Walgreens and AT&T are using your personal data to match you with the right customer service representative.  And they are using something called the “break point” to walk right up to, but not over, the line where the customer will leave.  They do this through analysis of tone and pace of speech.

It’s only at this modeling break point that the customer service representative is authorized to offer the price package or customer service fix to bring the customer back on.

This rewards what behavior?  A dangerous game of chicken.  Obfuscation on the part of the rep resulting in increased ire on the part of the customer and if and only the customer gets mad enough in tone and rhetoric will the machine call “break-point” and let the representative meet customer needs.

It also rewards hot heads and jackasses – those customers with unreasonable demands from the jump will get their back scratched quickly.

Machines are very fast and increasingly, very good at a number of tasks.  But, they still largely suck at fully analyzing and instantaneously and intuitively understanding tone, word choice, idioms, slang, satire and a whole host of other nuances of speech that we as humans possess.

Replacing man with machine in this instance likely has all downside, no upside.  You have a far less perfect, less attuned, less nuanced, context sensitive system that simultaneously undermines the autonomy, competence and connection for the customer and the representative.

In this telling of the customer service story, there’s a silver lining: they draw you back from the brink.

As nonprofits, can we say we are doing any better?

Here’s a verbatim comment from a nonprofit donor:

DO NOT SEND ME ANYTHING IN THE MAIL.  PLEASE DO NOT.  EVERY SINGLE TIME I DONATE I ALWAYS GET MAILERS.  STOPPPPP!!!

All caps and the sheer number of Ps and exclamation points means I’m thinking you don’t need an AI to figure out this person is near, at, or passed their break point.

If you are not collecting feedback, this person is having these feelings without telling you.  They pass their break point and they are lumped into the mass of lapsed donors who lapsed for who-knows-what-reason.  In fact, if you aren’t collecting feedback, you are probably mailing this person who said not to send them anything in the mail to try to get them back.

Once you are collecting feedback, the next step is to make sure you are asking the people who aren’t often covered.  If you are asking for thoughts post-donation, for example, you aren’t getting the people who hit their break point during the donation process.  You’ll learn what to do to avoid inconveniences for those who do complete the form, but not find the things that are fatal.  This is called survivorship bias when you only draw data from success stories.

If you have a robust listening program, you can now dedicate yourself to pulling people back from the brink.  The Canadian Red Cross modeled which of their F2F donors was most likely to lapse, then called them to recapture.  The calling program has over a 100% ROI.  You can see what they asked, how they asked it, and how they set up the program here:

Bottom line: this new world of bringing donors  to the brink of abandoning your organization sets a low bar for donor service.  But we want to be focused on jumping over it rather than limbo-ing under.

Kevin