Why Is Google Analytics Inaccurate?

by 44 01/31/2013

So you noticed, eh?

Google Analytics is inaccurate.

There’s a reason for that, actually there are several.

Here’s a fairly exhaustive list that I have divided into two categories:

  • Stuff you can fix
  • Stuff you have no control over

Ready to find out why your Google Analytics is inaccurate?  Here we go…

Google Analytics Errors You Can Fix

Let’s start with the things you can do something about.

Here are five:

1 – Failure to tag pages

Google Analytics works by putting a JavaScript “tag” into the code of your pages.

No tag on a page  = no data for that page

If you only have a 5 page website, you can easily check if the script is present manually.  If you have 100′s or 1000′s of pages, it’s difficult if not impossible to check for the Google Analytics tracking code on all your pages.

But there are tools you can use to verify that your pages are tagged and find the ones that aren’t.  Rachel Gerson from SEER Interactive details the use of Screaming Frog SEO spider tool to search every page for the presence of the same Google Analytics code in this article.

2 – Poor location of Google Analytics tracking code

Website owners are concerned about page load speed on their websites, and they should be.

As a result many web developers place code like the Google Analytics tracking code just above the </body> tag so that the visual elements of the website can load before the code executes (presumably increasing page load speed.)

However, placing the tracking code at the end of the <body> code will miss visitors that click quickly through pages on your site before the Google Analytics tracking code executes.

Google Analytics tracking code loads asynchronously and therefore doesn’t slow down page loading.  It’s best to place the tracking code at the top of the page before the </head> tag.

JavaScript doesn’t have time to execute = no data

3 – More than one Google Analytics tracking code on the page

I once had a client that had a 3% Bounce Rate on his website.  I couldn’t believe it.

Turns out he had put two Google Analytics tracking codes on his pages.

2 Tracking Scripts on page = messed up data

4 - JavaScript errors on the page

Google Analytics only functions if the JavaScript code on the page executes.  If you have errors in any of your other JavaScript on the page, the GA code won’t execute.

JavaScript Errors = Stopping of the browsers scripting engine = no data

5 – In-page Analytics is Inaccurate

The Google Analytics Overlay Report is inaccurate for a number of reasons.  Here are three big ones,

  • It doesn’t track anything that doesn’t have an HREF element (a click on a form submit button or play button on a video for example)
  • It doesn’t track pages containing Flash
  • It can’t separate the number of clicks from different links that lead to the same location  (UPDATE:  Big thanks to Aaron Bradley from SEO Skeptic for pointing out that this solution will fix this problem in In-Page Analytics by adding some extra lines of code to your tracking script.)

The solution to this is simple — use Crazy Egg for your Overlay Reporting and you’ll also get the Heat Map, Confetti Map, Scroll Map and List View Reports.

Google Analytics Errors You Have No Control Over

Ok, so these are things you can’t do anything about but they will absolutely help you better understand your data.  So, pay attention. :)

Here are 9 ways your Google Analytics data might be inaccurate that you have little or no control over:

1 – Some browsers have JavaScript disabled

Google Analytics tracks information for any visitor that loads the JavaScript tag in the page code.

Browser disabled JavaScript = no data

Yahoo! published data that shows that this problem is minimal.

javascript-disabled

2 – No cookie

Google Analytics doesn’t track a visitor unless it can accept a cookie.  Cookies are important because they are used to tag a visitor and aggregate that visitors behavior over the course of multiple visits.

No cookie = no data

There are a number of different ways cookies can cause problems:

  1. The visitors browser doesn’t accept cookies.
  2. The visitors firewall blocks or deletes cookies
  3. The visitor deletes cookies manually (Most research ballparks cookie deletion at 30% of visitors over a month.)

There are probably a dozen other ways cookies can be removed.  If you have some good ones, put them in the comments.

By the way, this is one of the reasons that search engine spiders don’t register in your Google Analytics — the spider doesn’t accept cookies.

3 – Cookies time out

Google Analytics uses two different types of cookies:

  • A persistent cookie – This cookie is placed on your device the first time you visit and remains for 2 years or until it is deleted, the browser is re-installed or some other such removal occurs.
  • A session cookie – This is a new cookie that is placed each time you visit.  You receive a new session cookie for each separate visit.

