We thought long and hard about how we are going to do sentiment analysis the right way. The challenges? No computer algorithm can detect things like sarcasm or dual-subject sentiment (a negative comment for one party is a positive for the other, but try mixing them in the same sentence).
There is only one fail-safe way to analytically calculate sentiment of your mentions and to get a result which is relevant and more importantly reliable.
You’re the person who knows the subject of your most important keywords best, and you alone are the only person (or machine) we would entrust to grade the sentiment correctly.
That’s what most of our users told us and that’s what we did. Next to each mention, there are three emoticons – click on them to set the sentiment – positive, neutral or negative.
Try grading a few, you’ll see how tricky it can be sometimes in the gray areas between neutral and positive. Would you still entrust a machine to understand the subtle hints in a 24-character-long Tweet? We wouldn’t either.
When you’ve clicked through your mentions (or the most important ones, at least), you’ll get a pie chart analyzing the sentiment around your keywords. Just click on Report and scroll down, past the sources analysis.
A More Advanced Sentiment Analysis
For a more in-depth analysis, we recommend grading the sentiment, then exporting your mentions and grouping them into topics (example: your product’s features, customer satisfaction, the event you attended last month and general), so you can get insights like “My customers love our new features, but our support is not performing up to par.”
So, there you have it. You can have an in-depth, data-driven report about the buzz around your brand, your competition, or your client done in a matter of minutes, ready to be presented or sent.