Interpreting The Consumer Story: Text Analytics in Market Research
Text data exist all over the internet. Emails, forum questions, social media posts, customer service requests, answers to open-ended survey questions, and even this blog post are all sources of raw text data. Text data are drawn from anything written by a human being, and these raw text data can be analyzed to reveal insightful trends.
What Is Text Analytics?
Text analytics, also known as text mining or text data mining, is a useful process for extracting high-quality, consistently analyzed data from raw text. In text analytics, a computer script automatically analyzes raw text data for patterns or classifications. One simple example is the extraction of basic frequencies of certain words. For instance, you might be interested to know the frequency at which people mention your brand in relation to other brands.
Another, and more interesting, example is the frequencies at which words occur with certain distances of other words. For instance, you might want to know which words occur near the phrase ‘customer service' most frequently. Are consumers writing words such as ‘great,' ‘responsive,' and ‘friendly' near mentions of customer service, or are they saying something else? Words which occur frequently together are termed ‘collocations,' and represent a potentially rich source of text information.
Another option is sentiment analysis, which tags comments as positive or negative. This information can prove useful for determining how a consumer feels about a product or company, or to determine which other text-based features correlate with a positive or negative sentiment.
How can text analytics benefit businesses and market researchers?
Market researchers already often gather raw text data in the form of open-ended question responses. They may also have access to text data ‘in the wild' (such as social media posts) or to client text data (such as forum questions or customer service emails) if a client chooses to share this information. A computer script can analyze such raw text data for interesting trends. One basic technique is to use frequencies to generate the familiar ‘word cloud' infographic. But with more advanced techniques, there are many options available beyond a simple word cloud.
Businesses can know overall sentiment, correlates of sentiment, frequent collocations, writing style, and even more, and use this knowledge to answer questions about consumers' beliefs and drivers behind these beliefs.
In the past, open ends have always been read by human researchers, and human interpretation does have its advantages. Although humans read slowly, they make accurate and intuitive classifications. Even with new technology and research methods, this remains a valuable skill. Machines, however, can read in large data sets and make quick analyses, enabling the human researcher to process a large quantity of data in a short period of time. Machines also apply identical analytical methods to each data point, leading to good precision. Human programmers and researchers can then apply their experience intelligently and use text analytics to draw informed conclusions. This makes for an ideal use of technology in market research.
Leveraging text analytics as a combination of human intelligence and experience coupled with machine speed and precision can open doors to new insights and quicker discoveries for companies interested in understanding how consumers really feel when they type out a message.