Getting Sentimental About Analytics

How do you feel?

I know you clicked through to this column from somewhere else. I can learn all sorts of things from analyzing your behavior on this site, but I have no idea what you’re feeling. Are you happy or angry right now? Will you feel differently after you read this column? Perhaps you’ll go blog about what you read here and express exactly how you feel.

Welcome to the next great analytics frontier: customer sentiment. As we push the limits of data-driven clickstream analysis to the max, there are new opportunities to get into visitors’ heads and better serve their needs. The advent of social media and user-generated content — Web 2.0 — has created great examples of people expressing their opinions online. Those opinions have real business impact. There are some powerful tools (e.g., Nielsen BuzzMetrics, Visible Technologies) available to track these discussions and measure sentiment. The missing element, however, is tying this all back to site behavior. If someone writes a glowing (or negative) blog post in the morning then visits your site in the afternoon, what exactly does that mean for your brand?

Industry luminaries like Eric Peterson have been talking about measuring customer engagement for some time. How can we tie this back to measuring how those engaged customers feel? Just because visitors stick around for a while doesn’t necessarily tell you if they’re happy or mad. Frankly, I know a few people that spend hours on sites they hate just looking for more things to pick apart in their blogs. Remember when we were all caught up in sticky sites? Time and experience taught us trapping people in a site that wastes their time isn’t good at all. Visitors want to find the information they need, even if that information is the customer service phone number so they can scream at you.

A Framework for Understanding Sentiment

OK, I’m not going to provide a complete framework here. But I am going to make some suggestions on how to think about this key issue and put it to good use.

First, you need a simple taxonomy of sentiment to serve as a foundation for discussion. There are lots of tools that measure this; you can do semantic and linguistic analyses all night long. Let’s keep it simple: positive, negative, and neutral. If you can segment your audience into these three buckets, you’ll be miles ahead of your competitors. You can worry about fine gradations after you get the basics down.

Next, think about integrating the data along the entire customer lifecycle. As I’ve advocated for some time, there are lots of data sources and every one can add value to your understanding of customer behavior. Do everything you can to drive integration and view behavior through a variety of lenses: surveys, behavioral analytics, focus groups, linguistic analysis, and good old-fashioned clickstream data. You want to always be able to put customer behavior into the richest possible context.

Finally, start paying attention to social media and user-generated content. There’s a vast mine of data just waiting to be tapped. Find the tools and experts to listen critically and learn from the conversations. Imagine tracking people who come into your site from a particularly nasty (or laudatory) blog entry. Be prepared to offer them something unique based on the sentiment that just “infected” them. If this doesn’t get your mind racing about new possibilities for optimization, I don’t know what will!

I’ll leave you with a little example. Imagine you work at a video game company with an elaborate site and hundreds of dedicated bloggers out in the marketplace. When you think about the discussions going on, you’ll need to create some specific examples of terms along the sentiment scale. In this case, you might place “nice” on the neutral/negative end (no gamer wants something that’s just “nice”) and “sick” on the positive end (to a gamer, “sick” is high praise).

But apply those terms to a family-friendly video site, and the meanings turn completely around. Sentiment can teach you worlds about your audience and help make the experience for everyone that much better. Now, that’s sick!

Related reading

site search hp
ga hp