Background - Going beyond Key Influencers
Gaining knowledge about the people engaging with your brand is a constant necessity across all networks. For Twitter, one of the publicly available data sets is Mentions. We have the full set of Mentions for the tens of thousands of Twitter accounts we are actively tracking. A Mention is any Tweet including a brand’s handle (@).
We use this data set in our Key Influencers metric. And while it reveals basic information about which users mention the selected accounts the most, it does not help with gaining further insights about these users unless you were to look up each one of them manually.
The idea of our Twitter Mentions Overlap metric is to take this same data set but go one step further: now we’d like to know which other accounts these users engage with. Our assumption is that the list of accounts a user engages with tells us something about their interests. The outcome is a relationship map between the selected accounts and any other Twitter accounts in our database that tells us more about the interests of our active users.
A basic example
Let’s assume the following scenario:
@userA tweets: “The best cars in the world are from @BMW”
@userA tweets: “I love the Big Mac of @McDonalds”
@userB tweets: “I have an issue with my new car. @BMW can you help?”
@userB tweets: “Just watched @BatmanvSuperman . What a great movie!”
@userB tweets: “I hate the new price increase by @McDonalds”
All brands (@BMW, @McDonalds and @BatmanvSuperman) are actively being tracked by our system because quintly users have added these accounts to the tool at some point in time. For actively tracked accounts, we import all Mentions on a daily basis, therefore we have access to all five Tweets above in our database. Content-wise these Tweets have nothing in common - both @userA and @userB mention a few different accounts in completely different contexts.
But let’s look at this scenario from a different angle: You are the Social Media Manager of @BMW and would like to know more about your active users. What we can do is look up all the users mentioning @BMW in a given time period. Let’s call these users your “active users”. In our example, @userA and @userB would be part of your active users. So to learn more about their interests, let’s look at what they do elsewhere.
This is where the following Tweets of both users come into play. We know the same user has also been in touch with other brands (@McDonalds for @userA and @BatmanvSuperman & @McDonalds for @userB). Next, we simply extract all the other accounts our active users have been in touch with. The outcome is @McDonalds and @BatmanvSuperman.
Beside just knowing which accounts they engaged with, we have a little more info: we can also count how many users have mentioned the same other accounts. Let’s do that for both @userA and @userB:
Other accounts mentioned by @userA: @McDonalds
Other accounts mentioned by @userB: @BatmanvSuperman & @McDonalds
@McDonalds was mentioned by @userA and @userB: 2 unique users
@BatmanvSuperman was mentioned only by a single user @userB: 1 unique user
So besides just knowing which accounts our active users have been in touch with, we can now also tell which “other account” has a closer relationship with @BMW - @McDonalds in this case. Our assumption is that an “other account” supported by multiple users represents a stronger relationship and thus should be considered more important when learning about our active users’ interests.
Interpreting the data
Here is how the Twitter Mentions Overlap metric looks within the tool. The account we are analyzing is once again @BMW. Visualized are the top 25 accounts with Mentions that overlap with @BMW during the selected timeframe.
@MercedesBenz ranks #1 with 782 unique users. Now what does this mean exactly? We can interpret this as follows: 782 unique Twitter accounts mentioned both @MercedesBenz and @BMW between September 1 and September 5. @Lamborghini follows in second with 621 unique users: 621 different users mentioned @Lamborghini and @BMW in the selected timeframe.
Note: The Mentions do not have to occur within the same Tweet.
Why is this metric limited to a maximum timeframe of 7 days?
Our current metrics only operate on data related to the currently selected profiles (via our Profile Selector). But while the same is true for the Twitter Mentions Overlap metric when extracting the active users, there is a next step which involves extracting the other accounts these users have been in touch with. This requires searching through our entire database of Twitter Mentions for tens of thousands of accounts. Because this is a huge amount of data, we currently limit the number of days that can be analyzed within this metric in order to be able to process all the data effectively and efficiently.