Is Beacon Inflating Facebook’s Visitor Numbers?
Erick Schonfeld
28 comments »
While Facebook’s Beacon program has been drawing the ire of many people because of the privacy issues surrounding it, advertisers might be equally concerned about whether Beacon is inflating Facebook’s overall visitor and traffic numbers. Yesterday, Compete reported that Facebook’s unique visitors went up 20 percent in the month of November, after a controversial dip in September and practically zero growth in October. Could the increase have had anything to do with the launch of Beacon in early November?
In a word, yes. Every time someone visits a Facebook Beacon partner (there were 40 of them at the time of the announcement) and performs a pre-defined action like writing a review or rating a hotel, a little Beacon toast pops up alerting you that your action is being sent to Facebook. That pop-up is actually coming from Facebook, and in some cases may be counted as a Facebook page even though the person seeing it does not normally click through to Facebook. It is triggered by a tag on the partner’s page known as an iFrame, which then tells your browser to load a page from Facebook within the site you happen to be visiting. This occurs even when a non-Facebook member visits that page and performs the same action. In that case, it creates a ghost iFrame, though, because the viewer does not see anything. But data is still sent to Facebook.
I called up Jay Meattle at Compete, who wrote the post, and he confirmed that of the 29.2 million unique visitors Compete counted for Facebook in November, those could also include visitors to Facebook iFrame “pages,” which are really nothing more than a pop-up on a partner site. So, for instance, if you write a review on Yelp, a Beacon partner, a JavaScript is executed for all users writing a review (http://www.facebook.com/beacon/beacon.js.php) and an iFrame is launched (http://www.facebook.com/beacon/auth_iframe.php). That review on Yelp can now count as a unique visitor on Facebook.
So how many Beacon iFrames “visits” did Compete mix up in its numbers? Meattle says that 2.3 million people triggered a Beacon iFrame in November. But he wasn’t able to tell me what the overlap is because a portion of those 2.3 million people could have already been counted as Facebook visitors when they visited Facebook previously, and thus would not be counted again. The Beacon pop-up would be treated in that case like any other Facebook page. (That is, after the first one, it would count towards page views, but not towards unique visitors). But remember, these iFrames are triggered by non-Facebook members as well who never go to Facebook proper. That would explain why Facebook had such a huge jump in visitors in November. If all 2.3 million of those iFrame visitors were counted improperly as part of Facebook’s total, that would account for nearly half of the 4.9 million jump in unique visitors measured by Compete.
Another strange thing about the Compete numbers is that they show unique visitors going up by 20 percent but page views only going up 2.58 percent. That could be due to lots of things, like Facebook users doing more stuff with apps on a single page, clicking off to Websites controlled by application providers, or Facebook just becoming more efficient in not making you click around as much to do the same things. But I wonder if that is partly Beacon-related as well.
Will wel see the same inflation in comScore’s November numbers when they come out next week, or in other measuring services such as Quantcast, HitWise, or Alexa. The folks at comScore assure me that they filter out any traffic or pages not requested by users such as pop-ups, so it might not be an issue for them. I don’t know how the other measuring services treat iFrames. But in an age when Websites are interchanging so much data and so many actual applications, where one begins and the other one ends is becoming blurred. Do page views even matter anymore? Do uniques? Maybe it is time for some new metrics.
Update (12/10/07): comScore just sent me their November 2007 numbers for Facebook. They show only a 2.4 percent increase in unique visitors in November 2007 over October 2007 (to 33.7 million U.S. visitors).
and the way apps work FB forces many installs that are worthless. Facebook has tons of relatively worthless pages views and results to “gaming” they system to inflate numbers across the board.
Also - Eric why not check quantcast the info is available there now
And before anyone says anything, this post was not in retaliation for Facebook poaching our Crunchbase product manager, per Mike’s previous post. We just both happened to be writing Facebook posts at the same time.
So we’re still looking for those negative story leads.
Quantcast shows no such jump
Compete has been known to show some very unreliable numbers
http://www.quantcast.com/facebook.com
Fourth!
Erick - some of us have been saying for a long time we need new metrics. Glad you have joined the campaign!
One thing I have heard over and over (for 10+ years) is that we won’t see a shift in metrics until the agencies accept it. I’d hope in 2008 we can band together and force the agencies to accept it.
Sorry to say this Erick, but I think that this post was a waste of time
.. I like your other posts btw .. 
Hope u take this as positive criticism …
All you guys need to go to this conference and learn about building the next Facebook, Google, Opensocial and more http://www.webguild.org/meetings/web20/2008/
They really pissed you off facebook, huh ?
Cheers
Jack - I’m confused. Quantcast showed an even LARGER jump in facebook traffic (24 million in october to 30 million in November), and they show a huge spike in daily UVs in the middle of Nov…
QUIZ TIME
Here’s the deal: PICK ONE
1) Facebook has incompetent people forecasting growth which translates into hardware, data center…….. costs.
2) Facebook clearly knows the numbers are tweaked because they have many intelligent people that have joined from Google and they know about SCALE.
