May 09, 2012

Control Limits for Social Media Trend Analysis

You have been making concerted efforts to increase the social media buzz about your brand. You do regular posts and Tweets hoping that talk of your brand will spread across the social landscape. One of the ways you measure your success is by looking at an overall trend in the number of conversations about your brand over time. 

As you look at this trend you see that some days there are spikes in the number of conversations about your brand—these spikes look like mountains on the line graph. Other days you may see valleys. Often times it is volatile and a peak can be immediately followed by a valley. But, how do you know if these peaks and valleys are normal or abnormal? Which peaks and valleys are part of the normal ebb and flow of conversation and which ones are not?

There is a helpful statistical concept called control limits that can be used in this situation to answer these questions. There is an upper control limit (UCL) and a lower control limit (LCL). These limits are calculated by taking the average number of social media conversations per day and then adding (for UCL) or subtracting (for LCL) the product of the standard deviation multiplied by three.

So, for example, let’s assume there is an average of 8 social media conversations per day about our brand. Using the control limit formula, our UCL would be 13 and our LCL would be 2.9. As we look at our social media conversation trend line graph we will see that some days the number of conversations falls between 2.9 and 13 conversations per day. Even though there may be 12 conversations in one day (which is 50% higher than average), we know that that day was part of the normal ebb and flow of conversation because it fell within the control limits. 

However, other days the number of conversations in a single day may be 25 or it may be 2. If there were 25 conversations in one day about the brand, that is not part of the expected ebb and flow of conversation because it falls outside of the control limits. That day should be analyzed to determine why there were such a large number of conversations about your brand.  You may find that it was due to a particular type of Tweet you did and this may be something you will want to repeat in the future because it caused a higher than normal amount of social media conversation about your brand.

Understanding the analytics behind your brand’s social media conversation is just as important as the conversation itself. Through developing a high level of analytics awareness, you’ll be ready to tackle your social media marketing strategy for the quarters and years to come.

 

[Contributed by Ben Hill]