Over the last few months I’ve noticed something with the Google Analytics stats for my small business forum that call into the question the validity of those referrer stats and are making me question Google Analytics stats in general.
I’m sharing some images of those stats below. I had to reduce their size to fit within the design here, but you can click on any of the images for a full size view.
Small Business Forum Referrer Stats
As the responsible SEO that I am, I periodically check stats for several sites under my control. On the small business forum I’ve notice that this blog seems to send a fair amount of visitors each month. It’s usually the #1 or #2 referrer to the forum.
At first glance I should be happy. This blog sends some pretty good traffic to the forum. There’s a very low bounce rate and visitors are spending a lot of time at the forum and viewing a lot of pages. Looking a little deeper though you notice something strange.
You can see in the image above that of the 136 visitors this blog sent to the forum over the month, 134 of them visited from the main page of the blog and only 2 visited from specific posts.
That’s certainly possible as the main blog page gets the most traffic for the entire blog, but still it seems a little odd that 98.5% of the clicks originated there. The home page of this blog does not get 98.5% of the traffic to the entire blog.
Digging a little further you can see in the image below that 133 out of those 136 visitors are using a Mac. Hmm? The market share for Macs is growing and this site has always had a few more Mac users than usual, but really 133 out of 136? Interestingly enough I happen to use a Mac.
Let’s change the dimension from operating system to city and see what’s there
The plot thickens. 133 of those 136 visitors apparently live in Denver. What are the odds? Wait a minute. I don’t live in Denver but my IP goes out through my ISP in Denver. Seems like all those visitors just might be me.
I can dig into some of the other dimensions, but it’s pretty clear that most (probably 133) visitors from this blog to the forum are me.
Now you might be thinking what’s so unusual about that. I naturally spend time here and also at the forum. The unusual part is I don’t ever click a link to get there. I type the forum address directly. Generally I just type the letter ‘s’ and my browser shows the forum right away. Other than when I tested to make sure I correctly added the links, I’ve never clicked from here to there. And that testing all took place several months before the stats you see above.
Are Referrer Stats Bogus?
At first I thought maybe it was because I would usually load up my browser with this blog in one tab and then open a new tab to visit the forum. I thought perhaps having this site open in one tab caused Google to think the traffic was originating there. Not a pleasant thought if you think about it.
However I’ve made sure the last few months never to have this site open in a tab when visiting the forum. And again I never click any link from here to the forum. Maybe I’m missing something, but doesn’t a referrer visit mean a visit through a clicked link?
So why does Google consistently show what would indicate me clicking links here and visiting my forum when I never click the links here to the forum?
I won’t post more images, but I will tell you that most of the top referrer sites month over month are sites of forum members. Checking some of the dimensions shows the same pattern you see above. It’s pretty clear the overwhelming majority of the visits are by the same person.
While it’s possible some forum members do visit through links on their site, I tend to doubt it. I suspect the situation is similar to what’s happening with my own visits to the forum.
What Do These Bogus Stats Imply?
Am I missing something here? Or is Google recording referrers where none truly exists. A couple questions come to mind right away.
- If these stats aren’t real what other stats that Google is showing also aren’t real?
- How exactly is Google tracking visits if it isn’t through clicks on a link and what does that imply about what they track in general?
Again I’ll ask if I’m missing something here? Could I be doing something with Analytics causing referrers to show like this? I haven’t noticed this on any of the other sites I track. Have you ever seen something like this on any of the sites you track?
Download a free sample from my book, Design Fundamentals.
Try clearing your cookies?
Assuming you visit your forum daily and also that you clicked the link from your blog to the forum within the last 133 days. It could be the tracking cookie has not expired so each day you visit the forum the same cookie gets called and reinitialized matching your IP with past ones and such.
That would be my best guess, but I am not 100% sure the methods Analytics us to track everything.
