Testing the Accuracy of Visitor Data from Alexa


SEOmoz.org had 13.8mm visits from 6.25mm one of a kind guests a year ago (2011). Those numbers are entirely energizing, however what’s not energizing is the outer recognition made by outsiders like Compete, Alexa, Quantcast, Doubleclick and Google Trends for Websites. These destinations report enormously lower and wrongly inclining information – and SEOmoz isn’t the only one in encountering this disappointment. We’re among many destinations I’ve conversed with who’ve gotten messages and remarks regretting our poor development or horrendous year on account of these awfully incorrect administrations.

Here’s a screen capture of our genuine traffic from Google Analytics:

SEOmoz Visitors Overview

Presently we should take a gander at the correlation with every one of those administrations:


Above is Alexa’s gauge of SEOmoz’s web traffic for as far back as couple of months. It’s difficult to tell how precise they are, on the grounds that they’re not appearing accurate numbers, just “percent” of “achieve.” They do effectively take note of that traffic was down in December (the most recent two weeks of the year were extremely moderate for us due to the occasions, which is something worth being thankful for – even SEOs merit a break) :- )

Verifiably, Alexa demonstrated an any longer timespan and significantly more wrong information, at one point assessing that our traffic had dropped year-over-year since 2009. I’ve had very much regarded VC reserves connect and inquire as to why we were battling and whether we felt the SEO showcase was evaporating a direct result of those graphs… Presently, Alexa’s positioning us as the 472nd most famous site on the planet, which is certainly way, way off.


Next up is Compete.com’s gauge of SEOmoz’s traffic. They’re considerably more explicit, yet appallingly, missing the goal. For a period, I’d trusted Compete would be a vastly improved contender to Alexa, yet those expectations kicked the bucket a couple of years back. This diagram isn’t simply wrong, it’s directionally in reverse (we developed when they demonstrated us contracting and contracted where they show us spiking at year-end) and off by just about two full requests of extent (our every day traffic is about 2X what they gauge our month to month traffic to be).

How anybody can believe that information is past me, since you can without much of a stretch analyze numerous locales who distribute their traffic subtleties (as we do) against Compete and see this disparity. To be reasonable, I’ve heard that for the best 1-2,000 most well known destinations on the web, they’re not terrible, however I can’t by and by affirm this.


Quantcast’s gauge of traffic looks similarly horrible to Compete. It’s directionally wrong and off by numerous requests of size also. Quantcast’s redeeming quality is their “Evaluated” program, which demonstrates genuine, really exact and estimated numbers for locales that pick in. I wish they’d adhere to that model only as opposed to giving these irregular theories on locales they’ve excluded in the program, however. I’m additionally truly attempting to see how 17,671 remarkable individuals could make just 11,005 visits… That is a mind secret.

Google Trends for Websites

Google’s my last, best expectation, and since they catch such an enormous level of locales’ traffic in Google Analytics, I’d expect they have a quite brilliant information demonstrating framework to work off. Clearly, that conviction is mixed up. Google’s in no way, shape or form as awful as Compete or Quantcast (and potentially superior to Alexa), yet it’s still way off. The directional information is kind of close, yet the every day one of a kind guests include appears at ~200K in December. Our examination says it’s ~47K day by day or 722K for that month.


Since Trends and Doubleclick are both under Google’s working umbrella, you may be enticed to think they utilize similar information… Truth be told, Doubleclick Ad Planner’s gauge of Moz traffic and Google Trends for Websites seem to have at any rate marginally various numbers (hard to tell without a doubt dependent on GG Trends’ fragmented diagrams). One thing I can tell without a doubt – nor is precise, nor even directionally right.

The after some time diagrams don’t exactly coordinate one another (however they’re close-ish); it would appear that Doubleclick is indicating higher traffic to SEOmoz for the most part than Trends for Websites. The nearest information point is their evaluated time nearby, however I don’t know I can give them kudos for that. In the event that you put on a blindfold and toss enough darts, one of them will most likely draw near to the board. It’s hard not to feel that route about these numbers, as well.

Presently here’s the rub:

As of late, Ani L√≥pez expounded on Comparing Google Trends for Websites versus Google Analytics Data and demonstrated a couple of precedents that recommended more noteworthy exactness than what we see with SEOmoz (and OpenSiteExplorer, too FYI). Consequently, I’m requesting two favors from you to help show signs of improvement sense for the overall value of these devices.

The first is to take the brisk review connected to underneath:

Take Our Traffic Comparison Survey

it would be ideal if you take me! (opens in another window)

The second is to, if conceivable, take screen captures of your own examination versus Patterns/DoubleClick/Compete/Quantcast/Alexa and offer them in the remarks beneath. For any individual who assembles a convincing next to each other, I’ll joyfully incorporate connections in this blog entry to your website and to the pictures demonstrating your traffic versus what these outsiders report. Ideally, that motivating force can help goad straightforwardness from those of you willing and ready to share some expansive site details.

Much obliged as dependably for your assistance – anticipating getting a more extensive perspective on these devices’ exhibition. For the present, I’d remain profoundly doubtful, yet we may return to the subject on the off chance that we get convincing information in the overview and in the remarks (else, I’ll simply refresh this post toward the week’s end with the study results, and since they’re unknown, give full information).


Please enter your comment!
Please enter your name here