Using the Median to Measure Performance for Social Media Marketers

  by    2   0

Measuring performance is key for social media marketers.  There is no reasonable way to predict performance without accurate numbers. Choose your methods of performance measurement  wisely. Consider the issues when  making inferences from data grabbed from social networks like Twitter.

As a marketer using social media  you also have limited data to work. On Twitter you can count the numbers of Retweets, Favorites, Replies and Mentions. The numbers alone will not tell the entire story. You will need statistics related to distribution, which are  valuable for performance evaluation.

The following discussion will be valuable for social media marketers. You will be able to account for the variability in your social media performance. Using the median provides another perspective on performance insulated from outliers.

The Limitations of Social Reach Revealed by Simple High School Statistics

Social media marketers are tasked with reporting the monthly performance of a marketing campaign. There are good social media KPIs, but they can be skewed by outliers.

Adjust you metrics when looking at daily, monthly, yearly numbers. You should account for variability in your data sample.

The Data Set

The data set used in this exercise was randomly generated. It replicates the performance of a highly influential Twitter account.

The data set is available for download.

 The Median is the True Middle in a Data Set

The median is a measure of the central tendency of a distribution. These measures show you how data spreads across a sample.

Typically, you would use the arithmetic mean for this;  which is also referred to as an average. In skewed data the mean is prone to error. Outliers, low or high, pull the distribution and change the value. The median in these instances provides a way to compensate for their effect.

The median is the data point in the middle of your data set when you list their values in numerical order. If the data set has an odd number then the two numbers in the middle are averaged. In a sample, the median separates the bottom half of values from the top half of values.

Determining Social Media Performance on Twitter

Take a look at the frequencies of Total Daily Retweets and Total Daily Favorites by Day of the Month.

The line in green represents the mean.
The line in yellow represents the median.

Retweets

Mean= 3,175.516 and Median = 2,434

Retweet-Freq

Sometimes outliers push the mean up.

Favorites

Mean= 6,026.645 and Median= 6,187

Favorite-Freq

Sometimes the outliers push the mean down.

Knowing the median allows you to find quartiles. Quartiles show the variation of the data above and below the median. They account for the middle 50% of the data set.

Quartiles for Retweets

Q1= 1,560
Q2= 2,434
Q3= 3,598

Half of the lowest values are found in the range of 1,560 – 2,434
Half of the highest values are found in the range of 2,434 – 3,598

box-whisker-renderer-retweets

This is a box and whisker plot which shows the interquartile range. The median is represented by the black vertical line. The middle 50% of values of the data set are contained within the box. The ends at the extremes, the whiskers, represent the min and max values.

Quartiles for Favorites

Q1= 5,579
Q2= 6,187
Q3= 6,713

Half of the lowest values are found in the range of 5,579 – 6,187
Half of the highest values are found in the range of 6,187 – 6,713

box-whisker-renderer-favorites

This is a box and whisker plot which shows the interquartile range. The median is represented by the black vertical line. The middle 50% of values of the data set are contained within the box. The ends at the extremes, the whiskers, represent the min and max values.

The median also tells  you the upper and lower limits from your sample. These are the outliers. They represent those data points which you would not expect based on your data set. Large variations in your monthly performance do not necessarily equate to these outlier points. That is why limits are so useful as an analytic tool. They only show you the unexpected values.

Upper and Lower Limits for Retweets

The purple line represents the lower limit.
The red line represents the upper limit.

Lower Limit= -1,497 and Upper Limit =6,655

retweet-outliers
Unusually High Performing Days: 10th, 13th, 14th, 22nd and 23rd

Upper and Lower Limit for Favorites

Lower Limit= 3,878 and Upper Limit =8,414

Favorite-outliers

Unusually High Performing Days: 13th, 14th, 19th, 22nd and 23rd
Unusually Low Performing Days: 20th, 28th and 30th

You cannot rely on counts to determine those high and low performing status updates. The median gives you a method for determining those points using statistical methods.

Using the median, you know the following facts about your monthly Twitter performance.

You can expect a daily total of 2,434 Retweets and 6,187 Favorites.

At the worst, you can expect a daily total of at least 1,560 Retweets and 5,579 Favorites.

At the best, you could expect a daily total of 3,598 Retweets and 6,713 Favorites.

These measurements could be used as benchmarks to predict performance for the next month.

Investigate the outliers and determine what makes them different or special.

Account for the Variability of Social Media Performance with the Median

The median is just one way to think about performance in social media marketing. Use the median when you are looking at performance daily, weekly, monthly or yearly.

The unpredictability of social media inflates, or sometimes deflates, the mean. This is due to the variability of  social data. This is why using the median is beneficial when investigating aggregate data.

Social media metrics like applause and amplification work best when judging  single status updates.

ArCompany is concerned with moving beyond vanity metrics to the key indicators which will provide insight into actionable interventions for clients propelling their revenue growth.

photo credit: alexliivet via photopin cc

2 thoughts on “Using the Median to Measure Performance for Social Media Marketers

  1. I love this post Susan. I show clients how to use simple math to gauge how they are doing over time and how they are doing compared to competitors.  Sadly you either are good at this already or you don’t want to do it. My favorite FB measure is how long it takes the average fan to interact with your brand on the page. It uses simple math and can be figured out in a minute. And you can tell who isn’t comfortable in the math world.

  2. susansilver says:

    Howie Goldfarb I might steal that time to respond to an update metric 🙂 As much as I love tools that figure out things for me, the companies that make those tools don’t share their methodology. If you don’t know the methods than you don’t really know what you are measuring and if it is valid.

    This month I’ve started culling the raw numbers from the accounts and doing my own math. It takes more time, but I am much more comfortable with results and reporting them to clients.

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.