Netpromoter scores minus 22 with Netpromoter users

As many will have heard me preach from the pulpit of the conference circuit, I don’t think asking someone if they intend to recommend is anywhere near as useful as asking “how many people have you recommended?” or “were you recommended?”

So it was interesting to see at the SOCAPie conference last week that Ipsos MORI have done some basic research to test Reicheld, Bain and Satmetrix’s contention that the one metric you need is to ask on a scale of one to ten “would you recommend to a friend or colleague?” and work out you Netpromoter score by taking the volume of 1 to 6s from the volume of 9 and 10s.

What MORI showed is that
•Correlations of all the attitudinal and customer behaviour variables were modest (say vs do).
•The single question didn’t correlate with business growth and that it performed “universally” worse than multi-variant scoring – that’s measuring several things to us non statisticians.
•What’s more intention to recommend as a single variable wasn’t as good a predictor as retention or share of wallet.

MORI challenge the robustness of the Riecheld and Satmetrix work and its claims for an ultimate question whilst acknowledging the favour done by getting “customer” firmly into the board room language, if not the psyche.

SOCAPie members who use Netpromoter had done some research of their own. They scored Netpromoter using the "ultimate question" and found it somewhat lacking with a score of minus 22. They counted plus 30 as average and plus 70 as good in the Netpromoter system.

As some wise gent once said "there are millions of complex problems with a simple answer that is wrong". And I naievely thought the ultimate answer to the ultimate question was 42.....