This begs the question:  What happens if someone:

  1. Visits a web page
  2. Goes to dinner for two hours leaving the page open
  3. Comes back and starts browsing

I’m glad you asked.

Google Analytics ends a visitor session after 30 minutes of inactivity.  In the scenario described above, a new session cookie would be placed when the visitor starts browsing again.  This will be considered a brand new session by Google Analyticcs.

Browser inactive for > 30 minutes = messed up data 

Actually, in this data hound’s opinion this isn’t messing up the data.  Thirty minutes of inactivity should warrant a new session.  But it is certainly a special circumstance.

4 – Same device, different user

This one is best illustrated by example,

  1. John visits ABC.com for the first time and a cookie is set on his computer.
  2. He then goes for a jog.
  3. Jane then gets on the computer and visits ABC.com for the first time.

These should be counted as two separate visits, but the unique visitor is tied to the cookie and the cookie is tied to the device.

Therefore these will be considered visits from the same “unique” person.

Visitors sharing devices = messed up data

5 – Same user, different device

Once again an example is best to illustrate this,

  1. Jane is standing in line at the grocery store.  S
  2. She starts researching XYZ product on ABC website from her iPhone.
  3. A cookie is set on her iPhone.
  4. She gets home, unloads the groceries and then purchases XYZ product from ABC website on her laptop.

In a perfect world, Jane’s behavior would be tracked as she switches devices.  As it is, the purchase will be attributed to a brand new visit.

Multiple devices = messed up data

And this behavior is more prevalent than you might think.  This is a portion of an infographic we created about the use of mobile in the marketplace,

cross-device-usage

 

6 – Google Analytics doesn’t reprocess information

Let’s say you have been tracking data using Google Analytics for two years.

You then decide to set up a Goal Funnel.

Google Analytics will not retroactively apply your funnel to the data it has.  This is just too much to ask of a free service that is processing the massive amount of data that GA processes every day.

In the scenario above, your Goal Funnel will be visible from the time you set it up forward but no data will be available before that time.

8 – Google hasn’t processed your data yet

Google Analytics isn’t real time, although there is some limited real-time data you can get through the Real Time reports.

For all other data, it’s best to assume that Google Analytics is 24 hours behind.  That being said, most small to average websites are safe to assume that Google Analytics is 6 to 8 hours behind.

9 – Google is sampling your data

If you have a very active website and you try to run a report that includes a large amount of data, Google Analytics might use data sampling.

In other words, they will not use the whole of your data — it’s too resource intensive for them to do so and speed is greatly increased when they provide you with a subset of that data.

So, they will run the report you are requesting using a sample of your data.

Learn more about data sampling here.

What’s the point of measuring if the analytics are wrong?

Let’s put it this way — it’s not ideal to build a house using a tape measure that measures a foot at 10 inches.  But, as long as the error is consistent, you can still build a reasonably solid house.

If you can fix the tape measure, do so.  If you have no control over the error, just learn to make the judgments that you can understanding that the data is flawed.

What you have to determine is whether the error you have is consistent.

For example, error from visitors that delete cookies can be assumed to be consistent from month to month and year to year.  It’s not ideal that it happens but, frankly, there’s nothing you can do about it.

Now that you understand the possible reasons why Google Analytics is inaccurate you will see your data completely differently.

What are you seeing in your Google Analytics that is inaccurate?

About 

Russ Henneberry is the Editorial Director at Digital Marketer. He's worked on digital marketing projects for companies like CrazyEgg, Salesforce.com and Network Solutions. You can connect with Russ on Twitter, LinkedIn, Google+ or on his blog.

Get our Daily Newsletter

Get conversion optimization, design and copywriting articles delivered to your inbox FREE

44 COMMENTS

Mukesh

Most of these issues or problems mentioned here are common to all analytics services (except web-server logs based tools). Also these problems are web issues in general as the same issues can be for other services / tools not just tracking. Are there alternatives that solve all these issues?

January 31, 2013 Reply

    Russ Henneberry

    Totally agree Mukesh, thanks for mentioning this. Any tag based solution is going to have these same problems. You aren’t going to find a solution that doesn’t have error — we just need to learn to live with it and make it as accurate as possible.

