Steve Jobs once said, “It is not the customer’s job to know what they want.” It’s up to you to discover and deliver what they want, when, how and where they want it.
Fortunately, most companies are already sitting on this knowledge. Their customers have provided it in the form of unstructured text feedback such as product reviews, open-ended survey responses, and social media posts.
Their biggest challenge is processing all that feedback in order to extract those key insights. How do you even begin to analyze hundreds of thousands of product reviews or tweets, short of hiring an army of interns to read through and summarize them all?
Fortunately, that approach isn’t necessary. (Interns everywhere are breathing a sigh of relief.) Natural language understanding (NLU), which uses artificial intelligence to help computers understand language like people do, enables companies to analyze their customers’ text-based feedback and identify patterns and trends within that data. Those insights can then be leveraged to improve the customer experience.
So how can you ensure that your insights from text-based data will actually improve customer experience? The secret is ensuring that your unstructured analytics program incorporates NLU and is also built into your business processes. Here are our five tips on how to make that happen:
Involve stakeholders and decision-makers from the beginning
Before you begin any analysis, it’s critical to identify and involve the teams who will use your insights to directly impact the customer experience. Including these teams from the start will ensure buy-in to the process and an eagerness to act upon the results. Some examples of teams you may want to work with include customer experience, market research, marketing, and product management.
Analyze data from every source available
At most companies, the analysis of customer feedback is not centralized. Multiple departments, including call centers, market research, social media analytics, and marketing often collect their own data. This feedback is typically only analyzed by the department that “owns” it… if at all.
It’s critical to compile and analyze data from all of these sources collectively in order to understand the full picture and have proper context for your findings. This is where involving stakeholders from the very beginning (Tip #1) can be especially helpful. Partner with peers from across your organization to make sure you’ve gathered all available data before beginning your analysis.
Don’t get hung up on hypotheses
You know your product or service better than anyone, so you understand what your customers need, right? Wrong. A common mistake is to develop hypotheses and then search through customer data to find evidence supporting or disproving them. This approach frequently leads to bias and can cause companies to inadvertently overlook important but unexpected insights in the data.
While
it’s inevitable that you’ll have some hypotheses about what you’ll find in your data, make sure that it doesn’t lead your analysis. The data should be leading you to relevant insights, not the other way around!Combine insights from your text with quantitative data to get the full picture
Make sure you’re leveraging every piece of data you have! Text data is invaluable for understanding the “why” behind customer trends and for pinpointing specific drivers of customer satisfaction and loyalty. Still, your quantitative data can help bring more context to the insights and trends you find in your unstructured text.
Get ready to act fast!
Your customers can reach you instantaneously via social media and email, and they expect an equally prompt response. The good news is that natural language understanding solutions are based upon artificial intelligence. As a result, they require less lead time and fewer resources to implement compared to many other types of traditional text analytics. Once you begin your analysis, insights will follow at breakneck speed, so get ready to respond!
To learn more about how Luminoso is helping customer experience teams get the most from their feedback, check out our solutions page.
