[An edited version of this post first appeared on Research Live here.]
I’ve been thinking a lot lately about the consumer research work we do here at Relative Insight, and I’ve come to the conclusion it’s not consumer research at all. It’s cultural research. And what’s more, we straddle the seemingly dichotomous worlds of qualitative and quantitative research.
When we talk about consumer research we usually mean interviewing. This includes sending out surveys, usually asking consumers questions which are specifically designed to give numerical answers so that we can process them more easily. “On a scale of 1 to 10 where 1 is disagree strongly and 10 is strongly agree how do you feel about….”. We can gain some insights and spot trends but it’s easy to spot the flaws in the method. Fundamentally, we humans express our feelings using words, not numbers.
Observer and participator bias
When it comes to qualitative research, it strikes me that there are two kinds of people who attend focus groups – those that want to give you a positive outcome because you’re buying them lunch, and those that want to give you negative answers because, well, that’s just the way they are. It’s a good example of the Observer Effect – by shining a light on something we change the phenomenon being observed. But we spend hundreds of thousands on focus groups because they do pop out the odd gold nugget, and there’s no other way, right?
Well what if you could get those qualitative-type insights, without the observer effect, and at a massive scale?
At the beginning of the year we were asked by a supermarket chain – which wished to better inform their summer campaigns – to look at how mothers talk about the season, theme and concept of ‘Summer’. We pulled out four months’ worth of posts from two large mothers forums where Summer was mentioned in the thread title. We processed two million words – to put that into some context, the entire set of seven Harry Potter books comes to a shade fewer than one million words.
At this point it’s worth explaining how we work at Relative Insight. We turn sets of language into fairly complex data models, store them, and because we always follow the same process we can compare those models to surface the differences and similarities to each other. We look at the words themselves, the grammar used, the semantic topics of conversation, the style of language, and any metadata we can pick up such as who, where, and when. It’s the culmination of ten years of research and development at Lancaster University, originally designed to catch bad guys online in the field of Child Protection (which we still do). We are now used by CMOs for a mixture of market research, consumer research, and ongoing language drift monitoring.
It’s not just what you say…
For our supermarket client, we found a lot of common areas of discussion in the Summer analysis, but the big insights came from the differences between the two forums. The first forum’s users talked significantly more than second forum’s users about Architecture (visiting stately homes, putting a wigwam in the back garden, and Yurts at Glastonbury!), Geography (where they are going on holiday and what they are visiting when they get there), and Education (not just where the kids are going in September, but also continuing education in the summer break). The way they described the topics was much more positive. Everything is ‘Beautiful’, ‘Lovely’, ‘Amazing’. In contast, the second forum’s users were a lot more practically minded – it was very interesting to see they discussed the kids overwhelmingly more than first forum, but they primarily discussed the kids getting ill, having to buy a bunch of stuff, and how much it costs.
The supermarket was able to use these insights to inform how they approach both audiences. As with all the brands we work with, they used our insights to target the language for different audience segments. We showed them that if they wanted to target the first forum’s users they should shy away from discussing the price and talk more about the experience and how they benefit when visiting stately homes or their holiday destination, as an example. Conversely for the second forum’s users, who have a more functional view, do talk about how cheap products are and how they benefit the kids. We’ve just rerun the project, looking at conversations around both Summer (to see any changes) and Schools. While it was without a doubt an interesting and useful piece of insight work for the supermarket, it was some of the unexpected results that gave us the most cause to step back and think about what we had uncovered and how.
The first forum we looked at is quite well known for its colloquialisms, but we were able to highlight a mass of previously unrecorded variations coming through from the users. This phenomenon of colloquialisms as well as a distinct linguistic style comes through on many of the forums we look at. People create tribes, with their own language. Culturally it’s quite an inclusive thing, though it does seem that if you don’t reflect a particular tribe’s language then you’re “not one of us”. The power of language reflection is well documented – as David Ogilvy said, “If you’re trying to persuade people to do something, or buy something, it seems to me you should use their language, the language they use every day, the language in which they think.”
As you can tell we don’t explicitly search for things. We don’t search for brand mentions, for product mentions, or for people mentions. We scoop up language in its entirety and, critically, we surface the similarities and differences. Because we work this way, we’re able to quickly highlight the use of colloquialisms, identify the style of language used by the consumers, and understand what matters to them. This is qualitative research at a quantitative scale. By using this approach, brands can start to better market to their consumers.