Information without context is rarely insightful. Yet many text analysis methods fail to adequately capture context. Instead, outputs like word clouds, sentiment analysis and topic wheels provide a one-dimensional view of a topic. Relative Insight is tackling this challenge with a comparative approach that delivers scalable, contextual text analysis and insights for market researchers, marketers and customer experience teams. Our newly enhanced ability to showcase verbatim examples makes it easy to communicate this context to stakeholders.
Data vs information vs insight
The terms information and insight are often used interchangeably. While both are derived from data, there is an important distinction between the two. What distinguishes an insight from a piece of information is context and a ‘so what?’ that can be used to drive action.
For example, knowing 200 survey respondents complimented your in-store experience is a piece of information. To make this piece of information insightful we’d need some more detail:
- How many total survey respondents were there?
- Were you more or less likely to receive praise for the store experience compared to last year?
- Which stores are receiving the compliments?
Context is created by building connections between multiple pieces of information. Only once we have answers to these questions should we feel confident making a judgement about the significance of the initial piece of information (the ‘so what?’). Once both things have been done, you will have successfully turned information into a contextual insight.
Contextual text mining
Relative Insight’s technology is designed with a focus on capturing context.
Our unique comparative approach means the results of an analysis will always be presented in relation to a relevant baseline (e.g. a prior time period or competitor). By delivering an analytical approach that captures context by design, we’re helping customers bridge the gap between information and insights. This approach saves analysts and researchers time while simultaneously helping them generate more robust, high-quality insights that hold up under scrutiny from stakeholders. Ultimately, this results in a higher proportion of insights being used to drive real-world actions – a top priority for any organisation wanting to become more data-centric.
Our mission to help customers uncover contextual insights doesn’t stop with our comparative approach. The reputation of text analytics has historically been hindered by concerns over the accuracy with which algorithms can understand the context in which specific words are being used. This challenge is a direct reflection of the nature of language in which words can take on context-specific meanings. Consider a few examples:
- ‘Park’ can refer to stationing a vehicle or a public green space
- ‘Right’ can refer to a direction, a 90° angle or a legal protection
- ‘Close’ can refer to the act of shutting down, or being near someone or something
To combat this challenge, our natural language processing algorithms that read and analyse uploaded texts leverage AI machine learning to make informed judgements about the context in which a word is being used. This contextual approach to text mining is done by looking at the entire construction of the text.
The secret to effective data storytelling
Generating contextual insights is important, but sharing this context with stakeholders is equally crucial. Doing so is an essential element of effective data storytelling that compels stakeholders to act. The ability to articulate context is essential to generating confidence that an insight accurately reflects the situation in which stakeholders are operating.
Our most recent platform update has made it easier than ever to select the most relevant examples for your stakeholders. Users can now select specific snippets of an analysed message to include on insight cards – helping them highlight the things that matter most.