A current hot topic of discussion is “how do you measure social media engagement?”.
In developing the Ackura PressRoom product I’ve built a tool that collects a lot of data but we are now in a position where we need to work through what all of these items actually mean.
There is undoubtedly a need to aggregate the data into some sort of metric or set of metrics which can be used as pointers towards positive action.
It’s all very well being able to say this video received x thousand views or that blog post recieved y comments but all that is doing is generating data when what we really want is information.
As an example, the PressRoom product publishes a Social Media Release which includes media items such as images and videos and also related links to other websites or pages. We use social sharing sites such as YouTube, Flickr and Delicious for hosting the attachments and to increase the reach of the content. These sites also allow other members of the online community to comment on, rate, tag or share these items – all of which generate some sort of metric.
As a starting point, here are some of the data items which we collect:
- Inbound links to the news release (via Yahoo)
- Blog reactions to the news release (via Technorati)
- The number of people who bookmarked the news release (in Delicious)
- For each image attachment we have:
- The number of times the image was viewed
- The number of comments for the image
- For each Video Attachment
- The number of times the video was viewed
- The number of comments for the video
- For each related link
- The number of people who bookmarked the link
This gives us lots of data to work from even though this is still a simple list. We need to decide what sort of weighting to give items e.g. should the number of people who bookmarked a related link have a lower weighting than the number of people who bookmarked the news release? The answer is probably yes but there is still the question of what ratios to use.
We have not yet considered the cases of user ratings (e.g. for Videos on YouTube) or people adding a photo or video to their favourites. Again, is a highly viewed but low rated video more effective than a video which doesn’t get as many viewers but is consistently highly rated?
All of these metrics produce numerical data which makes aggregating them simply a mathematical problem, however there are a whole other set of metrics to think about.
On many social sites, people have the ability to add their own tags to the items, as a way of categorising the content in a way that makes sense to themselves. We should be able to interpret information from the tags which other people assign to our content.
Finally there are the comments themselves, in terms of generating metrics, we could use textual analysis to determine the sentiments expressed in much the same way that Summize analyses Twitter posts. This is in no way a replacement for actually getting involved in the conversations, but it can help to quickly spot where, amongst the many conversations taking place, you might want to join in e.g. a discussion which is particularly critical of your brand may simply require talking with the participants and asking what you can do to improve.
Thankfully, the woolly nature of these measurements give us the freedom to build abstract metrics in which we are more concerned with getting an overview than absolute accuracy. I expect this to be a subject into which we will be putting a lot of thought, getting a lot of outside opinions and updating our methods on a regular basis.
