sentiment analysis is making you dumb
December 8th, 2009 • posts i've written
Tracking sentiment sure is all the rage these days. And why shouldn’t it be? It satisfies two crucial human needs: the need to know how much you’re liked by others and the need to look at more charts of how much you’re liked by others.
The idea is to dredge the conversation from the internet, apply some fuzzy natural language processing, and return a breakdown of conversation into degrees of positive/negative sentiment. Of course, it most often neglects the social context of that conversation, botches language, and fails at understanding the impact of specific conversations.
But worst of all, with the pretty charts and happy colors (green is good, red is bad!), sentiment analysis reinforces the behavior of mindlessly counting up high-fives and ignoring everything else.
The truth is, how a person feels about your brand is more complicated than you’d like it to be. The amount of time a person invests in being vocal about a brand is proportional to the extent that brand helps that person define their identity among their peers. In order to understand how someone feels about your brand, you need a tool more sophisticated than an abacus.
And when it comes to serving insights, these tools really fail; most offer the equivalent of keyword dumps and wordle graphs. Some attempt to deliver more, but there’s only so much valuable information you really pick-up when scraping the profile page from a social network.
But again, my beef isn’t with the technology, it’s the use of it. Sentiment is just one more mumbo-jumbo number brand marketers toss into an Excel sheet (when it suits them) to say, “job well done!” and move on.
If you want more reasons (five to be exact) to discredit sentiment analysis, there’s a post for that.
Related posts:
One Response (add your comment)
donate your two cents
NOTE: Anonymous and/or hateful comments get deleted. My blog, my rules.
I agree with you. There is no really useful way of measuring conversations to predict ROI or any other spurious connections. Dell makes money from Twitter because it is selling stuff simple as that. The rest is nonsense at this point and as useful as counting the number of waves on a beach to predict weather patterns – there may be a link but its not worth the hassle and is certainly not predictive of purchasing intent.