Evan Schuman's StorefrontBacktalk

Extensive analysis of a consumer’s Web interactions has been used for years to try and target pitches more effectively. But new research suggests that such analysis may pale in comparison to the next wave, where every digital comment made by consumers anywhere—in a product comment, an IM, on a social network site, in E-mail and via, exchanges with a live chat tech support person, coupled with Web traffic analysis—can be mined for hints as to emotions and other thoughts.

This story on an O’Reilly site is the latest look at what can be done. The science and technology of it is really not that far-fetched. The fact that so many consumers—especially younger consumers—today share so many of their thoughts and private moments in so many public settings alone would allow even a casual observer to learn quite a bit about someone.

Consider this scenario: You receive an E-mail about a prospective new piece of business from someone who you barely know. The first step would be a general Web search to find the most obvious details. The second might be searching a business site—such as LinkedIn—to learn more about this person’s professional history and perhaps get a peek at who they are connected with.

So far, this is pretty clean. But one offhand link sends you to that person’s Twitter, Facebook or MySpace posts, comments that are decidedly more personal. There you learn about a recent death in the family or perhaps a divorce or even that the person recently broke a leg.

At this point, having limited your search to easily available public postings, you have some good insights into this person—insights that could greatly help you craft your approach. Maybe, if you think it through, it might even give you a strategic edge compared with someone who didn’t bother to check.

Taking It To The Next Level

It’s been pretty tame stuff at this stage. But let’s say you happen to work for a large retail chain and can access that customer’s purchase history. Suddenly, you know an awful lot more about this person. Throw into the mix any customer service inquiries and complaints, plus instant chat exchanges with tech support and maybe 40 minutes’ worth of voicemail messages the consumer has left over the last year.

Or, what if you happen to work for Google and that consumer happens to exclusively use Gmail? If your quest is to learn about consumer emotions—and if you have the right access—the temptation can be extreme.

For emotional clues, voicemails and IM exchanges are golden as consumers tend to be less careful about phrasing during those encounters.

But to what end? What good does it do to understand a consumer’s emotions? Most consumers have very specific shopping reactions to happy and sad feelings. Depression might trigger a desire for certain products (perhaps chocolate or a “comfort food” video or piece of music) while happiness or a celebration would provoke very different desires (often higher priced items seen as a reward).

Different Consumers React Differently

The problem is that different people react very differently to emotions, and the products they are drawn to during those emotional extremes also vary wildly. That’s where things get interesting. By overlaying all of that emotion-revealing data over a long time period and matching it against that consumer’s actual purchases during those periods, it’s possible to chart a shopping-desire-to-emotion chart for that specific consumer.

Armed with such charts, the software could track all of the emotion-clues for 5 million consumers, predict when each is going into emotional extremes and send them customized product pitches based on how their chart says they react during such times.

What consumers receive is nothing bizarre: A pitch from Amazon or Borders or Walmart for a particular kind of product. But what they won’t likely know is that the pitch was prompted by an angry voicemail about tech support or a MySpace posting the software thought “sounded sad.”

Technologically? This is quite do-able. Psychologically sound? If the software is done properly, yes, these predictive packages can be frighteningly accurate. But here are the big two questions: What about privacy and morality?

Privacy And Morality

Those are two words that no CRM advocate wants any senior exec to ever bring up in a meeting. But let’s consider them anyway. Privacy. Like it or not, privacy is generally set by an informal majority rule vote by consumers. Something that would have been considered a major invasion of privacy in 1960 would not have been a privacy issue in 1990.

Many of the executives making these privacy decisions have very different privacy values than their older and younger customers. It’s most troubling when companies have age-diverse customer populations. Does a firm default to the most conservative or the most aggressive position?

Even more amorphous is the question of morality. Put another way, is it inherently wrong to use software analytics, purchase history, voicemail messages and Twitter posts to figure out if someone is sad and to then try and take advantage of that fact by selling them things they’re drawn to when sad?

I’d personally have to say that it comes down to how it’s being used. If Amazon sends me an E-mail pitch for a book that is exactly what I’m in the mood to read right now, has the company injured me? Will I likely feel that it has betrayed me or that it’s done me a service?

If a retailer uses this kind of information with restraint, professionalism and compassion, it’s a slippery slope—but it can probably be kept to the right side of the moral question. In a down economy, however, when making more sales could be the difference between having a job or not, such restraint can be a very hard thing to keep up.