You’ve Got The Numbers…Now What? Working With Social Media Analytics
Almost every client I’ve worked with in social media wants data tracked and reported for practically every post, tweet, comment and sweepstakes that they participate in online, and rightfully so. From a business perspective, Key Performance Indicators (K.P.I.s) are important to help guide decisions and craft strategy. The problem that so many companies have with this process is that they don’t see it through to the most important part: the analysis and interpretation.
Data, without insightful interpretation, is worthless. It’s like staring at the instruments of an airplane, but not knowing how to use them to get you where you want to go.
So you have 200,000 Facebook fans…so what? How many of them engage on a regular basis? What countries are primarily represented, and is it important to your business? What time of day is best for you to post so that you get the most exposure? These are questions that should be asked, but often are not.
I think that so many people and organizations are in the habit of asking for reports that they just do it automatically, and assume that the process is over. The way I see it, the process is just beginning at that point, and data can be used to make important business decisions, particular as they related to social media, looking forward.
A few guidelines and suggestions for how to use the data you capture:
- Flash reports are okay, but real strength from data comes by looking at a broad range. The more time you have to collect data, the more solid your numbers will be and the variance of peaks and valleys shouldn’t affect the bottom line as much
- Sentiment is quite subjective, and I have yet to find a tool that auto-scores and does it well. For example, if somebody tweets “Good Lord, my [brand] car is giving me a headache”, it’s typically scored as positive or neutral because of the inclusion of “good”. A human looking at that would usually score it as negative. I would rather hand-score a small number of data points than let a computer auto-score a massive amount
- Consistency with time and services are important. If possible, try to pull data from the same source and at regular time intervals. For example, it’s much easier to analyze data from a single source that you pull every Monday, than to compare data from many sources that you pull when you “want to get a good look at things”. Consistency is key
- I’m sure you are tempted to look at numbers each week as wins and losses, but it’s more important to look at data over a longer period of time. For example, when we presented numbers to clients after the 4th of July Holiday weekend, tweets, comments and likes were down almost across the board. It has to be taken into consideration that people were off of their computers and outside enjoying life, otherwise it seems like something went terribly wrong during that period.
I spend a lot of time looking over data that has been scraped from all around the internet, and an equal amount of time interpreting what it means so that we can help our clients make important strategic and tactical decisions. What I’ve learned from all of this is that, no matter where you are pulling it from, data that stands by itself without good interpretation is at best worthless, at worst dangerous.0