Big Data coming to a store near you
|Laurel Papworth talks about Facebook, Big Data and Social Media|
Howdy, I wrote this email a while ago but didn’t send it. But we have new News! Facebook has brought in a tool that let’s you advertise to, say, women in Canberra who have bought Cheerios, bananas and life insurance in the last 30 days.
They do this by working with big data brokers Acxiom, Datalogix and Epsilon, who in turn have done deals to get frequent flyer, supermarket rewards and mastercard purchases into a database and match it to Facebook users.
It means when you buy some vegemite you’ll get an ad for toast. Or something like that. The Facebook deal with Quantium (Woolworths), Axciom and Epsilon for Australian frequent flyer information, Everday rewards purchases and Mastercard etc was done a week or two ago (AdAge) and the datasets should be available to create Facebook ads against those buying behaviours soon. I heard a rumour it was August but hey, it was a rumour 😉
How do you feel about your purchases at the Supermarket, or flights you take, being made available to advertisers to tell you of new or competing products?
To understand Big Data, it’s a great idea to look at the four V’s plus a few more:
- Volume (1.1 billion people on Facebook is 300+ petabytes of data). That’s a lot of status updates. Care to read them all? I can’t keep up with my Twitter timeline without tools!
- Variety: being able to mash up unstructured metadata like location the photo of the new shoes were taken, the date, the device (iphone camera) with structured data from databases like credit cards and loyalty programs. New home prices with crime rates with school scores. Data is open and available and running free.
- Velocity: if you have done my measurement courses you know that velocity is a key metric. How much and how fast? Big Data can mean Too Much Data 😀 But not just the speed of a tweetstream (we get 16,000 comments during an Ad break) but also speed of analyis.
- Veracity: what is the truthiness of the data? if you run a sentiment analysis over thousands of updates/statuses can you really trust the “fully wicked sick” and “totes awesome” to mean what you think they mean? Data Trust is an ongoing issue – how many of you sign up for stuff with a fake email address? ^^
- Variety: how do you cross check unstructured data? is it possible to mashup geo-location (phone is in Bondi) with a purchasing behaviour (buying breakfast, newspaper). How about crowdsensing – Foursquare tracks your friends and can predict where you will be with them in 30 minutes even if you haven’t met up yet.
- Value: are you getting value out of the insights. For example, Facebook Lexicon clumps words together and then delivers content into newsfeeds based on Source and Engager and those keywords. Oh and Last Actor (the last 50 interactions) Knowing this helps you how exactly? Let’s not even start on data security, gaming algorithms and so on. Modeling, prediction, management.
- Vision: I’m adding this one in as I think it’s the V that marketers want. To predict behaviours not only on the past behaviour of that customer but also based on the customers tribe & rituals. If 80% of a customer’s friends have become pregnant and that customer’s searching behaviours indicate nest building, well, send ‘em some help. 😛
I know this might be a bit esoteric for early in the morning, but we need to address the what Big Data is before the elephant in the room – privacy and ethics – can be sorted out. I’m presenting on Big Data and Human Resources at the AHRI conference in a few weeks. I wonder what they’ll think???!! When I presented Big Data and Social Media to the National conference of Chief Purchasing Officers last month, they were astounded!
Hope you find this interesting!