Thanks Zerek. I had thought about the cookies and have cleared them between clicking and now. Even if that was what was happening it’s really not a referrer visit and Google shouldn’t be showing that. Unless I’m mistaken a referrer means someone has to click on the link. If they click once and on their next visit type the address directly that second visit shouldn’t really count as a referrer. It should be direct traffic.
At least that’s my understanding.
For example, based on what I’m seeing I might conclude it makes sense to advertise here to promote the forum, when in truth this site is only sending a couple of people to the forum each month.
Do you have any other monitoring to match it up against? Have a showdown between the two and see whos more accurate.
I have noticed a few peculiar things with Google Analytics. I think its a good tool, and not the end all of site tracking.
At the moment I only Google Analytics running, but it would be good to run another stats package to make the comparison. I know that no stats package is perfect and they all have their quirks. I did find this an interesting quirk.
Your browser could be pre-fetching pages. Firefox also has some problems with referrer handling if you’re using that.
I am using Firefox so maybe that explains it. I just checked some of the other member sites, though and it looks like they’re using Internet Explorer.
This is still calling to mind a few more questions.
1. If my visits are being incorrectly shown as referrers to the forum wouldn’t the same also be true with other sites I visit frequently?
2. Why doesn’t this seem to happen in reverse. Referrers from the forum back here don’t appear to be me?
3. If this is happening with my visits shouldn’t it be happening with the visits of others?
I certainly don’t think there’s anything evil going on here, but it still has me thinking it’s important to dig a lot deeper into your referrer stats to get a clear picture of what’s going on.
At the very least, you probably should add your IP address to your analytics filter so it stops counting your visits. This won’t address your question (which I also find curious and slightly disturbing) but it will stop your overall numbers from being skewed by your activity.
I used to do that, but my ISP would change my IP enough that I’d have to keep changing it in the filter. After awhile my own visits weren’t statistically significant so I wasn’t concerned.
With this I can easily tell when it’s me visiting. I’m more concerned with how well I can trust the visits from other sources.
I’m a viral marketer online and see the same problem. I get most of my traffic through referrers and direct keywords from SEO. I notice that more traffic comes from these referrers, even though i never though nothing about how accurate they were and trusted the analytics. I do not know how the bot handles my visits… from new tabs, new browsers, a whole new computer, or even checking the page on a 3G network for compatibility issues. I know the numbers aren’t perfect now, but it does help me see the unique visitors and parts of the world who are viewing my pages, even if its off by a couple hundred miles, lol. I hope google starts loosening up and give us more clues of how they manage SEO, keywords, and tracking of sites.
I’m not sure this is a Google issue and I don’t think it’s really about them being evil in anyway. As Dan pointed out above this could be a Firefox issue or something with prefetching pages.
The concern to me is if the stats aren’t accurate for one visitor how much can you trust the stats from other visitors. Digging a little deeper makes it clear what’s happening so maybe the answer is in us customizing Analytics reports.
I’ve used GA for a while and also statcounter for my blogs and the results have always been the same for both of them.
never had problems with GA (google analytics).
I don’t know that this is necessarily a problem. In general I still trust the stats I see. It’s just made me wonder about a few things being off and having me think it’s important to take a deeper look at your analytics.
You mentioned that you usually type in ‘s’ and then select the page. It may be that the browser is sending the same referrer with the request that it had when it created the cache entry of that URL.
It may do this to help servers know that your bookmark/cached entry was originally entered.
Try, and I know this may be annoying, typing in the full URL instead of the one your browser presents you from memory and see if that helps your referral stats.
I am keen to know if this is the case.
Thanks Guy.
I’ve noticed this with other people visiting the site and not just my visits. I know several of the forum members well and it’s pretty obvious Google is showing referral traffic from their sites when I know the sites don’t generate enough traffic to send us the amount reported. It’s clearly the one person visiting the forum daily and I doubt they visit their own site first to click a link there each day.
I’m not concerned at all about how it specifically reports me. It would be easy enough to filter myself out of the stats. It’s more that you can’t always filter out what other people are doing.
