What does Facebook know about my off-Facebook activity?
Analysing how much Facebook knows about my activity across the rest of the internet (i.e. lots)
We all know that it can often seem like Facebook are listening to your private conversations and part of the puzzle is that they have a pretty good idea of what you like and what you don’t like.
Another piece of the puzzle is that they can observe what you do on sites other than Facebook, which all comes down to businesses sharing your data between them.
In this article I’ll be looking at what exactly they know about my web surfing and also asking the question: who else knows it?
You signed up for this
One of the files in my data download from Facebook was rather ominously titled “advertisers_who_uploaded_a_contact_list_with_your_information”. This contains examples of when an advertiser has uploaded a distribution/mailing list, shared that information with Facebook and my email address, name or Facebook account has been on that list.
My first thoughts: OK I can see there would be maybe 10–20 times I’ve signed up for something over the years and linked to my Facebook account just for convenience…there are 89 companies who have uploaded a contact list with my information. 89.
Not only that, but of those 89 there were 13 companies who I have never heard of before in my life.
Of the 76 companies I knew, there were 32 who I didn’t expect to have my contact information. The remaining 44 were all companies I know I have signed up to at some point in my life, so were expected.
The 13 I had never heard of included 4 property/estate agents, 3 clothing companies, a solicitors and a special effects company. None of which I have any knowledge of ever giving my details.
This isn’t to say it never happened however, unfortunately there is no information on when they first uploaded my contact information so it is hard for me to work out if they are legit. Some or all of them may have simply bought my contact details from another company.
On a related but different note, there is also information on how many times I have clicked on an advert while scrolling on Facebook.
Again, I expected this to maybe be 10–20 times a year perhaps, when I see an advert for something that interests me…72 times in the last 2 months.
I am extremely surprised at how often I click on adverts that I see on Facebook. Looking down the list of adverts I have clicked (I can only see the title of the ad not the company or context), there are many which I have no memory of clicking on but can imagine why I would.
I think a key explanation of this is the use of sponsored posts for advertising. I’ll often see a news story, article or video which I think looks interesting and not stop and think ‘why am I seeing this?’ - and the answer to that question is somebody has paid for me to see it.
The Gold Mine
The data set of off-Facebook activity is fairly extensive - over 2700 ‘actions’ since July 2019. Each event gives a ‘location’ (usually a website) and a ‘type’ which can be one of 21 different types of interaction. Most of these actions (2594; 95.7%) fall into one of six categories, shown below.
Some other types of event not shown below include ‘ADD_TO_CART’, ‘SUBSCRIBE’ and ‘ACHIEVEMENT_UNLOCKED’.
The 6 most common events recorded by Facebook about my activity on other websites
By far the most common recorded event is ‘ACTIVATE_APP’, which is pretty much what it says on the tin: for certain apps that I have on my phone, Facebook is notified every time I open them. The vast majority (~85%) of these 1051 events are from two apps: Medium (ironic) and Monzo (my main banking app).
Speaking of locations…there are 326 unique sources from which Facebook has received information about my online activity. 326 different websites, apps or businesses who have shared with Facebook what I am doing online. That is a staggering number to me and includes websites I have no memory of visiting whatsoever as well as lots which I simply didn’t expect Facebook to know or care about.
Looking at the list of locations, it really is easy to see how this list would paint a picture of me as a person.
All my favourite brands and websites are on there - from the local coffee company I buy from, to travel sites for specific destinations I visited on a holiday to Switzerland last year.
Anything I have a subscription with is on there. There is every news site I’ve visited, every restaurant for which I’ve booked a meal online, every app I’ve downloaded on my phone, every time I’ve booked a ticket somewhere or looked at a job advert or even just websites from when I’ve looked something up once (i.e. the inclusion of ‘bumblebeeconservation.org’).
I am amazed at just how much information there is here when looking at where I go online, without even looking at what I’m doing when I get there.
Below you can see which of these websites have given Facebook the most information about me (in terms of volume) over the past 14 months.
Number of events shared with Facebook from each location
As mentioned above, the top two here are mostly notifications that I have opened the app. The next two are mostly informing that I have viewed content, although LinkedIn have sent a large number of ‘custom’ events so I’m not sure what they might look like - possibly something to do with looking at job adverts and descriptions.
Unfortunately it isn’t possible for me to see exactly what is included in these events that Facebook can see, I’m only able to see what type of event it is. Even so, I want to give you an example of how you could make deductions about my life based on simply where I visit and when.
The graph below shows the total number of events shared with Facebook, by time of the day. The colour represents the most common location for each half hour of the day (i.e. each bar represents total events in that half hour, not total events from a single location).
The number of events shared with Facebook about my online activity, by time of day and location of origin.
This shows again that Medium shares the most information with Facebook about my activities. Bear in mind that I do not think I use the Medium app very often at all so my impression is that Facebook can see every single time I open the app or press a link within it.
It seems I’m more likely to check my bank account at lunchtime or in the evenings (when I’m most likely to be spending money). You can also see quite clearly what time of day I’ll be eating if I get a takeaway (through JustEat) - between 6:30pm and 7pm. So if you want to advertise food delivery or takeaway to me, do it at 6pm.
I haven’t visited Hostelworld much at all (there are only 9 events in the whole dataset), but you can see it is the most common source of shared activity during the night. This will be down to the fact that I’ve only ever lookup up hostels when I’m abroad and the fact of the time difference.
This is the fourth part of my series ‘What does Facebook actually know about me?’, where I use Data Science to analyse the 7,500 files of data Facebook have about me.
See Part 1 for a first look at the data, Part 2 for visualisation of location data and Part 3 to find out what Facebook knows about my likes and dislikes (and why they aren’t listening to your conversations).
The results are in
So it seems that the answer to the title question of the series “What does Facebook actually know about me?” is “lots”.
Hopefully reading this series has made you think about the kinds of data you produce (and how you often don’t realise you are producing it), as well as the data which feeds in to everything you see on social media.
Nothing you see on your Facebook feed is there by chance. Although this can be convenient, it is also dangerous - leading to the well known issue of echo chambers on Facebook, which can be summarised in one ‘rule of social media’: you will be shown more of what you like.
Remember that next time you see lots of a particular news story or trend and think it is ‘all over Facebook’ - your Facebook newsfeed is a reflection of you, not the world.
Thanks for reading, if you haven’t read the other articles in this series then check them out at the links above. Also follow me and my publication Data Slice if you like this format of article and data analysis.
Tomorrow I will be posting the final article in this series (which is also a sequel to a previous article I have written) - ‘4.5 years of a relationship in Facebook activity’.
Data Scientist working in the UK public sector. https://email@example.com