Profiting from multi-dimensional sentiment analysis
How to translate sentiment analysis into business actions.
Imagine you run an online store selling bags. A customer bought a bag and wrote a review. He gave it 5/5 stars. Additionally, you implemented a sentiment analysis which tells you, that the customer is very positive about the product.
What can you do next, to improve your business? Well, it depends on the details of what the customer actually wrote in his review. And what is important and which actions to take strongly depends on your business.
While working with our customers, we discovered that it is always necessary to develop use-case specific solutions. This is especially true for sentiment analysis. A sentiment model that we developed for one customer in one domain usually doesn't work well for other domains. But there is also more than that, when it comes to classifying emotions.
For example, what actually means positive? Positivity is a broad term for many different kinds of emotions. Most of the time, it turned out that a one-dimensional sentiment analysis, i.e. positive vs. negative comments, is not sufficient.
In the case of your bag shop, you might want to differentiate between happy and excited reviews, which are both positive ones.
If a customer is happy, it means the customer has your trust and would buy another bag from your shop. You should strengthen this trust and also encourage the person to buy another bag. Like in the following example:
On the other hand, if a customer is really excited, you might want to do something different. For example, you could use him as a promoter for your shop.
That means you don't just want to give him a voucher. Instead, you could give him a personal promotion code, that he can use to invite friends to your shop.
And you put an incentive on top: With every friend that buys a bag in your shop, he earns something himself. For example like that:
Similar to positive reviews, not all negative reviews should be treated the same. In our case, we might want to differentiate between angry and sad customers.
If a customer is really angry, this means
a) that you have lost him as a customer and
b) that he is spreading a negative image of your brand.
You should set up an alert for such reviews and act immediately!
This means: Make sure that the problem is solved to stop negative promotion. And additionally, do something so that he might give you another chance. For example:
Sad customers are customers that were excited and trusted you. And you disappointed them.
You are about to lose their trust. Make sure that this doesn't happen! Like in this example:
Sentiment analysis has many more facets than classical positive vs. negative analysis. More importantly, you would take different actions in your business if you had a multi-dimensional sentiment model that is adapted to your purpose.
If you are interested in customized multi-dimensional sentiment analysis, please contact us here.