KDnuggets Home » News » 2012 » Mar » Data miners think Target was not wrong to use analytics to find pregnant women  (  12:n06 | Next > )

Data miners think Target was not wrong to use analytics to find pregnant women

Was Target wrong in using analytics to identify pregnant women from changes in their buying behavior ? Only 17% of data miners think so. Increased accuracy of big data analytics feels "creepy" when a big corporation knows too much about us. If understanding without empathy leads to "creepiness", should we prepare for a creepy future?


Target The latest KDnuggets Poll asked
Was Target wrong in using analytics to identify pregnant women from changes in their buying behavior ?

Only 17% of data miners answered "Yes" (Target was wrong), and 74% said "No".

Regional breakdown shows that although US data miners were more permissive, a majority of European and Asian data miners still did not think that Target was wrong.

Region%No%Not Sure %Yes
US/Canada  80%  6%  14%
Europe  59%  15%  26%
Other  61%  21%  18%
In the above table, bar height is proportional to the number of voters.

Even though a majority of data miners, including me, did not think Target was wrong in using analytics, the reaction among general public was very different. The story created a feeling that somehow Target crossed the line, although it was probably difficult to pinpoint where the line was. Perhaps the best explanation I saw was given by Todd Essig in 'Big Data' Got You Creeped Out? Transparency Can Help. He suggested that

people are used to anonymity, and big data analytics creeps people out because it violates our assumption of anonymity - we feel spied upon by an uncaring corporation. So understanding without empathy leads to "creepiness".

Todd's proposed solution - transparency - does not seem convincing to me. How can general public (a majority of which, at least in the US, believes in astrology) understand the details of any algorithms or big data models ?

With Big data analytics becoming increasingly accurate, we may need to prepare for a more creepy future.

Selected Comments:

  • Greg S, Target
    Target was not doing anything wrong in predicting who may or may not be pregnant. However, they did need to be more careful in how they used that insight. This type of prediction is what we are all after, in terms of predictability, however our business partners are not always the best at making use of the information provided.
    It is great to see that we are getting to the point of truly being able to produce something so accurate. We now have to be careful of the 'big brother' effect of using that information.
  • Italo, Target
    The results of the poll are amazing from across the Pond. Luckily in EU there are privacy laws which ban these kind of activities so that this discussion would not even occur. Beside all this, data is an asset but so is your company reputation: making data produce results while destroying your brand equity value is not a good business idea.
  • Naveen, Was Target wrong?
    Target is not wrong, but as this is sensitive, the data owner should maintain privacy. And based on these report one should not do target marketing.
  • Doug, Was Target wrong in using analytics
    I don't think you could argue what Target reportedly did was illegal in any way.
    But they're probably not enjoying the free publicity.
    So was it WISE from a business perspective? Probably not.
  • Guy, Target is a for profit company
    It would be stupid not to use their data to increase profit.
  • Doc Muhlbaier, Target Analysis for Marketing
    The only reason for doing the analysis was for marketing. The NYT article was thin on how they did the analysis (probably some variation of clustering), but they were given the data by the individuals through their affinity cards. Those affinity cards provide us benefits (lower costs) at the cost of lesser privacy. There was no evidence that Target violated the agreement from the card.
  • Francoise, Poll and Pregnancy
    Is the right to use consumer information outweighs the right to privacy?? I do not think so.
    The next step would be finding people with AIDS, High blood pressure and so on. These could part of your medical history and the question would be:
    Knowing that data analytics would give a company clues about your medical profile, should they use it considering that a medical file is private??
    Of course pregancy is not a medical condition but it is a highly private experience (at least for the beginning).
    I can see you guys coming saying: "It was meant to better serve the customerette" and "the current situation where the dad finds about his daughter pregnancy is an outlier"...
    In a society where everything has to be kanban, I guess that there should be no room for a women to think about her pregnancy and receiving ads right away for diapers.
    Also, comments on that topic are coming from hard working man using analytics. It would be nice to see the breakdown of gender in that poll. :)
    Statistics, predictive analytics can make a nice un-human argument.


I agree that Target was not wrong with its targeted marketing. To counter a possible perception problem (some husbands & wives consider pregnancy a highly sensitive family issue), corporations acting on predictive information like this could anticipate negative perceptions & write proactive PR/Marketing blurbs that demonstrate how the company cares about delivering value to its trusted customers, with an opt-out for any future prize offerings.

When I was target-marketed by amazon last year [tracking & analyzing not only purchases but purchases w/ searches to decrease the error margin], I felt totally complimented by what they said. It added in a positive way to my already strong view of the company.

The question is not whether it is right or wrong identifying a potential purchaser. I strongly believe that it is a "must" for every organization in our era, regardless of the fact which data has been used to identify those prospects. Predictive analytics leads to smarter marketing, but the right equation is "the right messaging at the right time to the right person". In this case, 2/3 of the equation had been probably not very accurate. I know many pregnant women, who would love to receive Target prizes, if it would be coming in the right package at the right time and wouldn't even think about how they got it.

The company cares that it has a list of 1000 (perhaps even 100,000 ?) pregnant women to whom it can offer new products, but the public perceives it that company knows each one individually.

The question is whether the purpose is benign or not. Do we care whether Jane Smith is pregnant, or do we care that we have a list of 1000 probably pregnant women to whom we can offer goods? Those are very different alternatives, just as using atomic energy for a electric power is not the same as using it for a bomb. The technology is here, and it is unreasonable to expect the same privacy on the street, even a virtual street, that one enjoys in one's home. Because we access the virtual street from our homes, we have the illusion that we are still at home, not on the street; however, that is a counter-factual perception.

KDnuggets Home » News » 2012 » Mar » Data miners think Target was not wrong to use analytics to find pregnant women  (  12:n06 | Next > )