The Importance of Slope - Social media Volume versus Time

As many people in the social media sphere are constantly looking for the ROI on how to apply this data in a predictable and sound manner, what continues to be lost in this discussion is METRICS.  As I stated in my last post, it has been a real challenge to get people to use what I consider strong operational definitions of social media (the listening versus understanding argument).  In fact, as I think about it I see this incredibly powerful emerging data set that is really a mess.

Why is it a mess?

Firstly, because the ebb and flow of information is so dynamic.  This incredibly malleable and wandering data set can actually make it hard to hold.  What works in its favor is the volume is so large that the room for error is probably pretty forgiving.  I mean when you can pull 1,000,000 data points on a particular event or topic at any given time (if you are lucky enough to have that amount of recall) you have to admit the room for error decreases a bit versus holding a focus group of 6 people who are being led by a "neutral" moderator. 

Secondly, social media as a data set is currently a frontier.  People are unsure what they want to do with it or how to approach it.  Can they trust it?  Will it augment what I do or should I replace what I do?  It isn't traditional and is new...how should I fund it?  The misinformation on the data set is troubling.  Only 15 year olds write on the web (even though 55-60 yr old women are the fastest growing segment).  This type of volatility makes for great angst and great opportunity.

And lastly, social media is a real threat that people are not paying attention to yet.  I am bewildered by the fact that large companies don't notice the threat it poses them.  With the right tool in their hands, small companies can now interact with consumers anytime, anywhere from the comfort of their office.  This is going to change the game on how we compete with each other.  In fact, I think this will be a deeper discussion for my next blog post.

What does this all have to do with the importance of slope?

Everything...because we have to first learn how to make sense from the noise before we can understand and predict it.

I have spent so much of the last five years thinking about how to apply social media effectively to produce business results that create Tangible ROI.  Whether this comes from saving money or creating new opportunity, I have always and still believe that application is the name of the game (I will probably keep pounding this from post to post because hey...I am an evangelist and it is my job).

But recently, I have really taken the opposite approach as I discussed earlier today.  I believe it is the development of METRICS that will help turn the tide and create scalability to social media application not the other way around.  So at this point I spend a lot of time looking at our natural language processing capability, our huge index and the power that words can now play in the development of metrics.

And when I think about the concept of Buzz, Sentiment and Passion and NetBase's ability to capture those three counts/measures accurately, I begin to see scalable patterns that can be apply to the issues at hand.  If you use the earlier analogy I made around trying to ring the bell at a carnival and that each measure represents a more sensitive measure of your social media effectiveness, these three counts become increasingly important to consider.

Why?  Because recently, I was looking at an event that intrigued me (Obama's speech on Libya about a month or so ago) and I noticed something very interesting.

 

Above is a net sentiment chart on the subject of Obama and his speech on libya on 3/28.  The chart gives a look at a selected time period before and after his speech.  The green part of the chart is positively expressed emotion about Obama in the context of Libya and his speech.  The Red part of the chart is negative emotion being expressed about the topic. The blue line is the Net Sentiment or %positive - %negative (form -100 to +100).  You will distinctly see the volume of sentiment expressed climb dramatically when he gives the speech on March 28th.  I have only tracked this study through the day after the speech.

The next slide is a look  the absolute buzz (non-sentiment based sound bites) expressed in the news and on microblogs (twitter) and the % change this buzz represents on the subject (the two lines).  You will notice here that twitter line is climbing at a much higher rate than the news (which tapers off pretty quickly).  In fact, it was pretty amazing to see the "lifecycle" on this event.

The third slide is the same view as the first slide, except it was rebuilt only using twitter content (not the sum of all social media sources like news, forums, blogs, microblogs social networks etc).  What is very interesting here is that the net sentiment for the twitter chatter on this topic is much hired 43% net sentiment versus 24% for all content types.  This actually shows in this case at least that there is a difference when looking at different sources of social media.  

Why use this example?  Because it brings me back to the importance of slope.  What this data suggest is that there is a life cycle for any social media event.  And what some might hypothesize is that each source represents a different point in an events life cycle.  And as you think about different counts you can measure buzz, sentiment and passion (three different but deeper types of social media measures) you have a very sensitive footprint for a social media event.  

The question is this...

If I take the rate of change of volume in buzz, sentiment and passion for an event AND I think about how each source type changes, do I have the secret footprint for social media?  Why is this a footprint?  Because these three sensitive measures tracked across different sources can be studied and become predictive when what an event occur.  Keeping it simple, if the slope is 4 on Twitter for one event and 1.5 for another doesn't this begin to predict the seriousness of one versus the other.

I think this is the foundation for using very simple metrics as a means of teasing out to become predictive of social media events.  And because they happen fast, you better know the event fingerprint.  

Of course this is all theoretical and you will need to be able to get those three sensitive measures to make it work...and that is where natural language processing comes in...doesn't it.  And an index that allows you to react to anything...welcome to my evangelism for a technology that is truly differentiated.

Sorry for the sale, but it is how a customer becomes an evangelist...and that is at the heart of MY social media passion.