For example I could type the full URL, but I can’t control what others do. I can tell in some cases, but not with most.
What I found for me was shortly after writing this post I cleared the cookies in my browser again and that seemed to take care of the issue. Of course as I continued to surf the web, other cookies were stored and I see the same thing on other referrer traffic as well as search traffic.
I’ve been thinking of writing a follow up post on this sharing more of what I’ve seen as well as some additional thoughts.
I’ve always felt that browsers shouldn’t just have a “View Source” option, but some kind of “View Entire Protocol” option, which would show the full HTTP GET data as well as whatever header preceded the page’s actual html.
hey Stephen, or should I call you Vincent Van Gogh (your avatar is his self-portrait), this has been bugging me for some time and made me question my understanding of what a referral is. I finally performed a more thorough search in an effort to find out what is behind Google Analytics reporting a site as a referral when I clearly am not linked anywhere on the site nor the reported page. I still haven’t found the answer, but hopefully I can be of some assistance in your quest. I am also a Firefox user and I have the Google Analytics tracking code on my blog. I notice a false or bogus referral only when I visit a blogspot blog other than my own, and at some point I create a new tab and edit my blogspot settings or content, and then visit my blog to view the effects of the change I made in either the settings or my blog content. This has happened to me 3 times where when I open a blogspot page (other than my own), then use that blogspot page as a reference to edit my blogspot site, then visit my blogspot site, that blogspot blog I originally referred to is counted in Google Analytics Traffic Sources as a referral.
For example, yesterday I wanted to edit my blogspot settings so that for my blog posts that contain computer programming code, the sometimes long code would be completely displayed within the boundaries of the visible column width associated with the specific blogspot template I have chosen. I first visited this page:
http://bguide.blogspot.com/2008/02/howto-add-custom-css-class-to-blogger.html
Then I opened a new tab, and went to change the settings for my blog. Then I viewed my blog to see the effects of that change. Today, I viewed my Google Analytics report for yesterday, and to my consternation, I see that:
bguide.blogspot.com
and in fact, the aforementioned URL, shows up in Google Analytics as a Traffic Source referral, even though I know that page does not link to my site. Like you, this makes me wonder how reliable GA is, as I’ve confirmed the bogus “referral” which we know is false is counted as a visit.
Steve is fine. Around the web I use the username vangogh a lot, which is why I chose to use that avatar here.
You’re noticing the same kind of thing I did. It did happen to me with a tab open and the site in the other tab would show as a referrer when clearly it wasn’t.
It doesn’t bother me so much when a site I visit is being counted as a referrer since it’s easy enough to discount it from the stats. But knowing it’s doing that for you leads you to assume it happens when other people are visiting your site too. It calls into question how reliable any of the stats are.
Since I wrote this post there are plenty of times I see something in analytics I know isn’t true. Reports showing one of my sites is being found for a particular search phrase several hundred times when it’s unlikely anyone used the phrase that often.
And again the referrers. The whole thing really has me rethinking how valuable the stats from Google Analytics are and my guess is the same thing happens with other analytics packages as well.
Some do:
use Firefox, install Web Developer extension, then you can use Information -> View Page Information -> Headers. You will see the Request & Response headers there.
Steven,
it would still be interesting if you could try out what Guy suggested.
If that approach works then this is a bug in the browser.
Of course, you should never trust referrer anyway (because of its nature), but it is interesting to see how it works.
Enjoy!
I haven’t really tried anything to test this since it’s not really my visits that I’m concerned about. It’s easy enough to filter me out od the stats.
It’s more that the majority of people visiting the site aren’t going to be doing any workaround and why should they.
I agree with you about not trusting referrer stats. That was really why I posted this. I knew they weren’t perfect, but I had thought referrer stats more accurate than they apparently are. I no longer trust them the same as I once did.
What Guy said is what I think is happening. I did notice not long after this post that when I cleared the browser cache I stopped seeing the same referrer numbers. It makes sense.
