The Limits of Data

In my last post, I discussed an old saw of  data-driven marketing: more data are always better.  Today, I’d like to discuss another central tenet: marketers can reduce all information that matters into data.

Given a reasonably sophisticated marketing database, a marketer can record every interaction between customer and brand--every visit to the website (with permission, of course), every in-store purchase, every call to the call center.  Again, given the proper technology, the marketing database can make decisions based on those actions: send good customers first dibs on new offers, send reluctant customers offers to make them buy and so on.

Indeed, modern marketing databases generate so much data that many marketers start to see their customers as piles of data.  However, data have their limits.

With these beliefs in mind, I am currently tackling a different sort of marketing challenge.  To wit: how many rapes equal one murder?


Along with my other clients, I work with a not-for-profit called The Stop Abuse Campaign (follow it here on Facebook and here on Twitter), an ambitious group dedicated to wiping out all forms of interpersonal violence within one generation, or 25 years.  While previous anti-violence campaigns have focused on bullying or sexual abuse or elder abuse, the Stop Abuse Campaign sees all forms of violence as bitter fruits of the same tree and thus need to be treated in toto.  

As quixotic as their goal might seem, the group have some solid science behind them as well as a plan.  They intend to fight abuse not as a law-and-order issue or as a sociological issue but rather as a public health issue.  Obviously, the group has more to it than that, so I encourage you to visit their website for more if you’re interested.

My contributions so far have been to help the team build a marketing plan.  My current assignment involves creating a survey.  This survey will look at various forms of inter-personal violence at the state level and create a comprehensive ranking from best to worst.  

Those of you with advocacy experience have probably figured out the marketing purpose of this survey.  A survey ranking violence by state can potentially draw the attention of national news media.  CNN.com headlines such as “Is Your State Among the Safest?” practically write themselves.  The survey will also resonate at the state level; no state wants to be number one on this list.  Stop Abuse Campaign representatives can work from Augusta to Honolulu to drive legislative action.  A well-produced survey can create marketing dividends for weeks if not months.

Here’s where the limits of data creep in.  I intend to use sources such as the FBI’s crime statistics database and various advocacy groups to determine the prevalence of these crimes on a per capita basis.  Easy enough.  But what then?  To say that Louisiana has the highest incidence of murder per capita but that Alaska has the highest incidence of forcible rape tells us...what, exactly?  How does one compare to the other?  Which state has the greater problem?

At first, I thought creating rankings for each violent crime and then averaging the rankings might provide a satisfactory answer.  However, this approach intrinsically equalizes all violent crime: #1 in murder has the same value as #1 in rape, hence the how-many-rapes-equal-one-murder question.  This approach sure gets grisly quickly.

Ultimately, I’ve decided to weight the categories by costs.  I assume that each crime has both direct costs and indirect costs.  For instance, in armed robbery (considered a violent crime by the FBI), the value of the property stolen counts as a direct cost.  Indirect costs for armed robbery might include medical bills or lost work time for an injured victim.  Indirect costs might also include investigation of the crime and trial and incarceration of the perpetrator.  Hopefully, advocacy groups can provide me with the numbers.

While this approach will create statistics that hit home (“violence cost the state of Nebraska $X per person in 2010”), it still fails, I think, to convey the magnitude of the crimes.  How can anyone put a dollar figure on a child who no longer trusts anyone in his family or on a parent’s grief?  I personally would hate to think that a legislator would choose to tackle one form of abuse over another because, as Willie Sutton might have said, “that’s where the money is.”

I wish I had a pithy anecdote or a counterintuitive answer to this question, but I don’t.  All I can do, I suppose, is remember that no amount of data, no matter how good, will ever tell the whole story.