"Weight for Actions"

As an analyst, I do come across quite a few methodologies that brands and individuals adopt to measure their performance online. Some are great, some are unusual, and some are downright bizarre. I'll do my best to list as many as possible on and share my thoughts on them. The latest methodology I've heard of is the "weight for action" one: some brands have difficulties measuring their Facebook performance accurately, and  some people just can't get their head around what they should take from Facebook Insights.

So, what some do is assigning a certain weight to the various interactions that create what Facebook calls "stories". A story follows this formula:

X [action] Y

where X is the user, Y is the Facebook page or content related to it, and [action] is any interaction from the user to the page or its content - Ben likes Oreo, Jay shared Oreo's page, Paul commented on Oreo's post. These stories will appear in your news feed, news ticker, and on the user's timeline.

The most common [actions] are: like, comment, and share. However, the list of possible [actions] is a bit longer than that, including tagging, check-in etc.

What some brands do is assign a certain weight to those [actions], taking into consideration that share > comment > like. After doing so, they come up with scores for each Facebook post to see which one was more successful engagement-wise than others. This brings us to this formula:

(# of likes x weight of likes) + (# of comments x weight of comments) + (# of shares x weight of shares)

So, for instance, if we give the following weights...

  • Like = 1
  • Comment = 2
  • Share = 3

...a post that received 3 likes, 1 comment and 5 shares has the weight of (3 * 1) + (1 * 2) + (5 *3) = 20.

Though I can see the logic behind this (finding a universal engagement score to compare Facebook posts within and across pages), I have to say that from an analytical point of view this thinking is flawed. This only works under certain conditions, mainly assuming that everyone has the same number of friends on Facebook and that users share to all of their friends.

However, the reality is way different and the truth is that a share from someone with 10 friends holds less weight than a like from someone with 1,000 friends, and a comment from someone with 500 friends holds more weight than a share from someone with 50 friends, regardless of whether you consider weight as potential reach, engagement prospect, potential exposure, or a mixture of the above.

Take into consideration the settings around which shares take place. Just because you see that someone shared your content, it doesn't necessarily mean that all of his/her friends saw it, because:

  1. All of his/her friends may not have seen this on their news feed or news ticket, and
  2. Not all shares have the same privacy settings.

A user with 5,000 friends can share content from your brand page while setting visibility to just his list of close friends, or in the case of like-and-share competitions, some (like myself) might just click on share and set the visibility to "only me". Marketers and page owners will see that as 1 share, but they won't see how many people that one share was targeted to.

Not only does this method produce largely inaccurate results, but it also involves lots of data leakage, whereby a lot of data isn't accounted for.

It's clear that any weighting that you put on any [action] cannot be universal. If you need to accurately measure engagement, set what you're looking to measure first. Though some brands go for PTA(T) [People Talking About This], you'll find more useful measuring how many likes, comments and shares your posts receive - something that you do not have to do manually if you have systems in place to do so for you, like Adobe Social.

MeasureCamp and Growth Hacking

#Engage2013: SocialBakers