3) Facebook is just shady.
— Personally, I hate to say it but I think it’s number #3. They are shady.— The only way you will convince me otherwise is if I see two large advertisers publicly say they were notified that these numbers were incorrect by X to Y pages.
On the other hand if the Facebook sales team has been sending out emails to potential customers with these inflated numbers as part of their sales pitch it will be another nail in the coffin and the kid (Zuckerberg) may be demoted to janitor by Christmas day if not sooner…….
It’s becoming a circus!
Sincerely,
Jeffrey Skilling, CEO of Enron
Ken Lay, Chairman of Enron (RIP)
Bernie Ebbers, CEO of MCI WorldCom
Dennis Kozlowski, CEO of Tyco
p.s. Erick excellent story. Very well written!
p.p.s. Duncan, if you can write a Facebook story by midnight, Techcrunch will have completed the hat trick (Mike, Erick and you) of covering Facebook today:)
I think Ben is doing it !
Another Conspiracy theory?
Sheesh!
sure looks like adware wrapped up in a beacon….bit. reminds of intermix tricks used to drive myspace UU’s before Fox buy out
for years a classic way to game metrics has been for publishers to purchase pop up ads to be served on a site and then get credit for the UU and PV for the hitting that ad on the site.
FB seems to have wrapped this up in beacon…perhaps NY AG Cuomo will step in like Spitzer did when he fined Intermix $7MM back in 05
Quantcast agrees with Compete:
Big bump in daily uniques in november
http://www.quantcast.com/facebook.com
Alexa agrees with Compete:
UP up and away!
http://www.alexa.com/data/deta.....cebook.com
3 out of 3 thus far.
Quantcast agrees with Compete:
Big bump in daily uniques in november
http://www.quantcast.com/facebook.com
Alexa agrees with Compete:
UP up and away!
http://www.alexa.com/data/deta…..cebook.com
3 out of 3 thus far.
Quantcast agrees with Compete:
Big bump in daily uniques in november
Alexa agrees with Compete:
UP up and away!
3 out of 3 thus far.
Erick, if you had invested the same time to get your hands on the revenue and profit numbers of Zuckerbook all of us here could get on with their jobs. That is sole metric that is of interest (unless you are from Fortune magazine). With regard to Zuckerbook we could then all doze off.
An iFrame you say… I think someone is a little too Apple-focused. Technically it’s an IFrame.
“these iFrames are triggered by non-Facebook members as well who never go to Facebook proper.”
That is a false statement and disproves the entire argument of this article. YOU NEED TO BE LOGGED IN TO FACEBOOK IN ORDER TO SEE THE IFRAME. Thus nearly all of those 2.3 million people were already counted and thus beacon is not inflating the overall numbers.
http://www.radiantcore.com/blo.....javascript
@19…the pageviews and sessions are pretty constant despite a 20% upkick in uniques. Its very unlikely that a 20% rise in “normal” unique activity would be exactly matched with a c 17% drop in average pages per user and session , especially as the ratios have held pretty constant in the last year.
@ 20 alan p - I think you’re discounting the fact that Facebook’s demographics might be shifting to a more (older) mainstream audience that doesn’t spend hours and hours obsessing with facebook.
@ 21 - not in a month it doesn’t - those ratios (eg pageviews / unique) are pretty much constant all year, but jump hugely in November - I’ve never seen a site do that organically!
By the way, this story seems to have disappeared off Techmeme in 3 minutes flat. Nothing does that in my experience
Here’s my blogpost on the matter:
http://broadstuff.com/archives.....versa.html
Have I got it right, or is my system having a bad day?
My response:
http://mediatrending.com/2007/.....r-numbers/
To Ryan M. my response posted on your blog and here:
How can the FB cookie be checked without a request being made to Facebook.com? It can’t.
The partner sites don’t have the ability, nor does any website, of accessing cookies based on your computer from another site, which it has to happen on FB’s servers …
Therefore if a request is being made to FB it doesn’t matter if you’re a user or not, if the tracking tool counts requests to domains like that then it will inflate the numbers.
Hi Erick, here is how you get definitive proof something is off. You add two columns to that table: Page Views per Visit and Page Views per UU.
A website, unless it goes through some major redesign, has a very stable PV/V PV/UU over time. Looking at the numbers above, these are PV/UU:
- September: 568 pages viewed by each user in September
- October: 580
- November: 495
The three key ways to interpret the 20% drop in October are:
1) Widget-dilution because of beacon (like you say above)
2) Facebook is losing its stickness factor, and each user is using the service 20% less than before.
3) Facebook went through a redesign in November that required less clicks (i.e. page views) to get things done.
The more likely culprit is a combination #1 and #2, and that can be a bit scary to advertisers (and investors) because the relative value of a Facebook user is going down. Can you say “tire kickers”?
Independent of all that, Facebook has a very impressive number of PV/UU and PV/Visit.
-Marcelo Calbucci
Founder & CTO
http://sampa.com