KDnuggets reader Lynne Mysliwiec sent me another story about Target data mining being producing unexpected results.
Congratulationson the front with the target. Inside it advertises their wedding registry services. (here are the pictures of the mailing).
WE HEARD YOU SAID "Yes"
The woman who received is NOT engaged, she is NOT living at her parents' home. Her mother called her to ask whether there was something going on in her love life that they didn't know about. Talk about awkward.
Lynne comments that Target said they had learned from their mistakes with the pregnancy packages, but apparently this is not so.
Are there some sensitive privacy areas that companies like Target should avoid predicting?
Given that there is a lot of money in weddings, can we expect that Target and similar companies will not try to try predict them? And also try to predict other personal events with big commercial implications?
What is next - a company learning of someone terminal illness and sending the spouse coupons for burial planning?
We are sorry about your upcoming loss, but you can save 25% when you plan the burial with us ...
Please let us know in the comments.
See here the discussion Data miners think Target was not wrong to use analytics to find pregnant women, prompted by New York Times story How Companies Learn Your Secrets, which gave an example of Target learning of a young woman pregnancy before her father did.
If I redesigned this piece, I would probably keep exactly the same targeting methodology and target, but CHANGE the lead in message. Perhaps something like:
"Love is in the air...each week, xx,xxx,xxx meet, xxx,xxx fall in love, xx,xxx people set a wedding date...."
Inside, modify the copy to speak in more general terms about the benefits of the Target Wedding Registry and the benefits that a hypothetical couple would reap -- without making it sound like "we think YOU are getting married" - instead, it's more like "We've got a service that YOU or YOUR FRIENDS might want to know about"
As marketers, we want to make sure that the right people get the message at the right time (down with spam!), BUT ALSO WITH THE RIGHT TONE AND MESSAGE. And that is really where Target misses the target for this marketing piece.
I'd rather have a dollar today than a dollar ten years from now. This marketing activity has put short term dollars at risk - it will scream through social networks (just like this one)
data miners and their management should realize that long term profits beat short term ones ... although such data mining leads to incresed profits in the short term, it will likely lead to decreased profits in the long term, when customers churn to other businesses where their perdonal lives will not be spied upon !!! is not target far too savy a retailer to continue in this vein, where fleeing cusomers will likely lead to fleeing profits ???
this is a significant example of that new dimension for invasion of privacy: not only being identified in a sample of individuals, but also having ones personal life continuously spied upon ... there is also another recent example of a major player violating this same dimension for invasion of privacy which was discussed on the evening news -- can you guess whom i am talking about ???
DO WE NOT NEED TO REALLY THINK WHEN AUTOMATED ANALYSES TAKE US SO FAR SO FAST THAT OUR FEELINGS DRIVE OUR ACTIONS, AND SERIOUS THINKING IS SIDELINED ???
Or hear the same question from one's spouse . . .
Totally agree Guest - Am totally OK with the predictive modeling, am totally OK with the application of data mining, but the marketers at Target are off base when they create a marketing message that takes probability (even a high probability) and treat it as certainty. If they send the piece and they're right, they risk creeping out the recipient. If they send the piece and they're wrong, they look like dopes AND they cause unintended fallout. "Honey, is there something about your love life that we should know about?" is not really something I would like to hear from MY mom. :-)
There's absolutely nothing wrong with the statistics here or how it was used - the problem was in the marketing. The the piece instead said "Spring is in the air..." and showed a lot of pictures of June brides, no one would haver noticed - we see untargeted pieces like this in the magaziner rack beside every supermarket check-out line.
The mistake here is not in the intelligence, it's in how the intelligence was used. We are free to use the highest level of scientific analysis available, so long as we don't go out of our way to tell our customers that we are working very hard to send them only those offers they are most likely want to accept instead of bombarding them with every last thing we have for sale. This is done by wording the advertizing piece correctly.