The real evidence of an outcome came later. In April 2011, the company reported a first quarter net income of $1.2 million, up 33 percent from $885 million the year prior. The baseline analysis alludes to the idea that sentiment may be predictive. In this case, maybe. American Express had just moved aggressively into online commerce.
But there were several other factors in play for the company. It had settled litigation with MasterCard and Visa. It had launched several premium products. Cardmember spending was up 17 percent.
One year later, the story was much the same but not nearly as strong. Cardmember spending was only up 12 percent and net income only grew 10 percent (without the benefit of settlement payments). And according to Digital: MR, it was still the most talked about bank on social networks.
It also carries a great introductory APR, but its regular APR is not nearly as competitive. And its stock performance, which is among the top ten, does not reflect the same exuberance as its conversation points.
Sentiment analysis can be useful, provided it is not a distraction.
Personally, I'm a big fan of sentiment analysis. It can be used as a benchmark for communication efforts. But marketers ought not mistake sentiment as the end all in marketing measurement, making decisions that upturns mean "do more of the same" or downturns mean "do less."
In fact, in digging deeper into the American Express sentiment, we found that much of the buzz comes from a smaller group of passionate advocates than, let's say, Citibank, which has considerably more reach from a broader base of people. My only point is that not everything is as it appears to be.
My second point is that if your marketing team is only using sentiment analysis as a means to track positive impressions and share of voice, then the research time is probably being wasted. There is a big difference between listening "for you" or listening "for an industry" and really hearing consumers.
In reality, only about 20 percent of research investment ought to be tracking impressions or attempting to snuff out complaints or improving positive:negative sentiment ratios. There is something much more important to consider: who are these people anyway?
The more you hear from consumers about everything else, the better your communication.
Instead of dropping every dollar on sentiment analysis, there are much more interesting things to learn about any particular segment of the population you might identify as customers or prospects. And none of it really has to do with your company.
What kind of music do they like? What were the last three movies they saw in theaters (and liked)? What were the underlying messages, if any? What kinds of books are they reading? Are they rigid in these tastes or more eclectic? Would they rather go to a fancy restaurant or buy new clothes? Is there a difference between what they buy and what they like? What kind of political leanings do they have? Are they aggressive about it or not? Do they cook? Are they struggling or secure? So on and so forth.
While you always have to keep in mind people feel this level of marketing research seems creepy, the takeaway is that marketers have a better chance of building a relationship if they hear what people are saying instead of listening for the latest mention. Or, in other words, marketing insight might be more powerful than marketing research.
It's a valuable lesson from old school copywriting — you communicate to a person, not an audience.