Social media strategists like to portray Facebook, Twitter, and other social networks as great marketing venues, but, frankly, they fall short of being able to measure campaign success effectively.
Effective social media measurement is much more difficult than it seems, mainly because it is almost impossible to get a representative sampling from social media, as Gary Angel, president of Semphonic and SemAngel blogger, discusses here. While social media measurement may prove rewarding when using social media sites to conduct online focus groups for market research, as representative samples wouldn't be required in such instances, that's not the case so much when measuring social media for brand sentiment, brand awareness, share of voice, share of sentiment, and other such activities.
As a result, I think we are largely measuring the wrong things and going about measurement the wrong way.
As I wrote in my book on social media analytics, I saw an opportunity to disambiguate the choices businesses need to make about which social media analytics platforms, analytics software, and processes they need consider for their social media measurement needs because it's hard to look under the hood and see what's really there. In that sense, I was trying to map a better approach that also put the analyst in charge, to give him or her a seat at the table as the person who is doing most of this measurement work, instead of the business development, sales, or brand person (although this is the conventional wisdom and DNA in many marketing communications departments).
One big issue in measuring social media is that businesses are capturing just a fraction of the data needed to calculate return on investment, and most of the rest is what I call "ultraviolet," which is either data that you can capture but don't, or data that you do capture but is not usable for analysis and business insight. As a result, capturing enough contextual data in social media with its vast troves of unstructured data is highly difficult.
Most social media platforms end up falling back on measuring friends, shares, tweets, retweets, and other quantitative measures that are relatively easy to get and calculate. But what we're really after is a qualitative assessment of effectiveness -- and then, by how much. In general, the formulas for calculating effectiveness do not map well to social media. This is because part of the formula for social media ROI, the cost of the investment in social media, often involves soft costs such as the value of time spent.
Common issues coming up with measuring social media are the confusion surrounding what to measure in social, along with the sheer volume of unstructured data and the lack of any universal standards on what and how to measure social. To make matters worse, many vendors and agencies obscure their marketing messages, misrepresenting offerings in order to sell more of their products and services. Clients and stakeholders often lack the necessary vocabulary and lexicon to articulate their real needs and wants. Another issue is finding the right people to do the work, along with clients willing to pay for the right systems to collect, clean, and structure the data being analyzed for social media effectiveness.
Distilling signal from noise in social media is pretty expensive, when you stop to think about it -- so much of it must be manually culled if you care about accuracy on sentiment and topic analysis -- and the already sizable data volume is growing constantly as the sources and reach of social media expand, too. As an analyst, I believe that measuring the effectiveness of social media requires having the right data at our disposal -- and if we don't have great data to work with, it's hard to get past that.