Technology experts and sentiment analysis software developers are claiming that we can now infer people’s feelings by analyzing big data. It’s based on what we say in social media. As direct marketers, we know our copy and content are most successful when we tap into the emotions and lift the feelings of our customers and prospects that motivate them to take action.
While I’m skeptical how sentiment analysis can be used without provoking consumer backlash, maybe we should reflect on this claim that software can predict people’s feelings.
In my last blog, I shared this thought-provoking quote from contemporary literature author Maya Angelou:
“I’ve learned that people will forget what you said, people will forget what you did, but people will never forget how you made them feel.”
Let’s take a deeper dive to see if this claim of inferring feelings from social media posts is not only possible, but if it’s smart. Or shameful.
A recent Wall Street Journal article on the topic of big data (“Marketers Want to Know What You Really Mean Online: Sentiment Analysis Aims to Decipher the Nuances of Social-Media Posts“) cites several examples of how it works. The article goes into more detail, but in summary, the process works like this:
- Software now can break down tweets and status updates to extract the literal meaning of what’s being said. This step is called natural-language processing.
- The software determines the emotion behind the statement. Was it written in earnest, or was it snarky? Was the emotion strong? That is: enthusiastic, angry or sad?
This technology has been used by pharmaceutical companies, hair product companies, food companies, political organizations, and even for the State of the Union address.
What the article doesn’t tell us is if the technology actually worked to increase engagement and ultimately sales.
The resulting analyses of sentiment analysis can be far from 100 percent accurate, but could be one of many resources used in your messaging strategy. Context, cultural and colloquial nuances, and length of message can lead software algorithms astray. The shorter the message, the more difficult it becomes for algorithms to correctly interpret intent. As we all know, people often misinterpret sentiment when reading each other’s messages (consider how many times you’ve read an email that was intended to be cute or poke fun, but backfired).
The CEO of a sentiment analysis software company is quoted in the WSJ article as saying that, “right when a person is first diagnosed with cancer, they are the most optimistic. So he advises pharmaceutical clients to target ads based on the emotion the person is experiencing in the moment.”
Is this smart, opportunistic, creepy or offensive? My mother is currently dealing with cancer and this feels to me like an example of cold-hearted marketers tapping into raw emotions and feelings of a vulnerable person’s emotional state-of-mind. I’m more personally involved, obviously, but using big data on someone just diagnosed with cancer feels shameful (and notice I’ve used the word feel or feelings three times in this paragraph).
On a different and more appropriately used level, sentiment analysis can be effective when monitoring social media for complaints. It enables marketers to more quickly address a complaint and correct a problem for the customer. This feels like a powerful and appropriate use of sentiment analysis.
If we take to heart Maya Angelou’s quote that people will always remember how you made them feel, taken across an emotional line in the sand, marketers would be well served to remember that the good feeling of the moment could quickly turn into a negative your customers and prospects will never forget.