Learning how to listen

Recently I have come across several articles, blog entries and webinars on using social media data to drive company innovation. While customer insights culled from sites like Twitter and Facebook can certainly be valuable they are not a substitute for market research in understanding what customers want and need.

Many of the largest brands in the world, including Microsoft, GE and Siemens, are engaged in some type of social media monitoring. The reason is simple: it’s a relatively inexpensive way to obtain large amounts of data directly from customers. Search Twitter for any major brand at any given time and you can immediately access dozens if not hundreds of tweets offering opinions and observations. If a crisis affecting a brand occurs, social media sites are often the best place to look for early information. And today’s social media monitoring tools like Radian6 do an excellent job of aggregating, arranging and analysing the conversations.

But there is a problem. The data might be indicative of something – perhaps something earth-shaking – but we can’t be completely confident about it for a couple of reasons.

First, there’s sample bias. Nearly half of Twitter users are in the 18-34 age group and according to 2009 figures from Quantcast most of those users are childless. Making a major business decision based on this demographic could alienate a large portion of your market. Second, there is lack of context. Search for ‘Microsoft Word’ on any social media platform and you will find plenty of complaints about sudden crashes. But without further information about the computers these customers are using, as well as the version of the software, taking a definitive course of action to address the problem would be a shot in the dark.

Add to this the challenge of automated sentiment analysis employed by many monitoring tools, which use algorithms to attempt to determine if a social media mention is negative or positive. You might search for ‘AT&T + love’. The hits will include people who tweet, ‘Don’t you just love when your AT&T coverage drops out at the airport?’ Sarcasm is common online, and it is difficult to filter out.

Obviously, companies don’t advertise the regrettable decisions they make after this kind of mistake. But that doesn’t mean it’s not happening in companies large and small today.

Social media can play an invaluable role in driving innovation and decision-making once we understand what it is: the first round of qualitative data. It should be used to identify issues and concerns that face customers and the market. Researchers must then employ quantitative analysis to discover if what is being tweeted and updated on Facebook is true for a significant sample of our market or just a vocal minority (and depending on the product, service or brand, even a few thousand tweets could represent a small group of outliers). To listen in a way that helps drive sound business decisions, both qualitative and quantitative data are needed.

The key to effective quantitative research is to find a way to listen actively. Take surveys for example. Active listening means following up survey questions with new ones that explore the topic more deeply. It is harder to plan surveys this way, but the deeper the respondents are engaged, the more insight you will gain. And that insight is what enables you to identify new opportunities and define your competitive advantage.

To listen actively, you begin with the intention of having a conversation – as if you’re having coffee with a customer. Your questions must be interesting and relevant to customers, especially at the beginning (so you won’t be starting with ‘What is your address?’) It’s not enough to ask how good your service is – you must ask why people like it or don’t like it. Leave room for answers you’re not expecting because that is often where you will find material to explore more deeply in your next survey.

Finally, the process must be iterative – continue going back and listening further. Too often researchers try to put everything in one all-encompassing survey, when a few shorter, more topical surveys would do the job better.

A quantitative ‘active’ listening programme will ensure that you’ve listened to an appropriate sampling of the entire market. Only then can you know if the opinions you hear on blogs, Badoo and Twitter are simply held by those with the loudest voices, or if they represent a real market concern. What does ‘I like it’ mean? Quantitative research puts an algorithm to ‘I like it’ and reveals who likes it, who doesn’t, and how much.

Of course, this isn’t a new challenge. In the early days of market research, a researcher would count how many times people used the phrase ‘I like it’ in a focus group and use that as quantitative data. Models and methods have changed over time to ensure such narrow samples are not the sole basis for major decisions.

Social media data can be an important piece of the research puzzle, but it needs to be used with caution. Essentially it’s a self-directing focus group selected by indeterminate means – it allows us to hear what we couldn’t before, but doesn’t always help us understand it. It gives us the ‘what’, but to get the ‘why’ and the ‘who’, we need to ensure we’ve listened to a broad sample of the market. Only then can we be sure that what we’ve projected as the concerns of the market are real and worth acting upon.

Thompson Morrison is CEO of i-OP Inc which provides software and services for market research and customer relations