SEOmoz.org had 13.8mm visits from 6.25mm unique visitors last year (2011). Those numbers are pretty exciting, but what's not exciting is the external perception created by third-parties like Compete, Alexa, Quantcast, Doubleclick and Google Trends for Websites. These sites report massively lower and wrongly trending data - and SEOmoz isn't alone in experiencing this frustration. We're among dozens of sites I've talked to who've gotten emails and comments lamenting our poor growth or crummy year thanks to these horrifically inaccurate services.
Here's a screenshot of our actual traffic from Google Analytics:
Now let's look at the comparison to each of those services:
Above is Alexa's estimate of SEOmoz's web traffic for the past few months. It's hard to tell how accurate they are, because they're not showing any exact numbers, only "percent" of "reach." They do correctly note that traffic was down in December (the last two weeks of the year were very slow for us due to the holidays, which is a good thing - even SEOs deserve a break) :-)
Historically, Alexa showed a much longer timespan and much more inaccurate data, at one point estimating that our traffic had dropped year-over-year since 2009. I've had well respected VC funds reach out and ask why we were struggling and whether we felt the SEO market was drying up because of those charts... Now, Alexa's ranking us as the 472nd most popular site in the world, which is definitely way, way off.
Next up is Compete.com's estimate of SEOmoz's traffic. They're much more specific, but tragically, way off the mark. For a time, I'd hoped Compete would be a much better competitor to Alexa, but those hopes died a few years back. This chart isn't just wrong, it's directionally backward (we grew when they showed us shrinking and shrunk where they show us spiking at year-end) and off by almost two full orders of magnitude (our daily traffic is about 2X what they estimate our monthly traffic to be).
How anyone can trust that data is beyond me, since you can easily compare many sites who publish their traffic details (as we do) against Compete and see this discrepancy. To be fair, I've heard that for the top 1-2,000 most popular sites on the web, they're not bad, though I can't personally confirm this.
Quantcast's estimate of traffic looks equally terrible to Compete. It's directionally wrong and off by multiple orders of magnitude as well. Quantcast's saving grace is their "Quantified" program, which shows actual, truly accurate and measured numbers for sites that opt-in. I wish they'd stick to that model exclusively rather than providing these random guesses on sites they've not included in the program, though. I'm also really struggling to understand how 17,671 unique people could create only 11,005 visits... That's a brain teaser.
Google's my last, best hope, and since they capture such a large percentage of sites' traffic in Google Analytics, I'd expect they have a pretty excellent data-modeling system to work off. Apparently, that belief is mistaken. Google's by no means as bad as Compete or Quantcast (and possibly better than Alexa), but it's still way off. The directional data is sort-of close, but the daily unique visitors count shows at ~200K in December. Our analytics says it's ~47K daily or 722K for that month.
Since Trends and Doubleclick are both under Google's operating umbrella, you might be tempted to think they use the same data... In fact, Doubleclick Ad Planner's estimate of Moz traffic and Google Trends for Websites appear to have at least slightly different numbers (hard to tell for sure based on GG Trends' incomplete graphs). One thing I can tell for sure - neither is accurate, nor even directionally correct.
The over-time charts don't quite match each other (though they're close-ish); it looks like Doubleclick is showing higher traffic to SEOmoz generally than Trends for Websites. The closest data point is their estimated time on site, but I'm not sure I can give them credit for that. If you put on a blindfold and throw enough darts, one of them will probably get close to the board. It's hard not to feel that way about these numbers, too.
Now here's the rub:
Recently, Ani López wrote about Comparing Google Trends for Websites vs. Google Analytics Data and showed a few examples that suggested greater accuracy than what we see with SEOmoz (and OpenSiteExplorer, too FYI). Thus, I'm asking for two favors from you to help get a better sense for the relative usefulness of these tools.
The first is to take the quick survey linked-to below:
please take me! (opens in a new window)
The second is to, if possible, take screenshots of your own analytics vs. Trends/DoubleClick/Compete/Quantcast/Alexa and share them in the comments below. For anyone who puts together a compelling side-by-side, I'll happily include links in this blog post to your site and to the images showing your traffic vs. what these third parties report. Hopefully, that incentive can help spur transparency from those of you willing and able to share some broad site stats.
Thanks as always for your help - looking forward to getting a broader view of these tools' performance. For now, I'd remain highly skeptical, but we might revisit the topic if we get very compelling data in the survey and in the comments (otherwise, I'll just update this post at the end of the week with the survey results, and since they're anonymous, provide full data).
We're even worse off in the UK than you, where we only get the stuntent, semi-functional data and features of SEMrush, etc., often significantly later than the the US data and in much less volume.
Rubbish, eh?
For my sites, SEMRush is actually pretty accurate with respect to it's trendline making. The raw data may be inaccurate but the overall trends it reports on seem to be pretty good.
Curious to know how it matches with SEOmoz. I'm guessing that the screenshot is total traffic, vs SEMrush which uses "SE Traffic".
+1 for SEMrush, it is a little better than the tools listed by Rand. However,the data showed by SEMrush is still lower than the actual data.
I agree SEM rush does it for me in my business
Just compared SEMrush to Omniture and the data doesn't even nearly correlate. I tried this for a couple of our sites.
The competitive intelligence space in the UK is fairly poor unless you're willing to pay £25,000 per year for Hitwise. We recently released www.similarweb.com which has what we believe to be the largest UK panel out there. Hope you'll find it helpful.
I had a client who reported their Alexa numbers to their Board of Directors. He ignored Google Analytics and our repeated pleas to use more accurate data.
