One of our strongest USPs on Pulsar TRAC has always been our network and graph capabilities. We know how central these elements are to your work for surfacing insights out of large datasets to help shape and inform your creative or audience strategy.
That's why today we're delighted to announce Segments 2.0, a hefty improvement to our initial Segments release in May last year, that will allow you to easily cluster conversations (or people), by how connected they are or how they frequently occur together.
It doesn't matter what data you are looking at! Whatever you got, we can cluster it: a group of authors, bios, images, topics or keywords, you name it, our network analysis algorithms can identify and group any of these data formats into informative segments.
You will find this feature, as always, under the Segments tab in the following sections: Audience (bios), Keywords, Topics, Content (images) and Influencers (network).
So what's new? We're glad you asked:
we've made substantial under-the-hood improvements to our clustering algorithm, which now can even better identify and group the nodes (authors, bios, images, influencers, topics, and keywords) into distinct segments;
and we created a visualization layout that is even more intuitive and easy-on-the-eyes, allowing you to easily take in the structure of the conversation.
Example: in the above segment from a conversation about vegan recipes, we see which types of posts are most common about the subject. In this case mood relating to the recipe (love), and ingredients (meat - likely substitutes here). This can help you build a data informed content strategy that aligns with your target audience's interests.
Let's dive in bit further.
You'll notice that we've updated the Keywords, Topics, Images, Bios and Influencers network graphs, resulting in a much cleaner, easy to read, and easy to export visualization.
Each segment we identify has its own colour assigned to it, and is now clearly labelled for easy identification.
When in the Keywords, Topics, Images and Bios tabs, we show you the nodes that are most connected and occur the most within the same context.
These are then grouped into segments, which we label based on the top 3 nodes that we've identified through PageRank (more on this later). This is a significant difference from the previous version of Segments, where we applied Betweenness Centrality.
Using the PageRank algorithm not only makes it easier to understand the nodes with more importance and relevance in a conversation, but it also makes it easy to identify the main behaviours in that segment.
Finally, when you click on each node, you can now see the sample posts related to that term, giving you more context on what's being discussed.
The Influencers Network graph has also seen some significant improvements.
For starters, we've updated the graph's visual layout and now show you the sample posts by each author in the network, along with a profile summary of that author, as shown below.
What now connects the authors and groups them into specific segments are their engagements, i.e. (retweets, replies, comments).
And unlike the previous Influencer Network graph, you can now effortlessly identify each segment in the graph, which consists of an original author, and the authors directly engaging with that post.
With this new clustering, it's simpler to see the connections between who is reacting to whom, and whether the authors reacting to an individual's posts are all from the same segment, or belong to a different one.
You can also see who is most influential in terms of their reach, by looking at their connection to authors from other segments, as shown below.
How should I use Segments?
Well it all depends on your search and use case, but here's a few ideas to start off:
Use it as a discovery tool, to discover your macro vs micro influencers: find out not only who is producing the highest level of engagement in your search, but also who is well connected to other segments, and therefore more influential in spreading a message through different segments.
Use it for content recommendations, for example as guidance on producing content around the key subject areas identified in the graph. You can even go further and look at this by specific social channel, and see how for example vegans on Reddit talk differently from vegans on Twitter. If you are using our image recognition AI, you could also use the image segments graph for some inspiration and get those creative juices going!
Use it to sift through the noise, and find the most relevant information in your dataset. For instance, if you've used our Communities functionality to break down your audience into communities of interests, you can carry over that filter and apply it to visualize intuitively how that community's conversations map out.
..and many other use cases that you will uncover as you begin to use this. We're excited to see what you do with this feature!
Any questions about the above, hit us up on the in-app chat, and for any feedback on the feature, feel free to add a suggestion on Collaborate.
✌🏽
Want to learn more: