post represents my opinions only. available under a cc-by-3.0 (unported) license.
In many ways the 2000 “dot com” crash was a misnomer – there was nothing fundamentally wrong with the technology. However, what was wrong was the model used to pay for the technology, which was primarily either display advertising revenue or (more commonly) venture capital advanced with the expectations of returns based on display advertising. There were expectations around the revenue generatable from on-line adverts simply because they were online that were just not realisable.
It is difficult to understand, more than 10 years on, quite why people thought that advertising would be more effective online than elsewhere. Advertising Effectiveness is a young science, and much is advanced (and then debunked) based on nothing more concrete than theorising. But many major advertisement placement conglomerates (Google, Microsoft, Apple… sometimes incorrectly referred to as service providers, hardware manufacturers and search engines) are focusing on one particular theory – the idea of personalised advertising based on user data.
(apologies for linking to a bad Tom Cruise film of a decent Philip K Dick book)
Google is widely reckoned to receive 99% of total income from advertising – around $28 billion in 2011. Facebook partnered with Microsoft in order to gain an estimated $3.8billion via advertising in 2011. And Microsoft itself (in partnership with Yahoo and AOL) are determined to break in to this market despite losing £2.5 billion a year on providing online services, barely breaking even. However, both Google and Facebook have seen recent valuations of substantially more than £100billion dollars, and Apple (a major provider of mobile-targetted adverts) has is valued at $362 billion (more than the UK national debt),very recently holding more cash on deposit than the US government.
Large amounts of these earnings, and much of the assumptions made regarding company values, are based on revenue generated by personally targeted advertising drawing on user data. The data these companies hold on our online (and increasingly, offline) activity represents their most valuable asset. Twitter, a company that doesn’t even yet have a revenue model, is valued at more than $12bn (£8bn) simply on the value of the user data it holds.
As we reach the bursting point of the bubble we see increasingly crazy activity. Only today Google launched “search plus your world”, using recommendations on social media (initially it’s own Google+) to serve you search results, and thus advertisements, based on the opinions of your online contacts. The “freezing out” of Facebook and Twitter is not the issue here, it simply breaks search. It relies on your G+ account being well-managed in order to provide you with tailored results. Forgetting that if I want the opinions of my online contacts I will most likely ask them, and most likely disagree with them too.
Tesco, the UKs largest retailer, does not allow you to set up an online account to make purchases without being signed up to their own “Clubcard” user data collection scheme. Simply and startlingly if you don’t give them your data, they don’t want your money.
Your user data, goes the theory, allows adverts to be specifically targeted to you. Should you buy, for instance, a decent bottle of single malt, you would be likely to receive advertising for other whiskies and spirits. And as a “single malt drinker”, you personal data becomes valuable to other companies selling other products and services that other “single malt drinkers” buy. Online, this is easier and quicker, due to data stored by your web browser such as cookies, and search and purchase histories stored by search engines and shopping sites. This makes personal advertisment serving quicker and easier (“looked at data projectors online recently? You’d love to see these ads about data projectors. Never mind that you just bought one, or were researching for a friend…”). Analysts estimate that targeted advertisements drawing on your user data (on and offline) are twice as effective.
To me, all of this seems to be based on a reversal of what I have previously termed “one of the odder beliefs that our culture seems to have developed about markets” – the idea of market efficiency and the rational consumer. Advertisement targeting draws on the idea of our observed behaviour presenting a coherent and realistic picture of our desires and needs. Bluntly speaking, it doesn’t. My past spending behaviour likely bears no relation to my spending currently or in the future – circumstances change, tastes change, opportunities change.
All of this sails wonderfully close to Stephen Downes’ recent post on learner data. He argues, and I would agree, that data is not “wrong”, but it is used in ways which are wrong in that it is used to generate conclusions that it cannot support. As Brian Kelly points out, the NMC Horizon 2012 Preview Report (you can’t read it unless you give them your data!!) sees Learner Analytics as a key 3-5 year trend for adoption in HE. And educational technology companies are putting serious money behind the idea.
You can see the effects of this cultural mindset even see this in UK funding policy. Students are expected to make decisions regarding their place of study (or indeed, whether to study at all) based on the Key Information Set [KIS]), an abstracted and highly summarised set subset of user data. This data, it appears, can fix broken markets.
To conclude: estimates of the value of user data are everywhere, and probably overestimate the actual realisable value. True in education and in wider e-commerce. Adjust your investment portfolio or educational predilections as you see fit.
At the moment, I suspect a lot of online marketing is based on information captured about the sites you have been visiting (and maybe terms you have been searching for) and responding to that. Would you be happier if transaction data could also be passed to the adserving cookie monsters (eg the fact that you had just bought a plane ticket or holiday, so maybe wouldnlt appreciate continued cheap flight/holiday ads?)Where recommendations are based on ‘missing item from basket’ estimates based on correlations of what was bought with what, would you prefer that instead the transaction data/fact that you had bought X was communicated to the cookie setting ad monsters, and they changed their ads to ones tuned by identifying where you are in some chain of likely related sequential purchases? So you’ve bought a flight, maybenow you need a local taxi firm ad?