More Facebook Insights: How To Calculate 30 Day Metrics
Facebook provides a wealth of data that can be used to calculate many useful metrics not readily available in the standard Insights report (see last month’s column for three examples). Unfortunately, Facebook does not segment data on a monthly basis. All Facebook data is segmented into daily, 7-day, or 28-day time periods. This means that […]
Facebook provides a wealth of data that can be used to calculate many useful metrics not readily available in the standard Insights report (see last month’s column for three examples). Unfortunately, Facebook does not segment data on a monthly basis. All Facebook data is segmented into daily, 7-day, or 28-day time periods.
This means that Facebook metrics can’t be easily incorporated into monthly reports. This is a problem not only for custom metrics that can be calculated using Facebook data, but also for the standard metrics displayed in the Insights interface.
For example, the Facebook metric “People Talking AboutThis” would have to be translated into a 30 day metric to compare Facebook engagement over time to engagement in other channels.
So, how do you translate Facebook data into 30-day metrics? Let’s take a look at one of the custom metrics covered in last month’s column, Average Organic Reach Per Post:
We used the 28-day organic reach number for this metric, but remember that Facebook provides a daily and 7-day organic reach number as well. It is tempting to assume that you could simply sum the daily organic reach number for a 30 day period and divide it by the number of posts in that time period in order to calculate the 30 Day Average Organic Reach Per Post.
Obey The Fruit Salad Rule
Unfortunately, this does not work. Here’s why: unique metrics cannot be added or averaged (at least, not if you want an accurate calculation). Let’s call this the “fruit salad rule,” because that’s what you’ll be serving up if you are slicing up lots of apples and oranges mixed together.
Reach is a unique metric. It reflects the number of unique people you have reached with your page and/or posts. If you sum the daily reach number for 30 days, your results will in all certainty contain numerous instances of the same person being counted multiple times.
In other words, a person who visited on Monday and then visited again on Wednesday would count as two unique visits when in reality it was only one. This may pump up your Facebook numbers somewhat, but at the cost of accuracy and credibility.
The fruit-salad rule applies to another custom metric that we covered last month, Engagements Per Engaged User. A “user” is based on a unique number, so this metric cannot be calculated on a 30 day basis. (Note: it would probably be more accurate to call this metric “Stories Per Person Talking About This”, since an “engaged user” in Facebook is defined as someone who clicked on your content.)
In fact, any data point provided by Facebook that is based on a unique visitor cannot be turned into a 30-day calculation by simply adding the daily values provided.
Plenty Of Metrics Aren’t Subject To The Rule
The good news is that there are plenty of Facebook metrics that are not based on unique numbers; these metrics can be translated into 30-day time increments. For example, the custom metric Average Engagement Per Post (also discussed in last month’s column) can be calculated on a 30-day basis by simply summing “Daily Page Stories” over the 30-day period and then dividing that number by the number of posts in the 30-day time period:
If you use the Facebook Insights interface, you will notice that many of the metrics provided are based on unique numbers. Fortunately, the Facebook API often provides a non-unique metric that corresponds with the metrics visible in the Insights interface.
Three examples are listed in the table below. Metrics listed in the “No” column (visible in the Facebook Insights interface) are unique metrics and thus should not be translated into 30 day metrics. Metrics listed in the “Yes” column are the corresponding, non-unique metrics found in the Facebook API, and thus are suitable for 30-day calculations.
By using the non-unique version of each metric (i.e., a metric listed in the Yes column), you can easily create a set of 30 day metrics to incorporate into your monthly reports. For example, instead of using Average Organic Reach Per Post, you can calculate 30 Day Average Impressions Per Post, by multiplying daily Impressions by 30 and then dividing that product by the number of posts in the 30 day period.
Facebook’s segmentation of data into non-standard time periods highlights a common problem in social media measurement: disparate types of metrics, definitions, and reporting protocols used by a growing pool of social sites make cross-channel social media measurement (not to mention integrated marketing measurement) challenging.
Until standards emerge (which would seem to be some way off), marketers will have to be creative about finding ways to compare apples to apples. Until then, some of the unwary will be serving up quite a bit of fruit salad.