Alexa did some kind of algo rejiggering and left things looking like traffic had plunged about 75% one month. He lost his job.
Seriously. I'm surprised we're still around as a species.
I'm glad someone did this research.
Now, we know which tools NOT to use for research.
Karan
It's a pretty serious issue and one I found myself in the middle of just before Christmas.
Market analysts actually cite Google Trends and similar in performance reports to the City. It doesn't take much to spin already questionable data to give the false impression that a company’s popularity is waning online; perhaps even going so far as to extrapolate that out, applying it in a much wider, and much more damaging, context.
And such reports can have a very, very real affect on a company's share price...
Wow. I never use data from Quantcast, DoubleClick, Compete or Trends on traffic in the first place - Google Analytics is too helpful to try branching out - but this is WAY more inaccurate than I would have ever imagined.
Here's our data from Google Analytics (click on the photo to make it bigger): https://www.flickr.com/photos/74022627@N03/6673622005/
Here's Quantcast: https://www.flickr.com/photos/74022627@N03/6673622189/
As you can see, Quantcast has us pegged at 21,697 monthly visits at our highest for the year in July 2011. Thanks, Quantcast, but that was actually our lowest traffic month according to Google Analytics, and the unique visitors were more in the range of 46,000.
The Daily Unique Visitors data was not available from DoubleClick, but I did access the Traffic statistics chart. According to it, our Total Visits are actually lower than our Unique Visitors (estimated cookies)! The traffic overall is grossly underestimated by DoubleClick but the time on site is a minute and a half higher: https://www.flickr.com/photos/74022627@N03/6673622123/
Finally, Compete: https://www.flickr.com/photos/74022627@N03/6673622081/. Our actual traffic is anywhere from 2.5 to 4.5 times higher than what they report. The trend of the graph isn't exactly opposite from ours, but it isn't accurate either. The one thing they seem to recognize correctly is that both April and July were our lowest traffic months of the year - however, the scale is all wrong.
Unfortunately, are data is not available from Alexa since we're ranked >100,000 - and Google Trends does not provide it either. Given what I've seen above, though, I'm totally okay with that!
SOMoz commenting form seriously lacks the formating stuff... I wrote data in such a nice format in above comment and look what happened to it... :(
You can use HTML.
I know that, but still my whole table that I made above is 'DESTROYED' and now I am angry because of it... All I meant to show was that for new sites these factors like MozRank, PA and DA are much like a dead factors which takes 'centuries' in updating... So we need something which is if not real time atleast should be updated on daily basis like Alexa!
For a month or two old(new) sites, there are no good sources to check the performance of site, I mean apart from analytics, the PR, the mozRank, the PA, and DA of SEOMoz, all are udpated so late, what will a new site do? Alexa is updated almost daily, thats why, though its random and rough estimate of Alexa, yet it gives an idea atleast.
Here is the Alexa rank, that I am noting for last few days... its a new site, a month old I guess...and this is how the Alexa rank is changing daily:(along with few other parameters)
4,422,871 ----- Moz Rank: 2.86 | Page Authority: 26 | Domain authority: 13 | PR - 03,261,573 ----- Moz Rank: 2.86 | Page Authority: 26 | Domain authority: 13 | PR - 02,851,462(30,31st jan)-- Moz Rank: 2.86 | Page Authority: 26 | Domain authority: 13 | PR - 02,582,054(1st jan) ----- Moz Rank: 2.86 | Page Authority: 26 | Domain authority: 13 | PR - 02,279,294(2nd jan) Moz Rank: 2.86 | Page Authority: 26 | Domain authority: 13 | PR - 02,064,024 (3rd)----- Moz Rank: 2.86 | Page Authority: 26 | Domain authority: 13 | PR - 01,896,531(4th jan) ----- Moz Rank: 2.86 | Page Authority: 26 | Domain authority: 13 | PR - 01897353 ----- Moz Rank: 2.86 | Page Authority: 26 | Domain authority: 13 | PR - 01555919 ----- Moz Rank: 2.86 | Page Authority: 26 | Domain authority: 13 | PR - 01469011 ----- Moz Rank: 2.86 | Page Authority: 26 | Domain authority: 13 | PR - 01325353 ----- Moz Rank: 2.86 | Page Authority: 26 | Domain authority: 13 | PR - 01282266 ----- Moz Rank: 2.86 | Page Authority: 26 | Domain authority: 13 | PR - 0Alexa Keep changing daily and even twice a day while others are just dead figure... Is there any solution for this? We are talking about tools other than analytics!
Hello again Mr. Fishkin,
I have another question about alexa. Do you think that your position in alexa makes a change in your position in Google search? In the case that alexa produce this changes, affects directly or indirectly?
I make this question because, you know if you install the Alexa tool bar, yourself can change the positions in Alexa... So the alexa factor affects in the google search?
Thank you again for your response!
I am also interested in knowing the answer to this question. Someone can help me too and tell us if the position in alexa influences in google positions?
"... I'm also really struggling to understand how 17,671 unique people could create only 11,005 visits... That's a brain teaser."
Rand, I wrote a post comparing Google Analytics to Omniture's (now Adobe's) SiteCatalyst. I had the same problems with SiteCcatalyst. I have two huge data tables in that post, one showing how SiteCatalyst reported huge amounts of traffic and would report more unique visitors than visits; the other showing how Google showed less traffic but did not contradict its self.
These are two fully installed traffic analytic tracking codes and the paid software coud not even get it right! But the free software hands down was more reliable.
