Current sentiment tracking technology is woefully inadequate

Listening tool vendors need to stop thinking about human emotions in binary. We are not computers. Positive / Negative sentiment is not enough to evaluate whether someone is likely to buy the brand. For example, if someone is cynical about a product, they could potentially be won over, if they hate the product they probably can’t, or the effort is better diverted elsewhere.

Yet under every listening tool I’ve used (Nielsen Buzzmetrix, Radian6, Visible Tech,  and Alterian SM2) you can only evaluate both of those as either negative or positive.

That last sentence was also to see which vendor eats their own dog food and picks up the brand mentions first. ;)

So rather than criticising something and not offering an alternate solution, I would like to propose an alternative. How about using sentiment tag clouds?

Automated sentiment is roughly as good as flipping a coin, so we tend to evaluate sentiment by hand based on a random sample. So instead of marking sentiment as positive or negative, there should be a free text field where we can enter emotive terms e.g., curious, uncomfortable, aware.

In public affairs scenarios you would be able to gauge whether the population is confused, angry, or cautious on a subject rather than purely Negative. You can’t do much with Negative.

So there. I said it.

Emotions in this post: challenging, hopeful