I have been deeply reading the Wiley published Audience Research Foundation text book on online listening (not an easy read) titled Listen First!: Turning Social Media Conversations Into Business Advantage by Steve Rappaport and my initial take, after reading the first 6 chapters , is that using social platforms requires a traditional market research grounding in order to make the best use of listening platforms such as Radian6, Sysomos, Brandwatch and so on.
For example, in just about all the case studies in the book, no one uses a listening platform without carefully vetting their web sources (forums and blogs, mostly) and writing extensive filtering to get out the material that is to be examined, the signal, without the noise (information you don’t want). Using a listening system without extensive source vetting is strongly discouraged by the ARF. Some of the source vetting can come from client leads while others will have to be discovered online, via examination, trial and error; this process should be budgeted for, along with regular governance activities associated with keeping up the quality of the data feed, the fundamental of all good listening reports and insights.
Usually, noise will be filtered out if we pick the right sources (certain blogs and forums) and write additional filtering. While many listening platforms have a query length limitation that is often an issue for the elimination of noise, correct source filtering will help.
For those who are not familiar with the setup of listening systems, certain business sectors such as pharmaceuticals and online travel, to name a few, will have large amounts of spammed content contained in the data feed. In order to filter that out, large queries are often necessary.
The same is true for Brand names such as Tide, Orange, and even WCG; extensive filtering is needed to give us just what we’re looking for, without all other mentions – sometimes there can be millions of mentions.
What is even more salient is that, under certain listening systems, we may be paying for the online mentions we don’t want, as many systems charge by mention numbers. So good filtering not only helps with better reporting, but also helps the pocketbook!
Finally, when we look at the performance of online listening systems, extensive filtering could slow down their performance – but improve analyst performance – sometimes we may have to choose systems that can give us what we want – but we need to understand how to ask for it.
Influencers are also identified beforehand. Whatever performance problems a listening platform has (and there many problems to consider such as accuracy, timeliness of data and query length) the more fundamental issue comes down to governance, their performance is predicated on the need to be calibrated properly and maintained to perform well, much like any system.
In our use of new technologies to augment communications, such as using one off improvements of text analytics, let’s remember that the fundamental research principals of vetted sources and influencer identification are as applicable and necessary here as they are everywhere else.
There are some best practices around around doing Ethnographic online research that also are playing into insights to be derived from the study of the behavior of online communities – Netnography: Doing Ethnographic Research Online by Robert V. Kozinets appears to be the only text book for this work that exists, as of yet, until someone writes another, and probably should be required reading for analysts and communications professionals as well.
Initial reviews seem to be mixed, and the book doesn’t appear to appeal to everyone who has read it, but given the difficultly of the subject matter together with the value of the offering makes an argument people should read valuable books primarily for the insights they contain, as the subject of Netnography will not make anyone’s best seller list.
Given the above, analytics deliverables will be measurably improved to the extent that best practices are standardized and applied to the tasks at hand. Understanding online behavior, while often daunting, when grounded with market research, becomes not only better, but far more defensible, especially with clients and stakeholders, especially with situations where fast turn around of a deliverable is needed. If a client understands that speed may sacrifice quality, they may opt for quality, or at least, if they need speed, they will be more likely to accept the tradeoffs.
Once sound principals are adopted, we can all rally around them making our discussions and improvements focused on the the right things.