That post garnered over 50 comments (that’s a lot for my little out dated blog), and it was fun to watch Omniture squirm for a bit to improve their product.
So I am all in favor for telling the truth about how bad these companies misrepresent our traffic. I’ll be happy to contribute as much data as I can from my personal sites and from my employer’s.
We dug into a few of our sites and found egregiously inaccurate reporting, most on the scale of 30% - 60% under-reported. Some tools showed the same levels of discrepancy, implying they both use the same (incorrect) data sources. I'd rather not make excuses for these tools (even if the long-term trends are correct) since they are in the business of providing data. A margin of error of 10% - 15% is not unreasonable to ask for. I have posted screenshots and traffic tool analysis on the agency blog.
I feel alexa and google data do not match up properly. Varied stats data.
Well done Rand.
I was actually going to write up something similar as a guest post for GetStat.
I was digging through data in their Codex and I noticed that there was an 135% upswing in rankings for guidetoonlineschools.com (9/24 - 11/22)
Compete shows their traffic as flat and then dropping: https://siteanalytics.compete.com/guidetoonlineschools.com/
Quantcast shows their traffic as dropping: https://www.quantcast.com/guidetoonlineschools.com
SEMRush however does show the upswing: https://www.semrush.com/search.php?q=guidetoonlineschools.com&db=us
I do agree the data is definitely largely imperfect but no more or less imperfect than basing projections (ugh) on the various CTR studies out there or even basing a link profile analysis on a tool that doesn't index as many links as Google does. As long as you consistently use one tool it should at least be accurate within itself and therefore does provide some actionable insight into your competitors.
-Mike
Well, at least it's good to know I'm not the only one whose business is being totally screwed by Compete and their bogus traffic stats. My site, singularcity.com, is a Los Angeles-focused online magazine and affiliated social networking community for people who are single -- how can Compete's alleged 2 million sample subjects spread across America possibly supply accurate numbers?
Yet in advertising and PR resource directories, like SRDS and Cision, the Compete number is presented front and center as THE source for advertisers to determine traffic numbers, which in turn, determines their media buys. (By the way, SRDS purchased Compete so they have a vested interest in promoting it as THE resource for traffic numbers)
In regards to Cision, they’ve followed SRDS and use Compete as well to provide traffic numbers. I actually had a PR account executive from one of the top travel PR companies in the world say, “Well, we don’t know why – we’ve just assumed Compete was what we were supposed to look at.”
The most egregious part of all of this is that advertisers are looking at these counterfeit Compete numbers presented by SRDS and Cision – then blowing off sites like mine because Compete says I’m getting less than 3,000 uniques a month. As a journalist, such faulty statistics being presented as facts is unconscionable, and as a business owner, I consider it to be corrupt and unfair business practice on the part of SRDS.
Kim CalvertLos Angeles, CA
Thanks Rand for putting together the survey and the mention to my article.
The two times I compared data from different tools I end up using Google Analytics, it is the measuring tool implemented in the sites I have access to and Trends for Websites as the public one to compare.
It is easy to guess Alexa, Compete and such are clearly far from the reality Analytics pictures so I decided not spent time comparing them.
For those who want to compare this is the procedure.
Share your conclusions please.
BTW Rand, the chart from Google Trends compared to Analytics one you show there in the article is pretty accurated in shape and numbers take a look https://ow.ly/i/pspX
Google Trends one starts Jan 2011, G. Analytics some months before but for the same timeframe they match
Numbers for March 2011 are around 100K Dayly Unique Visitors in both cases, they match again!
Why you say Trends is way off?
Ani - you're comparing the DoubleClick Ad Planner vs. Google Trends - they do look like the curves have a similar shape, but don't look anything like Google Analytics' data :-)
right I realized later it was not GA data, thanks for pointing that outthat different is the gap between GA and Trends? probably numbers are going to be but chart lines?
Rand
Very interesting, I was just pondering something yesterday - that is, whenever I am on a public wi-fi I notice Google "auto-detects" me as being in New York, NY or New Haven, CT (like the screenshot) when I am nowhere near either. See the screenshot I took and tweeted about: https://screencast.com/t/DQ0PKMRpNl
Someone responded to my tweet and said this would skew the data Google collects (Google Insights for instance, which you can drill down by location), my hypothesis being that Google would think I'm in New Haven, CT when I am not. If you, or anyone knows more about this I'd be interested.
I do have a site I can share the comparison data on when not on the run :)
-Dan
Your take on G Insights is actually good news to me. I use Insights quite a lot for determining/guessing regional differences. If you are a New Yorker visiting San Antonio, you are probably using New Yorker terms for your search...so, I'm glad that the wi-fi data records you as one.
Otherwise, the accuracy of Insights is science fiction, I'm afraid. The same query can return very different results from different computers on the same day. Accurate search data would make our lives soooo much better.
To attempt to predict traffic on a website without tracking code is an act of futility. Alexa, Compete, etc are all completely useless and always have been. I noticed this trend years ago and can confirm exactly what Rand has seen in his studies. There's no need for screenshots, Rand. I've worked on hundreds of websites and I've been in SEO for 7 years and I've never seen an accurate representation of traffic on any of the sites above. in fact, the only assumption to make regarding sites like Alexa and Compete is that if their data says "X" the truth is certainly "Y".
Really Frustrating!
These tools are a waste of time. It's got to the stage where I'm even beginning to doubt Google Analytics' accuracy due to a couple of problems and things I've noticed. I'm installing one or two other open source alternatives to GA to see how they match up.
