It’s pointless to analyze the wrong conversations, just as you would never survey the wrong audience. But that’s what most of us do when we analyze social media conversations.
We don’t analyze the conversations about the subject we need or the audience we need. We’re stuck matching keywords. So, when mobile phone network Sprint wants to know what consumers are saying about them, what happens when they analyze all the conversations containing the word sprint? Nothing good, actually.
Instead, ConveyAPI offers relevance classifiers that allow you to look at all those conversations and say “yes” to conversations about the company Sprint and “no” to all the conversations about NASCAR sprint car races and every high school’s 40-yard sprint. Don’t settle for Boolean searches that force you to search for sprint AND NOT car - they omit the relevant tweet that says “Sprint drops my call every time I am in my car.”
Now that you know how to identify the right conversations, you must extract the insights that drive your business decisions. ConveyAPI provides numerous insights from analyzing your relevant social conversations:
Sentiment. Understand whether social conversations are positive, negative, or neutral.
Emotion. Understand the emotions an author is expressing toward the subject they are discussing (based on Robert Plutchik's influential 8-category classification of human emotions).
Intensity. Understand the strength of the author's attitude toward the subject they are discussing. Sprint might care a lot more about intense comments that say, “Sprint is the worst carrier of them all” rather than “Sprint has fewer phones available than Verizon.” Both statements are negative, but intensity matters when you test opinion.
Custom insights. In addition to these “out of the box” insights, we work with clients to build custom classifiers designed to address industry - or domain-specific business questions.
Now that you know how to identify the right insights for the right conversations, don’t settle for analyzing the whole conversation when you really need to understand the sum of its parts.
If you’ve just released a new product, and dozens of blog posts have been published with reviews of every feature in your product, do you really want to know if the entire review is positive or negative? Most of the time, reviews contain both positive and negative comments, leaving most listening solutions calling those reviews “neutral” - that doesn’t add much insight.
While ConveyAPI can certainly work at this document level, you can also recast your conversations as a series of sentences and extract insights from positive and negative sentences. In fact, you can identify a specific product feature (Sprint might choose battery life or screen size) to extract insights on what the reviews say about those specific product elements.