Why Do I Need Semantics if I already Have Stars?

Last week we talked about how much time and effort is saved by sentiment analysis tools, which greatly reduce the amount of time and effort required to prepare your datasets for analysis and gain insight into what customers are telling you.

This week’s post goes into more detail about how Semantics tackles one particularly tedious data-related task: analyzing user-generated reviews related to your business that appear in external websites like Amazon or TripAdvisor.

A star is just a star

Third party websites like Ebay, TripAdvisor, Amazon and Booking.com have become busy communication portals that contain millions of customer opinions. Reviews are usually "scored" numerically or with stars, from 0 to 5 or 1 to 10. These values do gauge whether or not customers experience overall satisfaction with the product or service, but that’s it. Like or don’t like. Good or bad.

But what’s the point of knowing your customers have a low overall opinion of your brand if you can’t do anything about it because you don’t know what they’ve based their rating on? Being able to take action and use customer feedback to improve your business – so you can go from 2 stars to 5 – requires understanding the specific opinions, comments and feelings of customers.

The value of text-based reviews

Numeric scores fall short of useful because they don’t link opinions to topics. Nor do they tap into emotions, mentions of specific features, and all the other vital information that is available on third party sites.
When customers are unhappy with a specific service, love a particular product feature, or are generally pleased or displeased with your brand or other brands, they go right to the keyboard to spread their news and views. When harnessed, these opinions are a source of tremendous visibility for companies.

They point to weak spots, uncover competitive advantages and reveal other insights that can be used to improve products and services or develop relevant campaigns.
But getting at these insights is no small feat. Exhaustively reviewing massive volumes of customer reviews to find the ones that pertain to your company can take hours and hours when done manually. Not to mention pulling out specific opinions that are shared by multiple customers and thus point to real insights.

Enter: Semantics

Semantic tools simplify and speed up the task of gaining insight from the volumes of customer feedback generated by third party sites. They extract relevant information and organize it in a way that makes it easy to understand exactly what customers are saying. They also give a much more accurate picture of what customers are expressing than stars or numeric ratings.

Consider the following example. Let’s say a user has given a smartphone a numeric score of "1 out of 5," which would suggest that the user is very unsatisfied with the phone. But the review then reads, "This phone is wonderful. The camera is impressive, it has a perfect resolution and, on top of that, it is cheap. The only problem: the GPS works hopelessly bad. So, this is useless for me. A pity." What a difference it makes to drill down into the actual comments!

Not only do we know that our GPS feature needs improvement, we know that it is important enough to one of our users that he dropped a potentially 5-star review down to a 1. And we know that our camera, resolution and price are competitive advantages. This is information we can do something with.

Sentiment analysis tools are extremely intelligent. The identify entities (brands, people, products, places, etc.) and also concepts (customer service, price, cancellation policy, etc.) They organize opinions by topic, so you know exactly what it is they like or don’t like, and what you can do to improve, push up your ratings and win over customers.

For example, in the sentence “This phone is awesome, but it was much too expensive and the screen is not big enough,” a semantic tool will extract three opinions:

  • [“phone” + “awesome”],
  • [“phone” + “much too expensive”] and,
  • [“screen” + “not big enough”]

As we can clearly see from these examples, leveraging semantic tools to break down opinions in this way gives companies a big advantage.

Download a Real Case


Subscribe Here!