The sentiment potential tells you by how much a specific slice of your data can increase the global sentiment index. It is calculated by assuming that all the neutral and negative sentiments of that slice of your data were positive, and by how much this would then increase the global sentiment index.
To see how it can be useful, let's look at example distributions of sentiments across some categories.
Here we have three categories from the world of retail:
quality of products
availability of products
pricing of products
We see the sentiment distribution per category and their respective sentiment index. For example, the pricing category has a sentiment index of -13.5, indicating that there are slightly more negative than positive comments about the pricing.
Meanwhile, the availability category has a sentiment index of 23.9, meaning that there are more positive comments.
One might assume that pricing is thus the main issue here, however, note that the size of the categories is ignored when looking at the sentiment distribution alone. In an extreme case, we may have a category with purely negative comments, but only two members. Such a small category surely has little to no impact on the overall sentiment of your data.
Back to our example, we see that much more comments were made about the availability of products than about their pricing:
The sentiment potential hence combines these two important distributions to express in a single number how much potential a category has to improve the overall sentiment of your data.
Thus it gives you a clear indication on which issue you should focus on first.
By working only on the availability of the products, the sentiment could be increased by 25 points. Improving the pricing of the products could increase the sentiment by 12.5 points. Still a large margin, but only half the impact that the availability has.
Thanks for reading!