Facebook 'likes' and adverts' value doubted
bbc.com
bbc.com
It's not directly about roi or fraud or anything like that, not fundamentally (though obviously no one will advertise long term with a poor roi). It's more about vision. Adwords & Overture had some sort of vision about advertising online so do lots of (mostly unsuccessful upstarts). I really do think that with all of their problems they did a great job. Adwords is a great ad platform. It's more meritocratic. Scales down nicely for small/local businesses and is genuinely useful to the people seeing/clicking the ads.
Facebook on the other hand seems content to ride the buzz of companies keen to figure out this new medium. They don't seem to be taking advantage of it to implement some solid idea.
It seems like they have no interesting ideas about how their contribution to the communication-entertainment-advertising ecosystem looks. They don't really seem to understand themselves how/why/what their system (which is inevitably a new kind of advertising) is supposed to be working.
The @facebook email address mess is a kind of analogy. Think you can improve on email/communication? Great. We need it. Think Facebook is a unique position to do so? Great. I agree (at least for some people). Figure out a way of doing it that people want. Don't trick people into something that feels like its for the company instead of the consumer.
Now I want to state for the record, I actually like Facebook, I don't have an axe to grind, and I really believe that MZ's intentions are truly honorable. I would like to believe he knows nothing about this, but I'm not naïve - he owns 57%, and he's a very very clever man. But I am disappointed.
So to explain...
Because the BBC has a wide non-tech audience some of what was found was not fully revealed mainly because it's so difficult to explain to the general public the tech behind this issues raised.
The main point was missed both by TechCrunch, and many commentators on Twitter etc. and I guess general fans of Facebook. The point being that at first the ads were targeted to UK and US people. Hardly any clicks were reported meaning that advertising with this targeted criteria is pointless, and in fact more pointless than expanding the reach to countries that will artificially boost your return. The clicks/likes though high in number will never bear fruit, that much is not disputed by the BBC. But it begs the question if both "settings" are useless then what's the point in spending any money at all with Facebook?
Before I get a barrage of hate mail remember I had access to the real data. The BBC describes exactly how you can replicate the experiment for yourself, and I guarantee you'll find the same results. What's more you'll discover the same profiles are liking your fake page the same as they liked the BBCs virtualbagel. Take it from me.
What's more the list of Likers contained people who appear to spend their lives clicking likes. These people are clicking between 50-100 likes per day, all day, every day. In the 10 or so days in July many had racked up literally 300-1500 clicks. It doesn't take a genius to work out these people are being paid to click on ads.
Remember we're talking third world countries where it makes economic sense to pay someone a-dollar-a-day to click on ads all day, when the return is so massive.
The question is who is paying? Who really stands to benefit? Logically it just can't be a third party. This is out-and-out click-fraud.
In addition to all this, and I can't reveal too much (because it's still being investigated for later publication) is that there appears to be evidence that the process is also subject to some automation from within Facebook. We have comparative evidence that shows how click automation would look if it was carried out by a third party and how it would look if it was carried out inside Facebook. Unfortunately for Facebook the evidence is showing that again it's driven from the inside.
I'm actually quite sad, because it doesn't have to be this way, but from what I have analysed it is, what it is. If you don't believe me do the same exercise and look at the data.
I've been singing this tune for a bit. In my experience, this is a big problem. If Facebook is saying this is the actual number, then consider is much higher. Facebook, as well as every player in their ecosystem, is economically incentivized to inflate the usage numbers. It's why Facebook is "worth" what it's worth, and how every single application on the platform is measured for investment. Games are literally ranked by their daily active users.
I spent some time combing through the data of an oldish FB game (as an employee), with about 300k MAU and 150k DAU. We figured that about 40% of the accounts playing on our game daily were fake, created solely to take advantage of any mechanics where it was advantageous to have more friends. Some people would have 10 or more accounts. Now there are businesses sprouting up in countries where labour costs are cheap who sell "likes"; you can buy thousands for less than a hundred bucks. Mostly fake accounts.
And this is only the blatantly fake accounts. What about the people who actively, honestly use Facebook, but have lied about minor details that are important to advertisers? Things like their age, sex, name, jobs, location, whatever? I did this myself.
And yet, the value of the advertising (and Facebook directly) is derived from these measures.
>"One, going by the name Agung Pratama Sevenfoldism, showed his date of birth as 1997 and said he had been a manager at Chevron in 2010.
>Mr Tinmouth said this seemed "unlikely".
I'd say.
As far as disproportionate responses from the Phillippines and Egypt, if you don't limit the audience you show your ad to, you're gonna get bad results everywhere, not just Facebook.
http://news.ycombinator.com/item?id=4238075
Moral being, target your marketing in trusted markets and verify any feedback for validity and if you too find that flase people are comming as a result of the marketing, then it is perhaps time to ask for a refund. I'm sure even facebook don't want nor wish people to lie and cheat to give a false impression and in that they would be more than happy to refund your expences. Especialy as you are checking simple details that they should of picked up early on by them.
Just look at the http://www.freelancer.com/jobs/Facebook/. How do you think those jobs are possible otherwise?
That being said, it doesn't work as well as it did in 2008-09, so I see this as news 2 years old.
(2) and (4) have the monetary incentive.
I worked with the team that fight spam, scams, and the fake accounts that distribute them when I started at Facebook, and they expend a lot of time writing code and building machine learning models to detect fake accounts. There would be no fake accounts if there wasn't a massive financial upside to it, and so they do the strangest things to try avoid the detection.
You can read a bit about what the team does at http://research.microsoft.com/en-us/projects/ldg/a10-stein.p...
The only way to spoof that is to expand the targeted audience.
The automated ad clicks that I refer to is demonstrated by a gaping hole in the data consistency, which would not otherwise occur if the action was taking place outside of Facebook's own servers.