As for any third party tool that tries to estimate a site's visitors; you'd be better off manually looking at the number of comments, the number/quality of social media followers and any partnerships/associations with authority figures than trusting any of those numbers.
Jenny, some of the basic analytics concepts, like what is considered a visit, are handled differently so none of the decent analytics tools match numbers to perfection.
Supper-accuracy is not relevant but how you interpret data and the insights you get from them.
It really depends on what you're planning to do with the data and the cost of the potential impact.
Jenni.. even I have been testing out the efficiency of the GA and I have been using Piwik as an alternative(it's Open source) and I got to see some differences but those were not of that much issue.. the only downside of the Opensource is that their development gets slowed down(thank god it's not in the case of Piwik) and they need separate webspace to store the multiple types of traffic data so if anyone doesn't want to show his data to G, the open source (popular ones e.g. Piwik) are the preffered options!!
When I get a chance I'm going to do a proper comparison of GA vs. Piwik vs. Open Web Analytics (and possibly try to incorporate something like SiteMeter in there as well).
What if Google Analytics is the one that's off?
Ahhh, 2012 - End of the world!
No analytics tool is 100% accurate, we all understand that. the question is how much inaccuracy is widely accepted, 5% 10%? how do you know how inaccurate is your tool if no picture is perfect?
Now that Google Insight and Google Trends are now one single product, we lost the "Google Trends for websites" functionnality. What are the accurates ways of knowing the traffic of a website ?
Great to see some data that backs up what most of us already knew - these sites are pretty awful at accurately displaying traffic data. At least now you'll have an easy way to respond to the lamenters:
"Dear Whoever,
https://www.seomoz.org/blog/testing-accuracy-visitor-data-alexa-compete-google-trends-quantcast
Rand"
Thanks for sharing, Rand! I've never trusted third party analytics tools. My biggest problem is that they are useless for small websites. Unless you're getting tens of thousands of visits each month, they probably won't even be able to give you data.
We received 24,000 unique visits this year with our biggest month in August at 3,500 unique visits. So we're not working with nearly as much traffic as you, but still a respectable amount for our growing business. Out of all the tools listed in your post, only Compete.com was able to provide data on our site and their estimates were wildly inaccurate - over 600% off in some months!
On top of that, we work primarily with small, local businesses who only get a few hundred visits each month, so trying to use these tools for one of our clients would be pointless.
As I mentioned above, most of the tools couldn't provide data on our site. However, here's a comparison of GA and Compete - https://www.prontomarketing.com/google-analytics-vs-compete-com/
Numbers are waaaay under reported, but I guess the directional trends are close - kind of, sort of.
Thanks for keeping it TAGFEE as always Rand and sharing actual data. Very interesting [damning] indeed to see how off some of this data is. I know a lot of people rely on these third party data sets either for forecasting purposes or for attracting funding so it's nice to have this to point them to.
Obviously the small sample size (one site) doesn't conclude that these services are always wrong but it's certainly nice to know just how wrong they can be.
I'd love to see you look at some of the other paid services (e.g. hitwise) in future if at all possible :)
We are in the middle of doing this with Hitwise at the moment and it's not looking good. We're looking at data from the last 12 months and they are regulalty showing negative correlation to our data. Hopefully we will get some answers back from them soon....
I just had a discussion with a colleague the other day regarding the accuracy and usefulness of Alexa. Does anyone share Alexa data, or any of the others mentioned in this article, with clients?
From this discussion it seems like SEM Rush is the most reliable 3rd party data provider out of the bunch though. I'm going to start doing some testing of my own and compare Analytics and SEM Rush data for clients (both large and small) across a few different industries.
I thought that ad planner has more accurate trends. but sure only for fast comparison and not for all sites.
For max accuracy i prefer add multiplicator calculated from known site stats.
And here are stats : Adplanner , Alexa, Compete, Analytics
I may be a little late to the party. It definitely makes a person nervous to post stats like this, but I like the philosophy of being transparent. Like many of the other commenters here, my sites are all too small to have data from the likes of Alexa. So here goes...
Site: www.robotvacuumcleaner.org
Google Analytics
Compete
Graphed Together
As you can see, the data from the two sources is pretty much going the same direction until May, where Compete goes off the rails. I have two other sites that show up on Compete, and the results are pretty much the same. That being: too low and sort of an exaggerated caricature of the real graph.
Here's another of my sites with the data graphed together. And here is a third site with data graphed together (as you can see, this third one is a Panda casualty).
Here are percentages, figured: Compete estimates divided by Analytics reported uniques.
First Site: 22.3%
Second Site: 43.5%
Third Site: 29.9%
- Dave
All I am familiar with is the Alexa data for my site. Almost all of the info. is horribly inacurrate. Bounce rate off my 100%, % using search to find site understated by 200%, a few of the high producing keywords correctly identified by false positives greatly outweigh the few correct ones. The inaccurate data seems to be getting worse, not better. Often wondered why advertisers would have any reasonable idea which sites to pay what rates if all the data is *this* inaccurate.
I reviewed an account of mine that has had over 200,000 unique visits this year and the only site that gave any data was compete.com the rest said there was not enough traffic. The data was way off on compete.com however the graph was only slightly off.
How do these companies make such rash assumptions and are still seen as reputable?
Great post Rand.
But SEOmoz use any other analytics software more precise like Urchin or webtrends, which are based on log files of the webserver?
If yes, have you compared them to the GA? Is there any discrepancy?
I believe that the top 1000 websites can't rely only on Google Analytcs.
Did you ever publish the results of the survey?
