FB not very good at profiling

[rcjordan]

Based on the crappy ads they're showing my 10-month old test account (which only shows age, gender, and an unused google voice phone number), FB doesn't try very hard to build a marketing profile.  Even their ip-location ads dropped out after a week or two.

[Brad]

They don't seem to read cookies much like the other sites and then serve up more of the same.

[rcjordan]

>cookies

Agreed. I'm not seeing any evidence of off-site tracking ...or even on-site, for that matter.  They aren't even trying to match ads relevant to the groups I joined.

[rcjordan]

(BTW, this topic came up because I have been asked by someone to give a 'privacy review' of my test account. He is dealing with FB-paranoid users while migrating a forum to a FB group.)

So, if the above is correct, the vast majority of FB's "security risk" is derived from the composite the USER gives them (or, perhaps, your smartphone's contact roster).  Employment, colleges, circle of friends ...the various & sundry dribbles of info -when made available for assembly, even crude algorithmic assembly- that's where the risk lies.

[Rooftop]

I suspect that a lot more ad impressions on facebook are served through their ad exchange rather than through their direct system.  The targeting that facebook allows can be layered on top of any audience data that an advertiser (or their SSP) is bringing to the table.

So, if I am re-targeting past visitors to my website I might choose to only then target those who came to my site, but facebook tells me are in a certain age group or have particular other interests.  If it is via an ssp they might also have data on all the other sites of their clients (or ones running the DMPs they use) that you can also layer on top.  Add it all up and it gets pretty specific (most of the same applies with other exchanges).   

What you might find is that a test account doesn't pick up that targeting.  If it is just being used on FB it might not be one that advertiser want to target.

[Brad]

Click counting.  I get a lot of: "Your friends X and Y liked Tide."

I think you are correct in that most of the profiling is supplied by the user and their friends.  The problem is locking down your own account for maximum privacy so you are not sharing every thing you do: post, like, comment, friend or share with the whole world. FB makes this hard to do and just when you think you have it locked down they change it negating part of what you thought was locked down.

>migrating from a forum

Might just be me but I would stick with the forum.  

[BoL]

Other people's contact lists are a problem for those wanting to avoid being part of the DB. FB and LinkedIn both definitely remember any uploads of contacts, even after 'deleting' them.

[rcjordan]

Being a total noob at FB and its dashboard, I used this tutorial to do the original gutting of the account
http://www.komando.com/tips/12299/step-by-step-guide-to-lock-down-your-facebook-profile/all

Being unfamiliar with the menu layout, it took me about 45 mins to dig down and turn off all that crap.

I then let the account sit for about 3 months to see if there were any noticeable upticks in email spam.  Nothing happened.

So then I went in and joined a group of about 1200 members.  Nothing happened.

So then I went in and joined a newly forming group which was forking from the larger group. Nothin happened.

Lately, I've purposely been clicking through on some individual members' pages. FB weakly tries the "you may know" stuff, but they seem to have even given up on that lately.

>stick with a forum

Agreed. But the problem is demographics. IF you want younger members to be active and grow the user base, you're going to have to go FB.  They literally scream when you try to make them use a forum format, hhh.  I'm certain that this is largely a smartphone access issue, but the ease and ubiquity of FB apps --*particularly* photo-handling-- gives FB a huge edge in member recruitment.

>uploads of contacts

This is why I think a throwaway G voice number is crucial.

[Gurtie]

From an advertiser side I find FB ads perform very well which suggests they are well targeted. But if you're paranoid or have private accounts they're pretty random and largely based on what nventory they need to get rid of and what advertisers have failed to lock down their targeting imho Exactly like G and Twitter do, FB volunteer to 'optimise' the advertising of the unwary big spender and people end up spending their cash very badly

>> other peoples contact lists

I am being stalked on every bloody social channel by a man who's only connection with me was that he replied to one of those round robin emails with a political comment I found so offensive I actually emailed back and told him so. I feel like Linkedin wants me to be best mates with Nigel Farrage  

[rcjordan]

>advertisers have failed to lock down their targeting imho

I've had a whiff of that sort of ad, but not very much.  A couple looked like poor advertiser targeting but quickly disappeared --I assume they fried their budget pretty quickly and burned out.  A couple of ads smelled of very cheap run-of-network crap.

>private

I'll add that in the 1200+ group I've clicked through to other gutted accounts (by percentage. I haven't been to but a few dozen, so the sampling is very small) --three or four were as barren as mine, which was surprising.  Paranoid sumbitches!

[rcjordan]

<added>

> re-targeting > test account doesn't pick up that targeting.  If it is just being used on FB it might not be one that advertiser want to target.

I missed your post, RT, sorry.

It's the lack of re-targeting that puzzles me the most.  (I used my public email address in the profile, btw.)  Yes, my laptop browser defenses may be formidable but I purposely surf FB and other sites with an anemically fortified chrome browser on the ipad2.  I use a fair number of sites (tools, home hardware) that are notoriously good re-targeters, imo, and I'm not seeing them.  That 1200+ group (auto maintenance) should be a natural for re-targeting if only based on the group's focus.

>age, gender

Maybe they don't want old men, hhh.

[Brad]

>>cookies

If FB is using cookie data set by other sites it is subtle perhaps with a delay.

I've been researching cars visiting lots of car manufacturer sites plus third party review sites. Focus on Subaru's.  After a few days I started seeing some Honda display ads on FB Web.  Today I saw a sponsored post about a Subaru safety feature on FB Android app.  I have not posted anything about cars on FB.

It could all be coincidence.  Could be auto industry just not advertising on FB. Could be subtle targeting, definitely not in your face like Yahoo where you visit one car site and suddenly car ads appear all over every Yahoo page.

Data set too thin to tell for sure.

[rcjordan]

...Or it's new.  Just yesterday, I noticed the first instance of "vague" re-targeting.  One was Amazon, I forget the other.  I say 'vague' because they weren't specifically focused  on a product, more on a very general product category.  My thought at the time was that it was a coin-flip as to whether it was re-targeting or run-of-network coincidence.

[JasonD]

I get retargeted a LOT on FB

[rcjordan]

[update]

So, while the ads are still for 'Date women over 50' a maybe-not-a-coincidence happened today.  I happen to own a very rare antique car, one of 20 or so left from an obscure manufacturer in the 1920's. So rare, that there are only a couple of articles on the web.  But I have been searching on the brand 'McFarlan' over the years and even recently around Christmas.  Today, on FB, I noticed a group being promoted as one I might like: Classic American Luxury Cars.  I click through and it covers everything --but mostly seems to be 50s & 60s.  One car reminds me of McFarlan, so I search the group. Bam! 6 weeks ago, someone started a thread asking about McFarlans.

Coincidence? Clairvoyance? Cross-matching?