    January 31, 2013 Reply

      Henrik Hil

      Webtrends can track by 3. party cookie as well, can track by user agent tool to so you dont need java enabled. Cross domain tracking is by default therefore very easy.
      Next there is a plugin you can use that will transfer your 1. party cookie to another domain using a paramter in the url.

      October 31, 2013 Reply

    Henrik Hil

    @ Nicole, Webtrends or SiteCatalyst do NOT sampel data.

    October 31, 2013 Reply

Aaron Bradley

Thanks for this comprehensive look at inaccuracies in Google Analtyics, Russ!

However, you are in error when you say In-Page Analytics is inaccurate because it “can’t separate the number of clicks from different links that lead to the same location.”

Until recently that was the case. However, late in 2012 Google introduced “Enhanced Link Attribution” to address different links on the same page pointing to the same target URL (http://bit.ly/YCUjHG).

January 31, 2013 Reply

    Russ Henneberry

    Aaron– thanks a ton for pointing this out. I updated the article.

    January 31, 2013 Reply

Hannes

Thanks for the post as it is always good to remember the issues of a tracking tool like GA. I have just a little remark regarding your point that In-Page Analytics can’t separate the number of clicks from different links that lead to the same location. With a little change in the tracking code and the admin settings, GA actually can do that. For more information: http://support.google.com/analytics/bin/answer.py?hl=en&answer=2558867

January 31, 2013 Reply

    Russ Henneberry

    Thanks for adding this Hannes — I updated the article. Thanks again!

    January 31, 2013 Reply

Dan Coleman

This is a great article. All to often people just assume that GA is working. It’s worth noting that the free version of screaming frog only allows you to check 500 URI’s and the paid option is £99 per year.

It’s a great tool but if you just want to validate GA tagging you might be interested in the simple PHP crawler I’ve made specifically for this task. It’s a bit rough around the edges and will take you 10 minutes to set up the first time but it’s free and it does the job. http://www.exceptionalcommerce.com/Validate_Google_Analytics_Tags

January 31, 2013 Reply

Rose

Thank you! Kind of. With your help, I realized that plugin overlap caused my GA code to be in there twice. So, I’ll find my TRUE bounce rate in a few days, I guess. I may need some ice cream. On the bright side, at least it was in the head (both times)!

January 31, 2013 Reply

    Russ Henneberry

    Oh wow! Actually I have seen that twice before — now three times. Glad you caught it! :)

    January 31, 2013 Reply

Jacob Kildebogaard

Hi

Great list. As a danish Google Analytics consultant I needed a tool to check the code, so we have developed GAtective.com for this purpose.

You can free use http://www.gatective.com and scan your site. We check for code on all pages, same UA, same codeversion, and more and provide you with an excel file with all URLs. And all feedback are welcome.

February 1, 2013 Reply

    Asmitha

    Is there a limit to the number of pages that can be scanned with gatective?

    March 5, 2013 Reply

      Russ Henneberry

      I’m not sure, you will need to check with them about that Asmitha.

      March 5, 2013 Reply

      Jacob Kildebogaard

      Asmitha: We (Gatective.com) currently have a limit on 2.000 pages in the free scan, but we can help making much larger scan if needed. Just mail us, and we will find a solution :)

      March 6, 2013 Reply

        Asmitha

        Thanks so much Jacob! I have mailed you :)

        March 6, 2013 Reply

Brad Cumbers

The important thing, and I try to make sure all of my clients are on the same page, is not to be looking at the raw hard numbers, but to look at the trends and % weightings over time. Use these to base your decisions on – not that you had 2 more visitors today than yesterday.

February 4, 2013 Reply

    Russ Henneberry

    Good point Brad. It’s much easier to see trends than to try to tease out any really hard numbers to go by.

    February 5, 2013 Reply

    Henrik Hil

    The issue here is if you are using sampled data. Please notice that Google Analytices themself says: Sampling reports is relying on that asumption that your trafic patern is distributed identical all days and looking on small segments as conversion will be like findind a neel in the grass.
    I really dont understand big customres there use Google Analytics because the want to save $ 200 daily for the licens for Webtrends or SiteCatalyst. If you miss ONE conversion data you will miss an opportunity for optimization. Is that not $ 200 of value?