I never have doubt in Google until this post. Our site for fitnes gear www.ringsport.rs in Google Analytics have (not provided) for first phrase in organic results! That is just not fair!
To answer those who state SEMRush is "more accurate" -- it's actually way, way off.
Here is a screenshot using their info-
https://i.imgur.com/OZvUP.png
Only 84,070 unique US visitors for the month of December 2011??
SEMRush actually shows them LOSING traffic from Jan 2011 - Dec 2011!
Thanks for bringing this issue to light. When I first saw this post I couldn't wait to read it. This is a problem I have been dealing with for quite some time. Unfortunately, I am not at liberty to post our company’s side-by-side numbers, but I did just finish the quiz and much like your findings I have found great discrepancies as well. It also makes it very difficult to explain to VC as well.
Additionally, I have found discrepancies with comScore and Urchin. I think the problem with Urchin (since it is server side) revolves more around the installation and the framework of your website.
So what I have found is that it is best to simply present the Google Analytics numbers to VC just as you have here. I have found them to be the most accurate, very reliable, widely used and understood, trusted, and best of all FREE :-)
- Cap
Your sample size is great so the differences are stark. Without direct access to someones site it's hard to approximate, but that's all these tools are for.
For Advertisers you have to install these codes so they can get a better measurment.
I compare Alexa v. Google Analytics v. AWstats to graph what my stats look like for:
and usually ALL of these sources come up with wildy different stats. So I'm glad (kinda) to hear I'm not the only one vexed with how site analytics numbers vary from tool to tool.
Wow I couldn't believe that the Google Adplanner Data is so bad. Nevertheless it's great to see that your analytics account shows more than 55% returning visitors.
I have experienced the same issues with Compete and Alexa. As mentioned earlier by Mister G, I am also beginning to doubt Google Analytics, as it showed there were hits from two countries on a specific day ( UK and US ). However, in the unique visitors it just showed 1 visitor for that same day. I know this is a small number and rather insignificant compared to the numbers we are talking but this was a month ago. That doesn't make any sense, hits from two countries but just 1 user, what did they fly overnight? This is for the site www.litbloc.com
Very good article, a lot of webmasters denounce totally wrong statistics provided by these tools. In Europe, there are also Médiamétrie / Nielsen NetRatings, which is also very far in terms of accuracy, as i've seen for most of my websites, likes this one (which has a lot of traffic). I don't even see the point of those tools and of the money which is put in...
I have read through the entire discussion and find it tremendously useful.
Has anyone else been asked by Alexa to put something like the following script on each page? Is it actually hurting the page by taking time to run it through Alexa?:
<script type="text/javascript" src="https://d31qbv1cthcecs.cloudfront.net/atrk.js"></script><script type="text/javascript">_atrk_opts = { atrk_acct: "lV9af1a0mZ002M", domain:"homedestination.com"}; atrk ();</script>
I use Alexa Certified Site Metrics.
I did put the javascript code on my site, it's similar to the google analytics code.
Certified Site Metrics gives prety much the same results as google analytics (maybe a 5-10% difference in some stats).
If you have a website focused on SEO, social marketing, etc, Certified Site Metrics will hurt your Alexa rankings.
If you have a regular website (like mine), it will probably help your ranking.
My site moved from 200K+ to 50K.
my website is https://omgghana.com, and I use wordpress jetpack, alexa and Google analytics. Truth is, they are all accurate and special in it’s own way, i use alexa pro, with alexa pro, they let you put a code on your site like google analytics so it’s not solely about the toolbar as many have speculated it to be. They give you overall analytics. On a good day, I get 120k pageviews with WordPress Jetpack, that will amount 200k pageviews on google analytics and 110k on alexa so all I’m trying to say is they all use different ways, using all 3 makes me know that regardless I crossed 100k :) … it’s good to experiment with blogs, NOTHING IS 100% accurate not even almighty google analytics!!!!
I just had view at the Alexa ranking of Moz.com. It shows 281 as the Global Rank and 293 as the Rank in US. How does that work out to be? How can the global rank better than country rank?
Thank you for tthe article, thank you for your efforts!
I have tested the instruments you offered. I have also tested the instrument mentioned in the comments (similarweb.com) and found it to be the best in our days (4 years ago it was different, I suppose).
I may also offer another site: https://www.rank2traffic.com/
It is not as good as simillarweb, but it is not commercial and offers traffic history (more than 7 years, if I am not mistaken).
It's sad that with all the technology available to us, we still do not have a single accurate source for data.
The premise here that Analytics is 100% accurate is equally wrong; real traffic analysis shows that Analytics is off by 15% in that it misses quite a few real visitors. Anything that is JS based is going to miss short. I'd like to see an analysis of Alexa certified vs Analytics. Anything "Estimated" is not to be trusted.
One thing I appreciate about this article is the discussion of Compete.com. When I look at my own site's Google Analytics, and then take look at Compete.com, I am generally stunned at how bad they are at estimating my traffic. I am glad that someone else says the same thing. Compete is terribly low in estimation, and very often doesn't know if my blog (starbucksmelody.com) is growing or shrinking! If nothing else, I think Alexa gives you at least a good comparative idea of where your site sits as compared to others. Thank you for the interesting article Rand!
Please try with nextpr.info as well and compare the results with each other. And please also tell which service gives the best results to rely one.
Honeslty the analytics looks good - though it coulb be much better, regarding the step by step coming visitor on the site.!
Rand, are you going to publish our results?
