In advance of the CMA’s Social Media Conference (June 23, 2011), we asked one of our speakers, Joseph Carrabis, Chief Research Officer and Founder of NextStage Evolution, a few questions about social media and sentiment analysis. NextStage Chief Operating Officer Susan Carrabis also weighs in.
Q. CMA: Is social media good for business?
Joseph Carrabis (JC): One could consider that question flawed. However, the flaw is unrecognized and is the source of a problem in the current “social media” environment.
The question as stated implies a "one size fits all" response, something like "Yes, social media is good for business" and that's not correct (me thinks).
Q. CMA: Mea culpa. I was expecting you might say “It depends”. What are better questions, helpful questions?
JC: The better questions are:
1) "What types of social (whatever) are good for what types of businesses?"
2) "What types of social (whatever) are good for what types of business goals?"
And other questions in a similar, more specific vein.
This re-questioning allows for a logical development and implementation process with a self-creating evidentiary trail.
Should things go sour, one knows where, when, how, why and for accountability* purposes, who. Hence things are more easily fixable.
Right now we're seeing a one-size-fits-all methodology even though consultants and companies are "tailoring" the solutions. This methodology shows up in "best practice" manifestos that are either completely general (hence useless) or so siloed that nothing works outside that vertical. If "social media was good for business" then a solution in the automotive vertical would work perfectly well in the sportswear vertical, in the gourmet food vertical, ... and this isn’t true.
To use social media correctly as a marketing tool, you need to be able to use the tools to create truly specific solutions. Right now the "state of the art" is the equivalent of "if all you have is a hammer, everything looks like a nail" and that's sad.
People aren't spending much time understanding their audience (at least in the terms I consider "understanding their audience") and the results are starting to show up as mass exoduses from various social platforms.
Understand your audience and you’ll know how to trigger loyalty responses rather than simple recency (with no offense intended to recency advocates. It's a reasonable metric that doesn't quite go far enough, in my opinion).
Q. CMA: If gaining audience understanding is so critical, but difficult, do you have any suggestions improvement?
Susan Carrabis (SC): Social Media, if used incorrectly, without understanding the audience and other Joseph words, is a dangerous tool. And please get away from one-size-fits-all.
We do a training based on the experiences of working with companies for seven years, a training based on how we actually work with clients. The whole focus of the training is to create marketing material that exactly targets your audience. You don't need NextStage tools or technology to do this but we learned how to do this by using our tools and technology. By the end of the day students are inside their consumers’ heads in ways they can’t get from just reading a book (except Joseph’s, of course). They look at material the way their audience looks at material, respond as their audience responds, see what their audience sees… It can be either fun or scary and we go for fun.
Q. CMA: So rather than seminars, is practical roll-your-sleeves-up, put your brain-to-work training part of understanding what’s needed?
JC: Seminars are (by my understanding) overview or summary in nature. We are true believers in "Understand the theory and you can apply it anywhere" and the best way to teach this is “Teach Application to Theory to Application”. This means take the student from the familiar to the new so that the new becomes familiar, start with something they’re doing in the real world, explain/demonstrate what’s going on, extract the key elements/principles, then have them apply those elements/principles to their current and similar problems. We taught the class mentioned above to about thirty marketing managers and they were blown away. They all knew the stuff, they simply hadn't 1) thought of it in the way we presented it or 2) performed their practice the way we were suggesting.
Part of that training is to develop a real-world campaign during the class. They were all doing social media campaigns -- the techniques will work for any kind of marketing (radio, tv, video, gaming, web, kiosk, print, down, up or sideload, ...) -- and it was exciting how many things were becoming obvious to them as opposed to their standard methods for campaign development. Not only understanding aspects of consumer psychology and how to reach consumers that they'd never considered, but which channels and media to use when.
This was a very rewarding experience. For us, and when we asked, for them, as well.
So for us it all comes back to knowing the correct question to ask.
Everybody knows I collect quotes and one I use is "It is perfectly useless to know the answer to the wrong question." Right now the majority of businesses are (in my opinion) asking the wrong question. I'll be sharing what I consider the correct question -- and answering it, or at least demonstrating ways to answer it -- in my CMA presentation.
CMA: Joseph, you’ve said that ‘sentiment’ is an exhausted term and the word has been so bastardized and mis-used such that it no longer has any real meaning. How do you define sentiment?
JC: There are lots of good psych and related definitions, all close and few metricable (ie. meaning they are reliably and repeatably measurable).
Following in the steps of Attention, Engagement and Trust: The Internet Trinity and Websites and Meet Online Engagement's Little Friend, Satisfaction and after a day’s research, "Sentiment" is the measure of emotional versus cognitive activity demonstrated by an individual when their attention is focused. This translates loosely to:
"Sentiment is demonstrated by behaviors that are caused more by emotion than they are caused by logic."
This is a definition that is a) accurate, b) metricable, c) teachable to any self-learning, social interaction system (Evolution Technology, for example) and d) repeatably metricable via that system within a reasonable +/- distribution.
This Sentiment definition can determine a visitor's sentiment in real-time while they're navigating your site. It won't matter if they write anything, post anything, comment or whatever. It won't even matter if they bounce. I now know what to have ET look for and we can make a tool that determines a) if sentiment exists, b) if it's positive or negative and c) the degree of that sentiment (companies interested in using such a tool should contact Susan).
Q. CMA: Has sentiment analysis evolved to the point that it can be relied upon? Do we have precision or accuracy, both or neither?
JC: I cannot comment on other company's tools or definitions of sentiment analysis because I only know about them from hearsay.
However, one of my favorite quotes is from a C level person at a major Canadian social media monitoring/analysis firm, "All we need to do is to score a handful of words and then assign polarity. It doesn't matter that different audiences use words differently."
This same individual and company arbitrarily (my opinion) decided that they could standardize their dictionary, meaning that words don't change meaning over time and within groups. I'm sure this company doesn't have any linguists or psycholinguists working for them because such a concept would drive them nuts.
But when that's what vendors are saying behind closed doors, how can what they offer be relied upon?
Take the "sentiment" metric I defined above.
NextStage has (at this point) twenty years of data we can look back on when we create a tool or test a definition.
We know, for example, whether or not people acted favorably or unfavorably -- had positive or negative sentiment -- to something going back twenty years, so when we come up with a possible definition we can test it against lots of historical information to determine its accuracy. When vendors knock on our door with referrals and references we ask for historical demonstrations of accuracy or ability. That’s when they go away.
I don’t know how a business can rely on tools without knowing how they came about, how they're tested, their accuracy and so on.
I also know that our clients -- all of whom want to test our suggestions via A/B and like testing in the beginning -- eventually stop testing and just start doing. There's also the aspect that many of our tools (our Sentiment Analysis tool being one) provide suggestions for better audience targeting. There's nothing quite like holding up a bullseye and saying "Here. Test this" to demonstrate confidence.
As for precision and accuracy, no comment about other tools because all I know of others is based on hearsay. If anything, I’ll go back to the bullseye concept. If anybody’s tool is either making or saving you money, it’s a useful tool. If it’s not doing either, why are you using it? There may be lots of other reasons, and that’s fine, just stop telling yourself you’re using a tool because it’s providing a business ROI.
Joseph Carrabis is speaking on "Measuring Emotional Response - Attaching Dollars to Sentiment Analysis" on June 23 at CMA’s Social Media Conference. During the afternoon “social media monitoring roundtable” session, attendees can ask Joseph directly about social media tools.
Tags: Digital, Research, social media, Strategy, technology
Posted in Uncategorized | |
