Should we listen to every conversation?

Over on the Essential Research blog, I have responded to a post by a social media conversation monitor who eulogised the death of focus groups.

In that post, I have outlined why focus groups themselves aren’t the issue; rather it is shoddy application. Here, I want to expand on that a bit. It is my contention that conversation monitoring is more flawed than traditional research, and should not be used for major corporate decision.

Alan Partridge once declared himself to be a homosceptic, and in a not dissimilar way I am doubtful of the efficacy of social media monitoring.

In terms of numbers signing up, the social space is still increasing. However, the number of active users within this universe will remain limited – the late arrivals will be the more passive and occasional users. This space is increasingly asymmetric, with network effects and power laws distorting the flow of information.

Topics of conversation will by nature revolve around the major players – whether individuals, blogs or organisations. The larger the hub, the weaker the concentration of signal to noise.

As a small example, consider blog commenting. Aside from the odd spam comment, the contributions I get here are all genuinely helpful. Because this is a relatively small blog, there are few people commenting out of self-interest. Moving to the larger sites, comments are filled with spam, self-promotion and unquestioning advocacy/contrariness. Genuine debate and discussion still exists, but it is diluted by the inanity surrounding it. This on its own creates difficulties for sentiment analysis, but clever filters can overcome this.

But despite the internet being open, we will cluster around likeminds. Group think creates an echo chamber. danah boyd has pointed out that teenagers network with pre-existing friends. It is my observation that the majority of adults network with those in their pre-existing spheres. Planners chat to planners. Cyclists to cyclists. Artists to artists. Mothers to Mothers. These categories aren’t mutually exclusive, but the crossover is minimal compared to likeminds.

Remember the Motrin outrage? The mainstream majority remain blissfully ignorant. This may have been because it was resolved before it had a chance to escalate to the mainstream media, but it nevertheless shows the limited nature of social media echos.

Of course, some products or services target the early adopting, tech savvy ubergeeks and so for these companies they should obviously engage where their audience is.

But for the rest? Despite my assertions above, I do view monitoring as useful, but only as a secondary tool. Tracking conversations as they happen is a useful feedback mechanism, but few companies are going to be nimble enough to implement it immediately (once they have separated the meat from the gristle and verified that this opinion is indeed consensus).

Surveys and groups are indeed limited by taking place in a single point in time, and through these it is difficult to extrapolate long-term reaction. The Pepsi taste test being one notorious example.

But there are plenty of longitudinal research methodologies that are suitable. Long-term ethnographic or observational studies can track whether attitudes or behaviour do in fact change over time. These can be isolated in pilots or test cases, so that any negative feedback can be ironed out before the product or service is unleashed to the general public.

This is where traditional research still prevails: the controlled environment. Artificiality can be a benefit if it means shielding a consumer basis from something wildly different from what they are used to.

This takes time though, and some companies may prefer to iterate as they go, and “work in beta”. Facebook is an example of this – they have encountered hostility over news feeds, Beacon, redesigns and terms of service.Each time, they have ridden the storm and come back stronger than ever.

Is this a case study for conversation monitoring effectiveness? Not really. They listened to feedback, but only implemented it when it didn’t affect their core strategy. So, the terms of service changed back but the news feed and redesign stayed. Features intrinsic to its success.

Should Scyfy have gone back to being the Sci-Fi channel due to the initial outrage? Perhaps. Personally, I think it is a rather silly name but it didn’t do Dave any harm. If they have done their research properly, they should remain confident in their decision.

Conversation monitoring can be useful, but it should remain a secondary activity. A tiny minority have a disproportionately loud voice, and their opinions shouldn’t be taken as representative of the majority. When iterating in public, there is a difficult balance between reacting too early to an unrepresentative coalition, and acting too late and causing negative reaction among a majority of users/customers.

Because of this, major decisions should be taken before going to market. Tiny iterations can be implemented after public feedback, but the core strategy should remain sound and untouched.Focus groups and other research methodologies still have an important place in formulating strategy.

sk

Image credit: http://www.flickr.com/photos/jeff-bauche/