Is Sentiment Making Brands Stupid?

I loved the article by Nick Carr a few years ago that asked the question – Is Google Making us Stupid? Nick likened the use of Google and other tools as training us how to “power browse” and not really read effectively.

So I have been thinking about a parallel article that would apply in the social media space as people try to effectively read what is being said on mediums like Twitter and Facebook about their brand. Most often they end up with a tool that monitors for them brand mentions and breaks them into bite sized chunks of data and ranking them by sentiment.

So consider the fact that raw sentiment as determined by machine is often only 65% accurate – to me this is dangerously close to 50% or the equivalent of a coin toss. In fact most social media analytics firms do something to “augment” their sentiment analysis to make it more accurate – usually involving human intervention.

But in my estimation these firms are working on the wrong end of that equation. Why fix sentiment – its never going to be 100% accurate – since most people can’t agree on the spirit or intention of a tweet anyway and they never will.

I think brands need to focus on the broader content being used about a brand. Brand managers pride themselves on understanding the audience that surrounds their brands. The Internet and now Social Media have spurred two of the largest shifts in consumer behavior in the last 15 years; analyzing online and social activity is now a requisite to accurately anticipate trends. And focusing on positive sentiment over negative sentiment ignores 50% of the conversation that was classified as neutral – I guarantee you that there are key learnings in there that any brand would want to focus on. In fact in our tests, only 10% was either positive or negative (obviously, that can vary from topic to topic or day to day), so even more than 50% of the posts were ignored. So I ask you again – is Sentiment Making Brands Stupid?

4 comments to Is Sentiment Making Brands Stupid?

  • I really like your question Paul. For me it raises the wider issue of what happens if brands start treating these metrics as polling booths for their strategy. The risk then is that they could look to appease everyone and in so doing become everything to no-one. I post on this here: http://markdisomma.wordpress.com/2011/08/08/everything-to-no-one/

  • Wow excellent comment Mark – yes I guess its like driving only looking in the rear view mirror (reacting to what others do) then your strategy is truly not your own!
    Thanks for the comment and mention in your post!

  • Kyle

    First off, a coin flip is not the correct analogy. Random in this instance is 33% (one for each category). That means that your 65% number is 2x better than random.

    Also, 65% is very low for sentiment scoring, not sure why you use that value. 85-90% is more top of the line.

  • That’s fair
    But don’t forget in 2011 (when I wrote this) 65% was close to “top of the line” in sentiment

Leave a Reply

You can use these HTML tags

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>