The other day my father asked me if I was happy and I responded “I don’t know.” He laughed out loud, as if to suggest it would make more sense for me to say yes or no, something definitive.
I didn’t know what he meant by “happy” and wanted clarification before I gave an inaccurate answer. Happiness can be so complex. How did I know if he was asking me about my current mood or my long-term satisfaction with life?
Like any researcher, if I don’t understand the question being asked, I’m reluctant to give an answer.
This rarely happens in an interview. Never happens on a survey. People will always provide an answer as much as meetings will fill the entire hour. Inherent in those answers is an assumption that each other’s definition of a given topic/construct are the same, take happiness. Guess what? This happens in Listening too.
Although Listening implies gathering naturalistic data, it’s subject to the same bias as the forms of research above. Perhaps because automation is involved, people forget you’re still asking a question, despite how passively you’re listening.
Each of your “alerts” is a boolean query, which is just like a good old-fashioned research question. It’s what you’re asking of the internets. It requires clarity. Sometimes the bias manifests as simply as a query that contains only your brand name, the way you know it and market it, as opposed to incorporating slang and nicknames. Sometimes it’s more complex– you think the “functionality” of your product (e.g. cellphone) has to do with its feature set (e.g. app market), but it’s really about perceived value or how easy it is to use (e.g. haptic feedback). “Quality” is another good example of where definitions can vary widely.
Someone once cited some research (that I cannot find) suggesting that when expecting parents discuss their anticipation of a new child, the male partner envisions the imminent child as a 3-year-old; the female pictures a newborn baby. Although unspoken, the parents are completely misaligned in what they’re bonding over throughout pregnancy. This is the exact kind of vivid image to think about when you Listen– something that conveys how differently people could define your topic of interest.
To be a social business, we need to smarten up on Listening, or more accurately, anticipating answers. Question clarity in Listening demands both query precision and comprehensiveness. Be sure you’re capturing the specific data you’re hoping for, and being exhaustive in the ways it could be conceived by others.
This, by the way, is also why Listening demands data integration. More on that next time.
Fascinating line of thought, particularly about setting alerts. When we enter our keywords for an alert, the mechanism only Listens to the separate words and not the context. So it returns results with those words, if not necessarily the meaning we’re seeking. Conversely, when we Listen to the results, the words take on specific meaning as they tumble toward the Listener within the context of the words that precede and follow them along with the historical context and bias the Listener brings to the effort. Because of all the contextual baggage people bring to a conversation, it’s amazing we can communicate with one another at all … If we, in fact, do.
Thanks for your philosophical thoughts, Ric! The more sophisticated systems have gotten pretty good at gleaning context using various cues beyond content words in isolation. This is a good point of differentiation for something that’s alert based as opposed to information retrieval based, the latter being preferable for good Listening.
Hi Katie,
Thanks for this great post. Another issue is that the “alert” is about a conversation, not a person.
Let’s take the example of a software solution that is great but doesn’t scale.
Each single client will probably go through the following pattern:
- it looks great
- I’m going to give it a try
- i’m happy
- i’ve been using it for a few months and it works great
- then say it doesn’t scale and I’m going to the competition.
5 positives, one negative and the net is negative.
What I want to show with this example is that listening and measuring sentiment only make sense if you can attach it to a real person.
In addition, all conversations are not equal. Whether the person that mention is relevant, is a potential buyer or a prescriptor makes a huge difference on the value of the comment.
I am amazed that people use “search engines” or “search engine like” techniques in research and expect to get valuable output and I, as you recommend would stress on the question.
Valuable questions usually take more than a search engine syntax. One recent from one of our client: “Who are the influencers in fashion, talking tech who are not mentioning us in the last 3 months?”.
Takes more than a boolean querry to get that.
Best
Best
Merci, Dominique, c’est vrai! Listening is a multidimensional process, in many senses of the word.
Hi Kate,
It looks like even my writing sounds French
.
Best