Why Your Text Analytics Solution Is Failing You
September 24, 2014
When done well, text analytics helps enterprises weed through massive amounts of customer feedback, in real time, to gain insight into the customer’s wants and needs.
Unfortunately, text analytics is not always done well.
Many organizations suffer because they don’t understand the nuances of text analytics technology when it comes time to choose a vendor. It is a very sophisticated field, and it can be hard to know what to look for. Here are some of the key ways that an inadequate text analytics solution can fail:
1. Weak technology:
The field of text analytics is not brand-new; manual approaches (reading, scoring, and sorting of text by real people –probably interns) have been around since the 1980s. Today’s technology is far superior, but not every vendor is using the most cutting-edge techniques. You could be failing to retrieve the meaning of huge portions of your text if your vendor is using any of these outdated methods:
- Keywords – Just flagging the terms you know you should be looking for – so you never know what you don’t already know.
- Statistics – Ranking the frequency of terms present in the text – so you know “what” people are talking about, but with no context.
- Triples – Identifying three-word or three-phrase combinations – so you have a very accurate understanding of some of your text, but no insight into anything that doesn’t include the defined triples (up to 50% of your data!).
If the text analytics technology you are using can’t be specifically tuned to meet your needs, you will get inaccurate results – and you can’t make confident business decisions if you have inaccurate data. Inflexibility can relate to your business model, if your text analytics solution cannot adjust to the specific terms and phrases used in your industry.
Inflexibility is also a problem as it relates to your data sources. If you can’t account for the differences in text that comes from your customer service notes (with your internal jargon) versus social media (with its specific slang, hashtags, and abbreviations), then you are not going to have a cohesive account of your customer’s views.
3. Unequal language support:
Some vendors claim a huge list of supported languages; however, analysis of less-widely-spoken languages is often done after an imprecise translation. In order to analyze text effectively, the text analytics engine must support those language natively. This ensures that shades of meaning, grammatical construction, and language-specific idioms are not misinterpreted.
4. Limited data sources:
The text analytics solution with the most value is the one that can accurately analyze the most data. If your solution only processes survey data, or social data, or internal CRM data, you are not getting the big picture.
Is your text analytics solution failing you in one of these areas? If you don’t know, or if you are just starting to look into text analytics and don’t know where to start, the Cheat Sheet “Eight Questions to Ask Your Text Analytics Vendor” can help.
Lisa Sigler is Sr. Manager of Content Marketing at Clarabridge. For over 16 years, Lisa has used her writing and editorial skills to bring the value and benefits of technology to life. In her current role, she works to demonstrate Clarabridge’s position as thought leader and trailblazer in the Customer Experience Management market. Lisa holds a B.A. of English from Kent State University. Read more from Lisa on Twitter @siglerLis.