-->

2023 KMWorld Media Kit Available Here 

New tools provide deep insights from customer feedback

Article Featured Image

Customer feedback is of paramount importance to organizations in today’s highly competitive markets. Whether it is used to assess customer satisfaction, improve customer retention, or provide input to product development, this feedback provides an essential link between the customer and the company. Metrics, such as net promoter score (NPS) and surveys, in which customers rate their level of satisfaction with a product or service, provide valuable information, but this type of calculated number without context or explanation does not deliver insights into why the customer gave that answer. Text analytics offers a deeper understanding of customers’ responses based on their own words, or “verbatims.”

The text analytics market in 2021 was approximately $6 billion, and it is expected to grow about 18% per year, reaching $17 billion by 2027, according to the iMARC Group, a market research firm. This figure includes a broad range of applications, including customer relationship management (CRM), brand reputation, and fraud detection. Although only a portion of the text analytics market is accounted for by customer feedback applications, this sector is very dynamic.

Seeking a dialogue with customers

The Royal Automobile Club of Western Australia (RAC) provides its 1.2 million members with a wide array of services, including auto and home insurance, small business insurance, roadside assistance, financing, and home security. Given the volume of text that members provide in their feedback, RAC needed analysis at scale in order to enable its strategic objective of becoming more member-centric. Rather than coming up with ideas and promoting them, RAC wanted to take a collaborative path and seek more input from its members.

Developing a dialogue with its customers meant an increase in the use of open-ended responses, but although RAC had collected the data, much of it remained unused because RAC did not have the means to analyze it. After evaluating the tools available to accomplish this task, RAC selected Kapiche. This software solution integrates structured and unstructured data from multiple sources and analyzes it to explain what is driving customer responses.

RAC now has data-driven analyses, and the insights are arriving 30 times faster than they did with RAC’s prior manual method. In addition, engagement by departments and senior leadership has increased, partly because progress toward strategic objectives can now be measured. Finally, in just 1 year, RAC saved about 6 years’ worth of manual coding and analysis time, producing a highly satisfactory ROI.

Kapiche combines text analytics with quantitative analyses, so that the insights from written responses can be correlated with metrics that relate to factors such as demographics or purchase patterns. “Our product relies primarily on unsupervised and semi-supervised learning for text analytics rather than using an extensively pre-trained model,” said Ryan Stuart, CEO of Kapiche. “The goal is to allow flexibility so that the text analytics is not limited to what the system has been trained on.” Kapiche incorporates semantic search and uses SentenceTransformers for similarity search.

The NPS is one of the most commonly used metrics for cus- tomer feedback. Some studies show a relationship between NPS and corporate growth rates, but any single number rating has limits. “NPS is where our customers often are experiencing pain, because they don’t know the reason for the ratings,” continued Stuart. “Companies may have data from surveys, support centers, and online reviews, but can’t unlock it.”

Information being volunteered by customers across many industries has caused the volume of text to burgeon far beyond what was being obtained in the past via surveys. “A good response rate for surveys is 10%,” Stuart commented. “But people have many more avenues now to communicate with organizations, whether to write a testimonial, complain, or make suggestions.” The unsolicited data provides a rich field for text analytics, but without an effective means to analyze this abundant data, companies can easily be overwhelmed, or, more often, end up not using it.

KMWorld Covers
Free
for qualified subscribers
Subscribe Now Current Issue Past Issues