Of course how many people go around clearing their cache between visits?
I sell display advertising for an online news site in Canada. (I have also sold radio, TV, newspaper & magazines)
The frustration in trying to get clients to understand the value of ‘impressions’ vs ‘click-thrus’, can be overwhelming at times. I ran into a situation yesterday, where a client cancelled an annual contract based on ‘referral’ information from a “Google Analytics” report. Unfortunately, this client deals with a (self-proclaimed) local ‘agency’, which would rather cancel according to her wishes, than make any effort to explain to her WHY she should not place so much value on questionable click throughs.
I have a question I’d like to ask you, as I am at a complete LOSS as to how to explain the following:
Our closest competitor, which is the daily newspaper on line, has only 10% the traffic our site has. An identical display as was put on both sites, and the GA report claims that the other site generated 444 referrals with a bounce rate of 75.90%, and avg time spent as ONLY 32 seconds.
In contrast, the report claims our site generated only 21 referrals with a bounce rate of 42.86%, and avg time spent as 3 minutes & 22 seconds.
These stats seem completely illogical to me. I am inclined to believe that the daily newspaper may be quickly clicking on & off their advertisers ads, just to inflate the numbers….and in turn, mislead the advertisers as to the value/performance of their ads.
Am I grasping at straws here? Could there be any other explanation that makes any sense?
Hello Jean.
Instead of tracking visits referred from one site or another, I think your client should track both ad campaigns with the proper campaign tags (utm_campaign, see Analytics’ help for campaign tracking). Then he could tell exactly how many visitors he had from one campaign or another.
Furthermore, I don’t think that what matters is getting clickers to land on the website and then be “gone in 32 seconds”, or at least this should not be the way the customer measures the output/ROI of the two campaigns. Yes, time on page could be one thing. Goals and conversions should be another. It really doesn’t matter if one campaign is generating 1000 visitors and 1 conversion, if the other generates 2 out of 5. Does it?
Not to mention the fact that measuring conversions (sales, subscriptions, download of some material, whatever) instead of landings can filter out happy clickers who just reach the website and do nothing useful.
Good luck,
Calin / Romania
Jean, I agree with Calin. I’d rather have the targeted traffic your site is sending than visitors who quickly leave.
There could be a lot of reasons why the other site generates more clicks, one of which could be they’re clicking the ad themselves. You could study how they’ve incorporated the ad into their design and try to understand how they’re getting more clicks.
However remember that clicks aren’t everything to the advertiser. Ultimately they’re looking to make a sale and if you send one visitor each month that leads to a sale and your competitor sends 1,000 visitors who never buy, the ad is still better placed on your site than your competitor.
This is something that has always concerned me, my wife has a small art business online, and I always wonder how many visit are from her or me; although, we go into the admin section of the site so this should not effect the stats. The cookie theory seems plausible. It would be interesting to know, especially since you said you don’t have this problem with your other sites. Most sites have permalinks, if Google is using this to access your site than maybe this is one of the places the stats are coming from.
Scott, you can actually filter out your own visits fairly easily. Here’s a help page from Google Support on using filters. Look for filtering based on IP address
The one thing you have to watch is if your cable or dsl modem needs to be restarted it’ll sometimes assign you a new IP address, which means you’ll have to update the filter with your new IP.
Steven,
Referrer information is collected and sent from your browser so any and all tracking software will give you the same problem. Each browser has a different way of caching header information and I have not found a solution for resetting this data on a client’s computer and I believe for safety reasons all browsers will not allow websites to alter this data on the client side.
I know this doesn’t really help fix this problem, but at least we can understand why it is happening.
Sean
Thanks Sean,
Yeah i guess the problem isn’t specifically a Google Analytics one. Just where I noticed it. I don’t know that there will be a fix, since we probably can’t control everyone’s browser. Guess it’s one of those things we’ll have to live with.
If you do ever figure out a solution, even a partial one, do let me know.