Once again Rand, I applaud you for sharing actual numbers with us - TAGFEE is a blessed thing - and I'm just as glad to see this article because I've been saying for years how flawed such outside views are. Too many people, investors, advertisers, are lured by the data and quote these site's as if they were golden. It also validates my perspective that the methods used by these companies is unbeleivably ridiculous. Their premise is they get a big enough "sampling" that they can then extrapolate. What baloney! The web it too massive, there are too many mind models using the web.
People really need to wake up to reality and get out of 20th century concepts.
Alexa-silently the trend have evolved. People are using Alexa just for fun not for their business orientations, even relying on compete can disrupt the analysis.
Great article. I've found a basically identical problem.
One thing I'd like to add - I actually do have "qunatified" data for Quantcast. I opted in and added the code some time ago, but the data is still radically innacurate. The basic trends seem okay, but it still shows less than 50% the traffic that I see on Google. Very odd.
Randfish,
What's the answer? Which is the best service to rely on?
Thanks
I have put together screen-shots of our site's (i.e. sidebuy.com) traffic report from the following resources:
- Google Analytics- Alexa- Compete &- QuantCast
Here you can see them:
https://sidebuy.com/misc/stats.html
As you can see, we have purchased Site Audit from Alexa, therefore, their information compared to Compete and QuantCast is closer to the reality (i.e. Google Analytics). For a few months, we thought that Compete is capable of gathering good information. But recently and within the past few months, the reports on Compete are pretty off. I know that Compete's data only reflects US traffic, but still their report is not representing our real traffic. QuantCast for a while showed the right trend, although the numbers were very wrong. For the past two months as you can see, they are not showing anything for SideBuy.
I've watched this for a long time. Our site has respectable, but not huge traffic, since we are focused on just one city. However, the data that is shown for Compete and Quantcast is usually at least 1x order of magnitude wrong, sometimes less, sometimes more. Trends never follow the real numbers. Have never seen it get to +-200% on the real data, no matter what the traffic is.
Alexa is even stranger. Given how they use their toolbar, I did manage to get a jump in our traffic for them when I installed the toolbar! I'm guessing they deal in very small sample sizes because you'd be nuts to keep that damn toolbar installed.
Quantcast is spot on, if you enable measurement, but their estimates wihtout measurement suck.
Why would any of these tools be as accurate as Google Analytics? The data capture methods are all different. It's like comparing an indy car to a VW bus. GA = 1st party cookie vs Alexa toolbar data, Compete/Quantcast panel data, doubleclick 3rd party cookie, and whatever method GTrends uses.
That's why almost any third party tool is off from a web analytics tool ... data capture methods. And a 1st party cookie is always going to be the most accurate.
But you can use these tools to get soft, directional data in most cases as long as you compare your site and your competitors sites inside the bubble of that tool. Just be sure that when presenting the data to anyone outside of yourself or your group that you properly explain how the data was captured and set expectations.
Thanks for another great post Rand.
I followed your challenge, did the survey and got carried away and decided to make public the stats for one my sites, ruadireita.com
You can find the side-by-side here.
All you guys, feel free to use it anyway you see fit and if you need any more info just contact me, as always a link would be appreciated *wink wink*
BTW you can see there the hit we got from Panda Update, 12th of August when it went international and included the Portuguese Language. We lost +50% of our traffic from Google. That was a bad day :(
For my own sites, and often for clients, I run both Google Analytics and Piwik. Generally these normally come out within a couple of % of each other. The problem I have with using just one source of stats for decision making is that you do not know how correct GA really is so it doesn't hurt to double check.
I had just starting looking at Alexa and Quantcast for data last year, but you've made me think again. It's a shame that not even one of those services was close enough to recommend, :(
Thanks for sharing this data. Always good to have "real numbers" versus some projected amount to make business decisions on. Seeing these examples is a great way to properly demonstrate how inaccurate these tools are, and shouldn't be used as "matter-of-fact" but instead as tools that *might* be useful to indicate trends.
But, on the other hand, if all the data is skewed, then the comparisons should be similar.
Thanks Rand for providing such a detailed & insightful info. Just wanted to say thanks for doing all that you do….
We are way off every month. Our worst traffic month is 2-3 times higher than what they have listed for our best
Rand, have you thought about expanding your 'post analytics' to 'site analytics'? No better way to give third parties the truth than to lay your cards out?
I never trust third-party sites only take some idea and leave it because results are awful.
Great offer Rand
I would like to share analytics for my project but I have serious concern about privacy. Can I give screenshots without declaring website name?
Great question; I'd love to be able to help the cause but I am sure at a corp policy level there'd be concerns...
Yeah - definitely! Just anonymize the names and show the charts only.
Very good article, more and more webmasters denounce completely wrong statistqiues provided by these tools. In Europe, there are also Médiamétrie / Nielsen NetRatings, which is also very far in terms of accuracy ...I shared the statistics and data for one of the sites I manage, Allogarage ( https://www.allogarage.fr/ ), a comparison of auto repair shops in France (French traffic 100%), the data is visible on the blog here.
Hmm... yes in AU we get kinda ignored by most of these sites I've actually struggled to find a site that gets a reasonable amount of traffic in all of the tools, most of the sites I've worked on previously have mostly AU audience so many of these tools just choose to advise not enough data even with thousands of visitors/day it's not enough.
So sorry to say I'd more than happy to share my screenshot of data for my own projects but none will show data in all of those platforms. Some will show in compete but not in Quantcast etc...
