It's impressive that the internet can generate so much qualitative data through ordinary online conversations. Your Facebook page asks “how are you feeling?” LinkedIn asks for testimonials rather than rankings to endorse colleagues. We use comments as a way to evaluate the success of our blog posts. And while online reviews may show a star rating, it’s only by reading the full review that we can best appreciate the product’s true worth.
What Does All This Qualitative Data Mean?
For companies, it can be a challenge to process it all so that it translates into something more quantitative -- which is still the language spoken by the enterprise. This is where text analytics comes in.
How many of us have had an executive ask us “so what does it mean that the blog received a lot of comments?” or “what’s the ROI of Facebook comments?”
A decade ago, I developed surveys and performed statistical analysis for a non profit. Once surveys had been completed, they were scanned and the results were analyzed using SPSS -- the tool statisticians loved to hate.
At the time, I was asked to try out a new tool called Text Analytics for Surveys. For the first time, it was possible that I wouldn't have to manually transcribe the open-ended comments and painstakingly comb through them and analyze them separately from the data automatically collected from the surveys. Qualitative analysis was finally possible using SPSS, though it did still require manual set up and extensive coding for the words you wanted to analyze.
When I was in college, being a qualitative researcher was not as hardcore as being a quantitative researcher. While there may have been fewer statistics involved, it required much more work.
The time spent recording interviews with subjects, transcribing the text of the discussion and then having to assign keywords into categories so you could adequately summarize whether a respondent’s tone was more negative or positive based on the words used was laborious. The hardcore quantitative researchers could just run their data sets through SPSS and apply the parameters needed.
Fortunately, text analytics software has improved incredibly since I began developing surveys. Additionally, you no longer have to have a degree in statistics to use the software. Now, text analytics is integrated into CRMs so that all the conversations generated by your fans and followers can be analyzed to provide you with insights needed to strategize and evolve the customer experience.
What is Text Analytics?
At its most basic level, text analytics refers to the process of deriving high-quality information from text. The quality of the information is derived from patterns and trends, or some combination of relevance, novelty and interestingness.
Remember when IBM’s Watson competed on Jeopardy? It used Natural Language Processing (NLP), a component of IBM's text analytics offerings, to help figure out the context of the answers.
It's likely that you are already employing elements of text analytics. For example, you may have deployed sentiment analysis tools to help you predict trends based on the attitudes of users expressed on social media. However, the sentiment analysis process is not completely accurate because it doesn't account for the subtleties of sarcasm or body language.
Sentiment needs text analytics -- especially if you want to go beyond simple brand management like mentions of a company’s name. To dive deeper into how products and services are being received, you’ll need to apply advanced analytical rules in text analysis.
Just as predictive models analyze past performance to assess how likely a customer is to exhibit a specific behavior in the future in order to improve marketing effectiveness, text analytics can help companies infer similar insights from analyzing subtle text patterns to answer questions about customer performance.
CRM + Text Analytics
A recent Temkin Group report examined voice of the customer (VoC) programs within large companies and found an increase in the use of analytics, with nearly three-quarters of large companies with VoC programs using or considering text analytics.
What kinds of text analytics software are available within CRM or VoC platforms? Let's take a look.
Medallia's Text Analytics changes customer comments from a stream of anecdotes into “listening posts” that can be analyzed for trends. The platform can fully integrate with your CXM system, so you can use one system for both structured and unstructured data analysis.
The solution automatically identifies and prioritizes the most burning issues so that companies can target the highest-impact improvement opportunities. Homepage modules allow users to identify the most positive, and most negative, comment topics at a glance. Additionally, users can instantly see what new topics are emerging and drill down to get deeper insights on any topic they're interested in.
Lexalytics’ Salience Engine is a multi-lingual text analysis engine that is currently integrated into systems for business intelligence, social media monitoring, reputation management, survey analysis, customer satisfaction and more.
Its mining sentiment engine can measure the tonality of the discussions -- to show if there is a hot issue that needs to be managed, or to automatically route irate, important customers to a higher-level support representative. The engine includes several different pre-configured data directories to help process data more efficiently, including ones that are optimized for short content like Twitter, and one for longer-form content. Salience Engine can be integrated into your own application to handle text analysis and opinion mining in deep, fundamentally meaningful ways.
KANA Experience Analytics
KANA's Experience Analytics is a customer listening system that uses text analytics and advanced natural language processing (NLP) to analyze textual data -- including unstructured text -- across the channels your customers are using. The statistically-based analysis allows it to be "language agnostic," meaning it can work with many languages, dialects and slang -- the latter being especially useful for "Internet speak" and acronym-heavy SMS text messaging.
KANA text analytics can also correlate attributes such as age and gender with, for example, topics and sentiment, to help deliver a very precise understanding of customer conversations. Additionally, users can select automatic triggers that can identify and match incoming messages with the chosen criteria on a running basis, routing important comments to the CRM system and relevant staff for appropriate action and timely response.
Clarabridge provides a universal view of its customers through the transformation of text-based verbatim into customer experience insight. Clarabridge collects all sources of customer feedback, transforms it using Natural Language Processing (NLP), categorizes the content, performs sentiment scoring and delivers customer insights enterprise-wide though a variety of interfaces. With a full range of sentiment and text analytics functionality, Clarabridge helps to shed light on issues and compliments from the customer point of view in order to improve customer experience.
Preparing for Text Analtyics
These four software suites are just an example of the types of text analytics tools available. Before you find the solution that’s right for your company, there are few things you’ll need to sort out first.
What types of data will you access?
Gaining access to multiple streams of data can be challenging. Customizing the API is one thing, ensuring that you have the right to use internal or cross-company data stores is another. Before you select a vendor, it’s essential that all relevant parties, from IT marketing to your general counsel, are involved in the decision.
As well, because language isn't perfect, text analytics isn't perfect either. Thanks to words with multiple meanings, typos, colloquialisms and others, it’s important to manage expectations for the types of data that you use.
What are your expectations?
Text analytics can provide a lot of great, valuable information, but it’s not magic. While it’s definitely a more automated process than it once was, it will still require a deal of legwork before you start. Your company will need to build taxonomies to define the relationship between the terms a company uses so you can appropriately organize information into hierarchical relationships, making it easier to find and analyze text.
Additionally, you will need to clearly define the problems you hope to solve, or insights you hope to glean, by using advanced text analytics, otherwise it will be much harder to know if you’re targeting your goals appropriately.
How will you monitor the process?
While you don’t need to be a statistician to use these tools, managing the process will require some training. Text analytics involves learning and using a new form of data. No matter your skill set, it will take time to understand what to do and how to apply it to the business. Luckily, the text analytics industry is full of learning opportunities -- be sure to take advantage of webinars, conferences and classes.
There are many layers to optimizing the customer experience. As more people interact online, the way they interact evolves. As it becomes more sophisticated, it will require more advanced tools to understand what it all means. The conversations taking place online can impact the way your company empowers the customer. It's not just mindless chatter -- the words mean something and without text analytics you may not be able to read between the lines.