Tuesday, October 30, 2012

The maturation of social media analytics

In the last post we discussed the pillars of social media and the questions a buyer should ask its vendors when buying.  This is useful because the language discussed levels the playing field a bit.  I have included these questions if you didn't read my last post.

The post is here...(link to post)
The pillars and their corresponding questions...


But how does this apply to the market itself?  That is a more interesting question and one that needs a historical perspective.  In many of my posts, I have often written about the time analogy.  To simplify it (you can find it in many of my other social media posts) one has to think about the maturation of the social media analytics market as a march to high noon.  When I started in this PULL part of the market it was midnight.  This was about 2006.  As different market events have occurred the sun has come up.  In fact most of my posts would suggest it is about 930AM now.  Dawn hit late last year when people actually went from arguing about even buying analytics tools to deciding which one they would buy.  As people have grown more savvy and bought various tools, we are seeing cycling.  930 arrived when people decided to move beyond simply worrying about features and working in silos to actually collaborating to have a social media analytics program.  But how is the market actually maturing.  I would argue it still is.  And by my account there is a long way to go.  Analytics is still trying to find its way and companies are still pondering the ROI of their investment.  It is an age old business question.  Below is a model that I use to think about the maturation of the market.

Below is a description of the 4 pillars listed above and view of how the market is maturing. 


As I have described how you can ask your vendor about the 4 key areas of development, we can also use these "pillars" to describe how the market is maturing.  I will actually discuss next how this applies to the sophistication of an organization social media analytics program, but today let's just discuss how those buying are speaking and where the majority of that market lies.

Before the FEATURES PERIOD:  Before 2009

In the early days of social media analytics, people literally laughed at this idea.  In fact, many couldn't even imagine garnering insight from this data.  The owner of the company that I recently worked at, in is old fashioned and often wrong wisdom,would call it a really dumb idea.  He swore that the companies were the only ones out there marketing their message to everyone.  It was a lie.  I used to smirk at him because that is the closed minded way to look at things.  The rise of data changed that...the rise of twitter and facebook brought about the need to look at what the data said. The question at hand is how would we pick what we used to see what the data was saying.

The FEATURES PERIOD:  2009 to Late 2011

With the rise in the amount of data, came the rise of tools.  In the first phase, or the Features phase, it was all about what you could measure and how much data you could measure. Companies who were new to this idea would buy whatever tool made sense in terms of its ability to "show" them things.  They didn't think about content or whether there was any data quality; it was all about tracking trends.  Radian 6 did an excellent job during this time, because they built an engine that showed the PR function what it needed to know.  When things go wrong, what is the movement of the data.  In fact, they mostly wanted to be able to engage.  I would argue actually that this use case had little to do with analytics and everything to do with "marketing" away the problems because you could know what they were saying and tell them it was going to be OK.  This seems a little harsh to say, but a recall based use case (meaning show a trend from every piece of data regardless of quality) is one that is basically unsophisticated.  The features period was important because it established the fact that trying to make sense of the content was an important part of building a broader social media program.  It is clearly the foundation of true analysis, but in the early phases people mistook the ability to measure something with the quantification of a action.  Unfortunately, if you focus on features you are missing the depth the data from both a completeness and accuracy perspective.  The features based organization doesn't ask questions about is it all the data or how good is the data...they simply care if you can see the numbers of some set of data.

The CONTENT PERIOD:  2012 to ????

Today we are in the midst of the content period.  The rise of this period is clearly refreshing because there has been a greater evolution of most companies social media programs.  This period is the admission that I need to know as much as I can about as much data as I can collect.  I need the Twitter firehose so I don't miss anything.  I need to be able to see what people are saying within this massive data.  And most importantly, the features of my solution have to help me separate the wheat from the chafe.  It brought about looking globally at the social data (what languages do you have). It is bringing about the expanded desire to segment and understand all channels (facebook, twitter, you tube...).  As far as the content period is concerned, companies are really testing their tools.  They are choosing.  They are regretting those choices.  They are fixing that problem, by choosing another.  But people are still looking for the elusive ROI from the social data.  They are still stymied as to how they should apply the data to more use cases within the organization.  They are still more interested in having their fingertips on all the data and ways to slice that data.  But to make informed decisions with the data, companies need two things.  They need to realize that tools don't solve the the problems, but the process by which you apply them does.  The second is that if you have poor data it is really hard to make a repeatable decision.  The content period is a great leap forward in social media analytics because it acknowledges that completeness and thoroughness are key factors is knowing.  The question remains is when the market will skew towards the need for accuracy.

The ACCURACY PERIOD:  When will it come?

This is the question on my mind.  When will this period truly start?  I see the seeds of in the major partners I work with.  They are showing signs of thinking very comprehensively about their platforms now.  But how will this period of analytics be defined?  For one, it will be about creating meaningful correlations within the data to truly know you are making good decisions. It will help validate the value of social data.  I believe companies will be thinking harder about pulling content rather that simply creating it.  The rise of application and solution based thinking will overrule a single tool or dashboard to give you the answer.  Accuracy is what analytics requires and as we move into this phase the power of social can be tapped.  The questions I have for anyone reading this are these?

1.  Do you believe having every tweet is more important that accurately knowing the emotion behind it?
2.  Are you looking for solutions that accurate help you understand what the data says rather than that you have all the data?
3.  Are you considering whether your talent is more interested in creating content or accurate informing those who create content what content they should be creating?
4.  Do you consider solution based social analytics to be bigger than any single tool?
5.  How are you challenging your company to consider all the angles of how to build a holistic use case library that really impacts your bottom line?

The market is still evolving.  I challenge myself everyday to understand it more.  The greatest thing about emerging ideas is they are rife with the seeds of learning.  The more we challenge ourselves to think broadly about a problem, the faster we can invent ways to leverage the opportunity.

5 comments:

  1. This comment has been removed by the author.

    ReplyDelete
  2. A very helpful blog, thank you Malcolm. One question if I may, how to get analytics software vendors to give me the information I need on the accuracy of their models? As you have noted in later blogs, accuracy of data is imperative - I would add, so is current relevancy of the models. Without understanding the parameters used in the software models, and how they are being updated using methods such as streamed training / machine learning - how do I know I can trust the outcome of the software 'models'? Very few vendors are willing to provide this information - claiming it is their IP....but it's my business or my clients business that is going to rely on it!!. Any suggestions would be most welcome.
    Gail

    ReplyDelete
    Replies
    1. Gail...sorry for long response...been away from the blog for a while. I have been blogging on social media explorer exclusively because I am so frigging busy. I work at netbase and we do have the answer to your question We use natural language processing and actually track the accuracy by using human crowdsourcing to gain alignment on our work. If you go to netbase.com you can get the white paper taht describes this

      Delete
  3. Gail...sorry for long response...been away from the blog for a while. I have been blogging on social media explorer exclusively because I am so frigging busy. I work at netbase and we do have the answer to your question We use natural language processing and actually track the accuracy by using human crowdsourcing to gain alignment on our work. If you go to netbase.com you can get the white paper taht describes this

    ReplyDelete
  4. This clarity with your post is superb and that i may think you’re a guru for this issue.
    instagram shoutouts

    ReplyDelete