I picked a larger site that gets around 100,000 visitors/month and looked at the GA data and how it compared against the tools estimates... the basic insight I get is that Compete was mostly under reporting by 50-75% and it also showed the opposite ie it showed a spike in july when there was actually a dip as Australia started a new financial year. The same site still did not get enough data to satisfy Trends or Quantcast...
Google Trends and DoubleClick Ad Planner are the biggest let downs with Google Trends showing no data for that domain and Ad Planner over reporting daily visitors but under reporting total monthly visits by 55% and under reporting pageviews by 30%.
Also picked a bigger site that gets twice the traffic as SEOmoz does and DoubleClick under reports by 54% and pageviews under report by 20%, unique visitors under reports by 25%. Compete is even worse with under reporting unique visitors by 96% but the traffic trend in Quancast was mostly right and pageviews and unique visitors was only off by around 5% consistently over several months. This site was a global website with a large international audience...
Those tools might be ok for US or even international sites but outside of that it's a real struggle to make any use of them consistently... it might help if you can gauge an industry baseline but don't be surprised if you are well off the mark
Yes I agee 100% most of these tools are designed for the US market if you use a tool like Quantcast or Compete for a website which has around 500k visitors you see silly things such as 1k or less in some instances.
Good to see this data.
Personally, I never read into Alexa for sites (more for advertising deals etc) outide of the top 500, as it can so easily be gamed with the toolbar. I've seen services for x/mth to get your traffic rank into a particular range.
Interested to see the survey results for the Google Trends/Analytics.
Exactly. I can't believe people still rely on Alexa data. Just by installing the Alexa toolbar in our browsers, my brother and I alone have been able to get the primary site we use to rise from a rank of 300,000 to 120,000, although it's leveled off recently and it probably takes more than two people to rank in the top 100,000 (but how high could 10 people get it, etc). Obviously it's going to be more accurate for sites much higher, but they are going to be disproportionately techie/SEO/marketing related because those are the kinds of people that install the toolbar to see what the toolbar says... and thus also end up contributing to what the toolbar says.
Thanks for the great post Rand.
Like others have said, small UK-based sites aren't even privavledged to most of this wonky data. The only comparrision I can get is from Compete.com which shows US-only visitor data, and to be fair to them, comes with a warning about sample size. Their data for oushop.com fluctuates pretty wildly compared to GA and tends to underestimate significantly, but if you squint your eyes and are feeling charitable you might say that the trend is kind of right... https://i.imgur.com/72yPj.png
Hey Rand,
I'm sure you already know this, but it should be noted that Quantcast and Compete are showing data for US Internet users only, whereas the others are world-wide.
Just a thought but: in terms of Google's data; surely they do not show the data due to the fact you can get an idea of the traffic of your competitors (just like they did with link:domain.com etc) hence it is ~500k out on a certain month?
I find it amazing that so many SEOs actually rely on these tools for data (especially beginner-SEOs just learning about all the various web tools/apps that grab statistics and measure metrics.) I know I use to be obsessed with getting a low Alexa rank, but after learning roughly how the "Alexa Rank" was determined, I stopped focusing on it so much (they weigh traffic from different countries with a different weight/scale, claiming that US traffic is weighted higher than a small Country with low presence.)
Good to see your analysis drilled right into the problem with these sites, using real raw data to back it up.
I have long been annoyed with the clout that these third-party analytical tools have been given over the years, especially in the media, because they are more often than not, wildly inaccurate. However, unless a website is willing to provide access to their internal analytics, these third-party tools are unfortunately the best/only options currently available.
Here is one of my favorite blog posts on the subject matter, in which Reddit shows how flawed these tools are with regards to their own traffic - https://blog.reddit.com/2010/07/experts-misunderestimate-our-traffic.html
This should go some way to emphasize the fact that you should be very aware of your data set or metrics analysis before quoting and making decisions based on it.
Good post, I see soo often people talk about statistics from these tools yet I always have advised to not take the data, as they are a very very rough indication of the overall sucess.
This only adds further fuel to my points =) so great post Rand =)
They are, I guess, accurate enough overall for the biggest site I work on, CiaoItalia.com. Some months it's way off, others are spot on. I posted some screenshots to Twitpic:
Analytics, Compete, Quantcast
People use these tools because they provide a quick answer to their question: how much traffic does this website get.
Anyone can answer any question with a made up answer. Look at Yahoo Answers!
Rand,
I think it is well known to all of us that the DATA shows in free tool can be always have some statistic issue...the accuracy you cannot judge. But one think is clear that Google Analytic also skips capturing data if it comes from same IP or for unknown reason !!! :X ....but if you check Hosting Traffic Track it provide more accurate data....
:o)
Rand this is really interesting post actually it is more shocking than interesting. Well Alexa and Compete are not the tools we can trust but what shocks me most is the huge difference between Google analytics and Google Trends and Doubleclick. Your post really an eye opener for those who uses the Alexa Compete metrics while concluding their site performance.
Well after reading your article I am pretty sure now that even Google is also providing imperfect results frankly speaking I am feeling option less.
Ok, this is quite shocking and frustrating at the same time. For research and other stuff I somehow believe on compete data but the way this graph is inaccurate to the real numbers (example SEOmoz data) is quite frustrating to me and to lot of folks here...
All i can say is that there is a strong need of a data provided that provides (if not exact numbers) then at least more or less similar numbers... i believe Google trends can do that as Google already have lot of analytics data with them!
#Justathought
Hi Rand! I think that you are right, however, the objective from these tools is not to be exact, is just provide a Intelligence Competitive to compare with other players like Search Engine Land, SEOBook, etc.
