If you’re just getting started with text analytics, you have the advantage of learning from others who have taken on a similar task. Here are five tips to keep in mind:
- 1. Start Small
Those more experienced with text analytics recommend starting with a pilot project so that you can demonstrate ROI, get management buy-in and use the knowledge and experience you gain through the pilot to expand into other areas. If your objective is to enhance brand and product image in the market, create competitive strategies from competitive analysis, develop actionable items from the customer feedback and improve customer service, that’s probably too much to take on. Start with a smaller, focused, realistic goal.
2. Establish an Objective
Don’t expect to throw all your text data together, run an analysis and have answers come jumping out at you. As with every data mining project, you need to clearly establish the purpose and objective of the project – the key questions you’re trying to answer or business problem you’re trying to solve that is important to senior management. When you state your objective, manage expectations that the project will end with a series of black-and-white 100% accurate results.
3. Identify All Relevant Data Sources
Limiting yourself to one or two sources of data may not give you a complete view of the problem. The “voice of the customer,” for example, is heard in customer surveys and call center logs, in customer research such as focus groups and usability tests as well as in social media product reviews, blogs and tweets. Adding another source of data doesn’t necessarily mean adding more complexity to the project. Today’s tools are very versatile, and the same tools can often be applied to different kinds of data without any modification. At the same time, more data is not better if the data isn’t relevant to the objective you established for the project.
4. Don’t Limit Yourself To Just Text
A big portion of text analytics is pre-processing to convert text into structured, quantitative information that can be data mined. The greatest insight comes from combining these results with other quantitative information. In many cases, the quantitative data that accompanies text, such as the date of a comment, is crucial information you’ll want to use in the analysis.
5. Measure ROI
Translate data insight into actionable information that can be applied to the business and measured. To reasonably expect funding to expand text analytics beyond a pilot project, you need to be able to demonstrate a link to improved business results, such as increased sales. While the data you analyze may be from the past, you can’t change the past. Focus on future applications such as actions to predict, prevent or avoid future problems and quantify the positive effect of those actions on the business.
Subscribe to our blog to stay informed on text analytics and other data analytics topics.
Spotfire Blogging Team