When it comes to behavioral targeting (using clickstream data), where is the fine line of benefit vs. “big brother”?

With petabytes of data available to ISPs, auction sites, search engines, and social networks, these companies are naturally seeking profits via targeting advertising based on the interests and inclinations of users as determined by clickstream data.

However, “targeted advertising” based on the recordings of queries, clicks and mouse scrolls is quite controversial as some privacy advocates think such targeting is invasive, while others worry about companies amassing–and storing–too much information on our behaviors.

Adding to the challenge, a recent New York Times article, “Web Privacy on the Radar in Congress“, August 11, 2008 mentions, “There is no broad privacy legislation governing advertising on the Internet. And even some in the government admit that they do not have a clear grasp of what companies are able to do with the wealth of data now available to them.”

Why is behavioral targeting getting such prominent play?
Annoyed with non-relevant advertisements, internet users are increasingly ignoring banner advertising, skipping flash programs, and deleting email offers. Behavioral targeting provides promise because based on the analysis of clickstream data, powerful applications are able to calculate customer affinities with fine precision and then tailor advertising according to predicted customer needs.

More relevant advertising suggests satisfied consumers as offers pitched relate more closely to web queries, companies can charge more for advertising, and advertisers benefit with higher click-thru rates and hopefully more revenues.

Companies that rely on web-based advertising for their business model see much promise in behavioral targeting. Indeed, for many social networking websites, behavioral targeting may be their only viable path to profitability.

MIT’s Technology Review, “Part I: The Business of Social Networks“, July/August 2008, highlights the difficulty of turning a profit for social networks like Twitter, Ning, Meebo and others.

The Technology Review article notes that, for example, when users search Google, they expect display advertising and even in some instances welcome paid advertising results because they’re shopping for an item, or looking to supplement information in a buying decision. For social networks, however, Jason Calcanis, founder of Mahalo.com notes that users are, “are busy in conversations and don’t want marketing messages.”

Thus the challenge for social networks is to provide value to their user base while creating a revenue stream that can sustain the business and pay back shareholders and/or investors. Increasingly that path is behavioral targeting of relevant advertisements in order to turn a profit. Let’s be clear–”profit” is not a dirty word. However, some companies are pushing the limits of what many would find acceptable in behavioral targeting.

Case in point, the same Technology Review article details the well documented travails of Facebook’s roll-out of the Beacon platform.

The article mentions, “Working with commercial websites like Blockbuster and eBay, Beacon tracked Facebook users’ purchases and displayed them to their friends. The problem was that users were enrolled in the program automatically. If a user went to, say, the Blockbuster site and rented a movie, that information was automatically sent to everyone in her Facebook network. Online petitions and negative press ensued, and the program was clumsily scaled back.”

Perhaps there was nothing inherently wrong with Facebook’s approach to behavioral targeting; however the lack of disclosure was particularly galling to users and privacy advocates.

So where’s the “fine line” where advertisers, consumers and companies win?

With the debate on privacy swirling, and Congressional action looming, I have questions for DailyFix readers:

* When it comes to targeting advertising based on customer behaviors (online or off-line) where’s the fine line between customer benefit and “spooky big brother”?
* Is there a way to target customers based on their affinities–without actually invading their privacy? Barneys of New York seems to be doing it right. What do you think?
* Does “the line” get crossed when a consumer is targeted across multiple sites vs. one site?
* Done right, will behavioral targeting be Web 2.0’s salvation (from losses to profitability)?

Chris Anderson, editor of Wired Magazine and author of “The Long Tail” penned a provocative article in the July 2008 issue titled, “The End of Theory: The Data Deluge Makes the Scientific Method Obsolete.”

Mr. Anderson makes the claim that in “the Petabyte Age”, it’s more important for companies (and the marketers within them) to identify and act upon correlation first, and worry about context later. For instance he writes, “Google’s founding philosophy is that we don’t know why this page is better than that one: if the statistics of incoming links say it is, that’s good enough. No semantic or causal analysis is required.”

And dismissing many of the sciences that attempt to bring us understanding of the world around us, Mr. Anderson notes, “Who knows why people do what they do? The point is that they do it and we can track and measure it with unprecedented fidelity. With enough data, the numbers speak for themselves.”

As companies collect more data about their customers, competitors and macro/micro environments, Mr. Anderson makes the claim that our approach to science (hypothesize, model, test) is becoming obsolete.

Science looks for causation. Scientists hypothesize as to why something works/reacts/behaves like it does and then attempt to build a model to represent reality. The goal is to use the model to test and learn, thereby gaining an understanding of a particular phenomenon.

Modeling is not only confined to the realm of physicists and quants. In the field of marketing, managers often work alongside statisticians, to build models that help predict customer proclivities such as items they might like to buy, or identifying customers who might churn to a competitor. Models can be tested and refined and with enough tweaking, models generally get more accurate over time. The output of those models can be used to piece together more complete customer profiles.

However, in the Petabyte Age, Mr. Anderson claims, “There is now a better way. Petabytes allow us to say correlation is enough. We can stop looking at the models.”

Let’s suppose Mr. Anderson is on to something. Mr. Anderson makes the claim that since modeling is often a poor representation of reality and tends to over-simplify things, we’re much better off with less explanation and more action based on the identification of correlation.
In a sense, he says, in an age of massive data, we’re better off with fewer discoveries of knowledge and understanding.

I’m not sure I agree.

Peter Atkins, author of “Galileo’s Finger“, says it much better than I can: “With the rise of the computer and its ability to handle huge numerical calculations of the greatest intensity, we are seeing a shift from analysis to numerical computation. (This is dangerous because) resorting to numerical solution can distance us from understanding.”

It is true that building models is an imperfect science, and cannot in all instances be 100% accurate–it is after all just a model!

That said, models help us test our assumptions and verify our forecasts. They help us test what we think we know, with the ultimate goal of improving our decision making in real-life situations where marketing budgets and return on investment are on the line. Modeling helps us piece the world together, and look beyond the patterns towards discovering “why” things happen as they do. And models help us transform reams of raw data into intelligence thereby helping us predict outcomes with greater accuracy.

While focusing on correlation can help us make better decisions to a degree, philosophically I am concerned with marketers potentially losing our interest in truly understanding what makes our customers, prospects and partners tick. For me at least, I want to know more than “my customers do this, or they do that”; I want to know “why“!

In all fairness to Mr. Anderson, every coin has two sides. There is a fine line between too much pontification and too little action. Indeed, there are many instances where it doesn’t make sense to dig deeper in understanding–where it doesn’t matter “why”, only that a given solution produces results.

I’d like to open this discussion up to the MPDailyFix community.

* As a marketer, do you want to understand and be able to explain the causes of why customers/companies/competitors do what they do, or is observing and acting on the behaviors good enough?
* Building accurate models requires tons of data collection, the proper analytical technologies and applications and of course the know-how. It’s not easy work. Do you agree with Mr. Anderson that for most decision making opportunities in the Petabyte age, correlation is enough?
* In a complex and busy world, where a marketer’s time is at a premium, will brainstorming, learning, and piecing the world together, become passé?