFrom Bad To Much Much WorseJune 18, 2015
When we last wrote about Google Analytics we noted that the web analytics package only captured a small percentage of your actual traffic, the metrics for top pages, referrers, entry pages, exit pages and pretty much everything was completely wrong.
It’s been five years and the requests to install Google Analytics on some of our sites in spite of our warnings still come. Typically we just add the line of code to the site and go about our business. Recently we decided to check again and see if Google Analytics has become any more accurate. Maybe Google Analytics has gotten better. Maybe they have fixed the flaws in their methodology, maybe we missed something. We looked at the performance of Google Analytics over two different time windows, one week and one month. We used the same methodology as in our previous article. Here is what we found: One Week 1. Site Traffic Over the course of a week of data collection google analytics missed 95.98% of the example website’s total traffic 2. Top Page Listing The top 20 pages Google listed were wrong. In fact Google’s top 20 pages covered the actual top 30 pages of the actual data for the site. 3. Top Page Order In addition to having the wrong pages in Google’s top 20, the sequence of their pages is incorrect even discounting the pages that are missing from the actual top 20. 4. Top Page Count As noted in item 1 above, the page views for each page are inaccurate. Worse yet, the inaccuracies range from missing 84% of the actual page views to just 7% of the actual page views on one page. The random and wild variance in Google Analytics inaccuracy makes it difficult to discount the Google errors, i.e. you can’t just look at a result and know that it is off by 10% or 40% and adjust accordingly, it could literally be off by any percentage you could imagine. 5. Missing Pages The top 20 pages Google lists misses 10 of the actual top pages. In other words in order to cover what Google thinks are the top 20 pages you have to look at the actual top 30 pages. One Month After looking at one week of data we thought we had confirmed our results from five years ago, but when we looked at one full month of Google Analytics data, what we found surprised us. 1. Google Analytics Missed Most, Let’s Call It Virtually All Of The Data Like last time Google Analytics missed a range of the data for the page views on the site. Over a month of data we discovered that Google Analytics missed 89% of the actual page views. 2. Top Page Listing As with the one week sample, Google Analytics top 20 pages spans more than the top 20 pages and over a month of time actually covers the top 98 pages. 3. Top Page Order Consistent with the one week sample, the order of Google’s top 20 pages is incorrect and wildly erratic. 4. Top Page Count As with the one week sample, page view counts for Google’s top 20 pages are incorrect and strangely inaccurate. Incorrect page counts ranged from 63% to -150% (* more on this later) 5. Missing Pages Google’s top 20 pages fails to list 7 of the actual top 20 pages on the site. Over a month of data Google’s error widens to require looking at the actual top 98 pages in order to cover what Google thinks are the top 20. * 6. Adding Page Views That Don’t Exist We found something over the course of a month of Google data that shocked us. We know that Google Analytics does not provide a complete or accurate view of your web traffic, but what we had never seen before was Google Analytics actually adds page views that were not there. One example page Google Analytics claimed 250% as many page views as actually took place. Conclusion The tools you choose to use to answer organizational questions are entirely up to you, but it’s important to have accurate data when answering those questions. In the age of the Cloud and free software it’s easy to understand why so many never stop to ask the question, does this tool actually work? Free and now, trumps accurate and professional, even when professional is incredibly affordable. Quoting from our previous article: Google Analytics is free and you get what you pay for. It certainly provides some views that we haven’t seen in other web statistics tools, but it is extremely difficult to draw conclusions about your web traffic and user behavior when the data you are seeing is both more and less than the actual data, completely at random. Finally, the percentages and discrepancies you might observe will vary based on how you have your web analysis tools configured. The precise percentages presented here are not the key point, but rather that Google Analytics is only capturing a small percentage of your actual web traffic in some cases, and adding page views that never existed in others. We have looked at data across multiple clients for 5 years and have confirmed the errors and trends presented here. If you would like to see the data we used to prepare this article, shoot us an email or a tweet, we’d be happy to share it with you. | ||
Recent News
Cascade Centers 2019 ReDesignFall 2019
University of Chicago Consortium on School ResearchSummer 2019
Federal Laboratory Consortium
Sal Albanese for NYC Mayor 2013February 2013
See More
CHICAGO 346 N. Justine Chicago IL 60607-1010
NEW YORK 564 72nd Street Brooklyn NY 11209
WASHINGTON DC 6031-F Curtier Drive Alexandria VA 22310 |