What we do
Relative Insight is a comparative text analytics software that helps organisations generate actionable insights from text data using technology originally developed for law enforcement.
Most decision making relies on quantitative measures that tell you what is happening, but taking effective action requires an understanding of how and why. We believe that qualitative text data – i.e anything with words, represents one of the biggest and most valuable – but often, most under-used sources of business intelligence. It’s one thing to understand the what, but that’s irrelevant unless you understand why things happen, how people feel, and how best to engage with them.
Traditionally, qual data has always been messy, dispersed and deemed too tricky to analyse, so we approach it with our key principles in mind:
- Comparison is key. The differences matter.
- We can analyse text data from any source. If it’s words, we analyse them.
- Text is a strategic and valuable data asset. Our customers store and combine text data for new insights continually, producing at minimum five times more value from one data source.
How does Relative Insight work?
Relative Insight’s platform combines AI-powered natural language processing with advanced comparative linguistics to analyse any source of text data and drive enhanced contextual understandings of target audiences, competitors and trends. This approach reveals what makes data sets (and the audience groups, brands or products they represent) unique and similar.
What is Relative Insight used for?
Voice of the customer
Analyse survey open-ends, reviews, interviews and other customer feedback to inform marketing, customer service and product development
Market research
Transform relevant public discourse from social media, forums and the media into valuable market research insights.
Competitor intelligence
Compare website copy, advertisements, customer reviews and public discourse to understand how competing brands and products are different and similar.
Extract maximum value from your text data
Too often, text data is analysed in a ‘one and done’ fashion. With Relative Insight, the average piece of text gets analysed in five different ways. Using the split functionality to breakdown your data sets, you can develop 360° understandings of the topics you are researching and ensure you are getting maximum value from your data assets.
The challenge with traditional text analysis
The text analytics capabilities commonly embedded in social listening and survey analysis platforms are one-dimensional and typically rely on frequency-based analysis. This approach involves counting the most common words and phrases and presenting the findings in in word clouds or frequency tables.
There are two problems with this approach:
Topic-specific language
The most common words and phrases aren’t always the most insightful. Frequency analysis tends to surface obvious topic specific findings.
Supermarket reviews are likely to include many mentions of the words ‘shopping’, but this isn’t interesting or surprising.
Lack of context
Context is necessary to judge the importance of a finding, and one-dimensional frequency analysis does not incorporate any element of baselining.
It’s great to know 20 customers praised the quality of your product, but is this better or worse than the competition?
The power of comparison
All analysis in Relative Insight is organised around one or more comparisons. Comparisons can be built across time, audience segments, geographic markets, competitors and more.
This approach overcomes the challenges of traditional text analysis methods in two important ways:
Insights that capture context
- A comparative approach produces high quality insights by enabling you to learn about a particular audience group or brand in relation to a relevant reference point or baseline.
A focus on the differences
- Relative Insight cuts through the noise and focuses your attention on the statistically significant differences and similarities that reveal the most important and insightful elements of the text.