    October 31, 2013 Reply

Ankit

I also have observed that sometimes Google Analytics doesn’t give the right results. There is always a difference between the visits that are shown by c-pannel tracker and Google Analytics.
I always thought that it was due to javascript issues, but you have given some other very good reasons for why this difference happens.

February 5, 2013 Reply

donOld

Last Dec, goDaddy cancelled Site Analytics, a web server stat tool. Site Analytics reported 800-1,000 page views per day. After switching to Google Analytics my page views have dropped to 80-100. Could 90% of my traffic really be from spiders?

March 5, 2013 Reply

    Russ Henneberry

    If I’m correct the Site Analytics program was reporting activity from your server logs is that correct? If so, don’t forget that Google Analytics is a tag based solution that can only register data for visitors that can execute JavaScript and accept a cookie. So, your server log solution was actually more accurate in this way.

    March 5, 2013 Reply

      donOld

      Thanks Russ. That’s right, Site Analytics was a sever-side app that processed my log files. I am now using GoDaddy’s free analytics program Analog 5.1 (also a server-side log app). In the last 7 days it reports page views to be 6,552. Google Analytics says I have had under 100 hits. All of my main pages are tagged properly at the end of the section as per Google’s instructions. I can’t fathom the discrepency, or how Google Analytics managed to get so popular.

      March 8, 2013 Reply

        Lasse Liukkonen

        I also think that Google Analytics is missing A LOT of traffic to my website. I am considering If I should remove the whole thing from my pages as I trust and find more useful the logs on my cPanel. However, I am a bit worried if the removal of Google Analytics code would affect adversely in Google Search rankings..?

        Google Webmaster on the other hand is great!

        May 9, 2013 Reply

          T

          I’ll 3rd these GA accuracy concerns. I recently implemented GA on a small site as a test case and the data is significantly less than our long-running web server logging tool. Visits alone are off by 700% with GA compared to our server tool. The implementation is quite recent so perhaps it will take GA some time to catch up and stabilize, but if the current data patterns hold I will advise the client to ignore (and will not implement on other sites).

          October 7, 2013

Barbara Weibel

I sure hope you can help me. A while back I hired a designer to revamp my webste. Afterward, Google stopped tracking my traffic, so I researched and via Google support pages, realized the tracking code was not in my header. I added it, and weverything went back to normal. That was a couple of years ago. Today, through a series of events I don’t need to bore you with, I discovered that my site had previously used the traditional synchronous code and it was still in the footer. So for the past two years, at least, every page of my site has had both codes installed. Before I do anything, I hoped you could tell me what the implicatins of having both tracking codes active, other than bounce rate. Is it possible that my traffic has been counted twice all this time? And is it safe to just remove the old tracking code from my footer? Appreciate any help you can provide.

April 8, 2013 Reply

    Kathryn Aragon

    Hi Barbara. Great questions! In most cases, Google tracks unique URLs, so the duplicate code should NOT result in duplicate hits on your page. That being the case, it should be safe to remove the code from your footer.

    April 8, 2013 Reply

Lumia 925

i have some problem too
in alexa, my website growing fast
but in Google anaytics, visitor and page views every time down :(
thnks for your info :)

May 14, 2013 Reply

Md. Nasir

Great info… I have a small query, my blogger page views shows a huge page views but when I check my analytics there are very few traffic, why this happens? Is there any problem with my Analytics account?

Thanks

October 26, 2013 Reply

Nikole

A web advisor placed GA code on several pages of our 1000 page site and now tells me he trusts Google analytics results more than our Website OS. Why would he not realise that visitors enter the site through many pages on the site, not just the index page or the pages he marked;- the site is 15 years old, regularly updated. Google analytics, as he set it up, shows between 2 and 5 visitors per day (!) and website OS shows 200+. Even if some of these are robots, etc. I’m in a state of shock and hate that I have to debate this with him. Isn’t it obvious?