Yeah, but if you buy ads from SELand, they'll show you their traffic, and I can say confidently it's nothing like what's suggested above. Aaron from SEOBook would likely say the same. The comparisons are as useless and random, in my experience, as the numbers and trends themselves.
Thanks so much for the info, I will pass the word along to discourage use of these sites. I appreciate you doing this research and sharing the info with us.
off by almost two full orders of magnitude (our daily traffic is about 2X what they estimate our monthly traffic to be).
An "order of magnitude bigger" normally means that something is ten times bigger. So two orders of magnitude would be a hundred times bigger, not twice as big. Doesn't take away from your point, though.
Natan - Compete shows 26K monthly visits, but our analytics is showing more like 1.35mm monthly visits - I think that's about 52X, so much more than one order of magnitude, but not quite 2.
Not all the third party sites would work for our site.
But for compete.com, the difference was astronomical.
SITE: https://www.stadriemblems.com
GOOGLE ANALYTICS SCREENSHOT:https://www.stadriemblems.com/images/stadri-google.jpg
COMPETE SCREENSHOT:https://www.stadriemblems.com/images/stadri-compete.jpg
Thank you for pointing this out. I was just noticing this last week when trying to evaluate sites I wanted to advertise on. The site owners were stating their traffic numbers were X, while Compete showed Y. I wondered if there was some exaggeration going on until I compared my own companies site traffic to Compete. Yes, it was WAY off.
Why even have such tools out there (such as Alexa, Compete) when they are so horrible - it just confuses people who don't read posts like this.
Wow, this couldn't be a more timely article. Just had this question. Again. "how much traffic does competitor xyz receive each month?" It's always the same answer.. these tools are incredibly unreliable.
I have an unfounded, off-the-cuff guess about this, @Rand.
Google over-estimated the traffic on the site. I noticed that on the AdPlanner site, it says:
Daily Unique Visitors (cookies)
Would it make sense that a lot of people in the SEO community are clearing their cookies regularly to try and circumvent personalized search results and are counting as another unique if they are returning the to site 2-3 times a day?
I'd be curious about getting traffic screen shots from highly trafficed sites to see if that is, in fact, correct. I have a feeling if it were a site outside of the SEO community, that the data would be more or less accurate.
I suppose that is the point of this article... To analyse data from other sources.
Love this. Thanks Rand.
Good review of the different data sources - I've often wondered who actually uses these kinds of trending data and what for.
I've had a quick look at the data from Google Trends and I think it's showing about 60% ish of the actual trends.
The deviation I've seen comparing Google Trends and Analytics 9 different sites goes from 0% minimum to 57% maximum what makes an average of 25% (higher for GA)
SemRush is the most accurate after Google Analytics.
Can you describe better why?
I compared all the services that were highlighted in Rand's post. And I too noticed that the data were inaccurate from these sites. Google Analytics tracks data like no other tool and it looks very very accurate. For keyword analysis and competition research I use Sem Rush and the data they provide makes more sense than Alexa, Compete etc.
Thanks Group16.The thing is Rand's post neither mine is evaluating Google Analytics accuracy.
Rand is comparing Alexa, Compete, Doubleclick, & Quantcast against Google AnalyticsHe talks about Google Trends but not adding any Trends chart to the mix. What they see is not shared here.
I compared Google Trends for Websites vs. Google Analytics Data. no other one.
Most of the comments agree all these tools are inaccurated BUT are they saying the same abot Google Trends? I don't think so.
Did they take the time to compare and draw conclusions? I don't think so.
You state "SemRush is the most accurate after Google Analytics" and I ask you to bring some evidence to certify what you state.
"What can be asserted without evidence can be dismissed without evidence" Hitchens' razor
It's almost as if "the powers that be" are cooking the books in order to justify their existence, and be able to sell more services. (in a comical voice) It's a conspiracy man !
Seems to me that "visitor measurement" needs to be brought down to a real world clarity, that happens to be useful where MOST businesses are, and that is SMALL. So measuring where it really matters is not even a capabilty for those that measure ! WOW !
The gaping void between IM world and the real "Brick and Mortar" world businesses makes this transition to the "new media" very difficult and confusing for average business people, and that's interesting, because, they are the ones who (a lot of) the money will come from to support the new media.
I think that instead of having some select bunch of "know it alls" telling us what's going on, we need to determine on our own, what works for us, and keep the focus on creating authentic content (tell our individual stories), and develop a growing number of relevant paths back to our solutions. That's a much wiser use of our time than worrying about how some larger than life measurement guru authority sees us.
For most businesses (especially MOST local ones) high traffic means nothing, because there usually isn't anything to sell to the (huge number of) visitors, especially if they're outside of the typical geographic service area of a given business.
Media (much like the government) seems to forget that it is here to serve us, and not the other way around.
So what's the point of sweating about how were rated by the rater's? What do they know (that's accurate and helpful) ?
Get back to work and tell your story authentically and widely so you can get more customers.
When comparing the viewing figures via Joomla Stats of sites I work on with Google Analytics figures, I've always found a massive discrepency, not 10 or 20 views out, but hundreds. Is there one source that is generally the most accurate? My guess would have been on Google Analytics but it seems that can be inaccurate at times too!
Great, insightful post - thank you.
try use joomla, alexa pro, google analytics, compete, i prefer comparing ... the average is always better right ..
Great analysis, I have always doubted the accuracy of these data, but never thought especially GA will be off this much, it is probably the sample data (average) they are reporting back.
I'll do a similar analysis for one of clients to see the similarties/discrepancies of the data reported by various resources and send you the link.