Using AI to address backhanded compliments in service recovery...
AI doesn't have to be that complicated you know. People complain that it's all pilots and concepts and too expensive to scale. Meh!
Start simple. That's our motto.
A wise man taught us a long time ago that all good ideas come from hypotheses, and testing those hypotheses. That's how our new Sentiment Mismatch feature was born at JabFab.
Upon procuring a shiny new AI and Natural Language Understanding engine for our experience management system, we sat in a room one rainy Thursday and started throwing around hypotheses. After using up most of our pink Post-It stickies, one of our product managers came up with a genius idea. Genius in its simplicity.
"What if we made the hypothesis that a LOT of people hate being negative and prefer to hide their negative feedback in a positive feedback event?."
"Hmmmmmm, go on...".
"Well, what if we use our AI engine to read all the positive comments we received from people during service events, and look for negative comments too? I'll bet that 10% of our thumbs up people are leaving passive-aggressive backhanded compliments!".
He wasn't far off being completely right. Hah. Who knew?
This hypothesis led to another hypothesis that negative feedback thumbs-down people were also likely to be offering up passive positive feedback. Again, true.
One thing led to another, and the next thing we found ourselves doing was to use the AI / NLU engine to analyze this phenomenon on the fly, while the patient (customer) was still in the building, giving the services coordinators time to act and affect the overall outcome of the experience. Soon it just became an embedded feature of our ExperienceAI toolkit in the system. And yes, someone got promoted.
From the simplest of ideas was born a highly scalable and pragmatic use case of AI and NLU in motion. No elongated pilot. No expensive investment. Just a simple and well executed idea.
Moral of the story. Start simple, build from there. Bittersweet value from the most unexpected places. Come and test drive JabFab for yourself and chat with us on your hypotheses!