When it comes to science, “thought experiments” can be just as useful as highly controlled experiments done in labs. These hypothetical case studies present problems that allow for creative solutions we can use to ensure our tools are capable of anything our customers and partners might need.
For example, you run a restaurant in an area with a lot of foot and car traffic, ample parking, and your food is well-reviewed. Despite all of this, your sales are slumping, your restaurant is never full of customers. What are some of the ways a business owner could address such a problem?
The most likely solution would be to run a new ad campaign on a legacy channel, promote your restaurant on Yelp, and perhaps buy a marquee for the front of the restaurant. The goal here is generate demand and bring in new business. Advertising is foundational to every business strategy, but what does it say about the business if the ad campaign doesn’t work? Try again? Move locations? Change the menu? Sell the business and move down to Zihuatanejo with Red and Andy Dufresne?
While those are all viable options, they are proverbial shots in the dark because you don’t know what the problem even is. Changing the menu could make things worse for your restaurant, especially if the menu isn’t what people objected to. This is where text analytics and sentiment analysis can help.
Using named entity extraction, our tools identify reviews and social media posts that mention your business specifically. It will also recognize the names of your competition, highlighting how your customers are comparing you to them.
By utilizing categorization, you can use text analytics to determine the root cause of the problem—Is it décor? Maybe it’s the bathrooms? Perhaps the service isn’t up to snuff. You can target specific complaints about the food, prices, the location, or anything else you think your customers might be upset about.
What if the problem is something you never even saw coming? While you may customize categories manually, our tools identify most common themes in the data automatically out of the box. In this hypothetical case study, let’s imagine all the specifically targeted categories show neutral or positive sentiment, but there is a surprise negative theme around the term “bathroom.” Using text analytics to identify sentiment bearing phrases in these reviews, you can discover not just that customers are talking about the bathrooms in the restaurant, you can discover exactly what they dislike about it.
Are the customers saying the bathroom is too dark or not clean enough? If so, then you don’t even need to incur the expense of an ad campaign. Instead, emphasize restroom cleanliness with your staff. Perhaps install an attractive external vanity counter for diners to use for washing their hands and checking their appearance so as to maximize access to the lavatory.
Remodeling the bathroom might be a bigger project than you wanted, but it’s a much better solution than running an expensive ad campaign, or, worse, closing or moving locations, especially when you don’t have to. Or In the days before social media and online reviews, companies paid a premium to collect the kind of customer data that is out there on the web just waiting to be mined.
Stay tuned to the blog for more case studies like this, backed by real text data analysis. In the meantime, why not reach out to our experts to see how text analytics can be applied to your business.