Back in the 90s I did quite a bit of ‘commercial ethnography’, mainly in exploring problem spaces and user experiences for tech companies. The mix of methods available were either deep and narrow, or broad but shallow.
One choice was the kind of rich contextual work of an observational nature that we called ‘deep hanging out’. At the other end of the spectrum was (largely) survey-based research. The deep stuff gave you rich insight on a limited population; the shallow stuff was broader but more representative.
I would argue that today, we have another option and that this has forever changed the landscape of information.
The alternative is rich, behavioral, use level data from a broad population, often our entire sample universe. In the days when software came with manuals, we used focus groups to help design the interfaces. While we still do some of this, the activity is overwhelmed by application data coming back to us in real or near time. What we clicked on, how much of the video we watched, where we were when we were watching it. This data is rich because it represents real behaviour rather than claimed behaviour or stated preference. We know that us humans are bad a predicting what we will do or explaining why we did it after the fact (e.g. famously, Nisbett & Wilson 1977).
But there is another dimension that makes this data valuable: immediacy. While we used to have to wait a month for a researcher to code the fieldwork and write up a report, behavioural data gets to us in real time.
The digital intelligence we get back from behavioural channels is ‘on the bang’. Imagine a timeline in which an event occurs. The event could be a catastrophic system failure or a bad service experience. Let’s call that event ‘bang’, ground zero. Anything we do after that event, to describe it, diagnose it, and react to it is ‘right of bang’. The closer our information is to the event the better able we are able to react. Our aim, of course, is to get left of bang but I’ll leave that for a future post. Getting timely information on the bang is of huge value because it allows us to make better decisions faster. Not perfect, just better. ‘A good solution applied with vigor now is better than a perfect solution applied ten minutes later’.
Back to the landscape. Today, when I mentally sketch out what data is available to help make better, faster decisions, something interesting happens.
The dimensions I think with are Latency and Richness. Traditional survey data is about half-way up the arbitrary richness scale because it’s claimed or stated in response to contrived questions. Choice-based approaches might be a nudge higher up because they are a more accurate predictor of actual behaviour, but both of these are slow (high latency). Ethnographic data is very rich (although not necessarily representative) but also relatively high latency.
What is interesting here is that the new kids on the block: big behavioural data, search intent, social sentiment, and even the re-invention of intercept polls through mobile, are all low latency. This puts them on the ‘action’ rather than the ‘insight’ side of the line. If your data is ‘now’ you can react to it tactically at a granular level. Service poor in one of your branches? Contact the manager and remediate. Video streaming problems across your east metro network this Friday night? Manage and communicate in social.
If your data is aged, the best thing you can do is roll it up and develop broader strategic solutions. Whilst these are important, you probably already know that you need to get better at managing your service roster and sort out that edge cache issue. Sounds like motherhood.
Where ethnography (or any of the traditional approaches) need to move to retain their value is across the insight/action line. That means getting either faster (more immediate) or richer. Currently, ethnography is useful to the degree that it provides depth, context and richness that cannot be realised elsewhere but the more developed the social and environmental streams of big data become, the more risk ethnography runs of losing its mojo.
But I think a bigger question concerns survey data. There are currently several industries that base a big part of their ‘intelligence’ on what customers say they will do in surveys. Evidence the growth in Net Promoter and related c-sat approaches in recent years. Compared to whether a customer says they are likely to recommend you, isn’t a much richer data point whether a customer is advocating you?
Today we can track this in email, and social (to name two channels) as well as get a handle on influence & reach. Of course we always hear that these digital channels are not representative but, frankly, how representative is the population that responds to surveys?
This article is by Jason Juma-Ross, former Digital Intelligence Lead for PwC.