October 27, 2013 Reply

Henrik Hil

TagMan wrote that Google Analytics changed refering domain for direct trafic so up to 50% of your trafic are reportes as Google refered trafic. This discussion started back in 2009 and is now started again.
Other topics that TagMan can be found here: info.tagman.com/rs/tagman/images/TagMan_Truth_About_Google.pdf

October 31, 2013 Reply

donOld

One thing I have learned since my original post here is the difference between “visitors” and “hits”. Hits are simply requests made to your server for the resources needed to load (display) a page in a visitor’s browser. A single web page may need many resource files to load (images, text, buttons, style sheets, etc.). I use Fireworks to assemble my pages. Fireworks breaks up my finished pages into 40 or 50 smaller image files and exports them to my graphics folder. When a visitor requests a page, 40 or 50 requests are made to my server for the image files necessary to load the page. In my server logs, this shows up as 40 or 50 page requests but in reality all this activity was initiated by a single user requesting a single page.

Google Analytics doesn’t track server requests. It only tracks the 1 html page file that contains their javascript tracking code. I now believe this is the main reason why the GA numbers are always much lower than the server-based data programs.

October 31, 2013 Reply

satya

hi , based on my findings GA – not 100% accurate . I am using combination of Aw stats and GA together .. THis helps me to certain extent , but Is there any other tool , which gives more promising data ???

November 7, 2013 Reply

Dan

Just what I was looking for.

We have both GA and Statcounter(visit length set at 1 hour) installed with quite a difference in Unique visitors.

I realise they track differently but have to give the stats to some advertisers.

Think its ok to add the 2 together and then take the average as accrding to some of the forums GA under reports and SC over reports!

Page views for both are actually more consistant, just need to have the answers ready.

Thanks Dan

December 11, 2013 Reply

naga

HI, i am using google analytics for my website, its fast growing, GA Showing per day visitors are 700+ everyday, when i am searching for my site growth value, alexa showing only 27 visitors perday.
Even i gone through my hosting control panel , that figure GA providing is correct. alexa showing wrong figures.

January 4, 2014 Reply

Shannon Lewis

I have noticed this happening… example would be best to illustrate it…

Log into google analytics on Friday, review Thursday’s data. Vistors = 480.
Log into google analytics on Saturday to gain more insight to Thursday’s traffic and now vistors = 240.

This seems to happen all the time. When I review yesterday’s traffic, the data is always higher than when I review that same day a few days later.

Insight as to why this happens?

March 11, 2014 Reply

Siew Ann

Google Analytics seemed to have missed out on 1 transaction that I had. I don’t have many transactions on my site, hence why I noticed. Any idea what happened there or how I can get Google Analytics to show that transaction under E-commerce Conversion numbers? Otherwise, this affects my total overall % . I’m running on Magento for my ecommerce site. Googly Analytics, however, have not failed to track all other transactions that occured thus far. I would very much appreciate any sort of assistance I can get.

March 17, 2014 Reply

Sports Backlinks

I’m still not sure what the problem is. Perhaps everything listed in this article. My analytics report typically shows for 30 days what my server logs show for 1 day. It was so far out of whack that I just stopped using it entiirely.

March 18, 2014 Reply

Geo

Nice post in pinpointing drawbacks that EVERY js/cookie-based tracking tool out there has. If I am to sum it up I would say “Nirvana Fallacy!” (appeal to perfection). I mean, come on, there is NO tool that solves these problems without introducing new ones (in most cases worse than the ones solved).

Let’s hop straight to the “unsolvable” issues:

Those 1-2% of visitors with no javascript or who don’t accept first-party cookies? Who cares? You are measuring *trends*, not trying to see what John Doe does in this exact minute. The first can be used to act against, the second – not so much, unless you really care about this John Doe in which case you are better off hiring a private eye.

Real-time? Come on, this is bs eye-candy, no real >data-driven< decisions come from those reports except for *very* large companies with hundreds of thousands of visitors per day.

Sampling? Look at "trends" above. Sampling is a fine thing with large datasets. It's happening in your brain as you are reading this very comment. You are still alive and doing well, right?

Now, if you suspect you are a victim of any of the solvable issues, I suggest you try our tool "Google Analytics Health Status" for free here. It will detect most issues and give you advice on how to solve them.

If you really want to bash Analytics, you could try "User-level insights" and you'll be nailing it. Well, not that much anymore, since they've finally launched User-ID tracking, but still they lack USER-level conversion calculations and other things marketers really need and care about.

April 11, 2014 Reply


Leave comment

Some HTML allowed

Get conversion optimization, design and copywriting articles FREE >>>