Structured and unstructured data analytics for marketing. ‘Alien Vs. Predator’ or ‘When Harry Met Sally’?

What text analytics can do for marketing is impressive. More impressive is what it can do in conjunction with non-verbal data.

Stephan Shakespeare, the founder and CEO of poling colossus YouGov, recently wrote a piece for Conservative Home on the relative popularity of Boris Johnson and David Cameron among voters. His analysis uses a tool called the “Opigram”.

This acts as a searchable, Facebook-like profile-compiler for YouGov’s panellists, which they use to better describe themselves. It also gives YouGov background to straightforward survey results. You could map people’s preference for Boris against their appreciation of Moby Dick, for instance.

The ideal Conservative Prime Ministerial candidate?

The ideal Conservative Prime Ministerial candidate?

As you’ll know if you regularly read our blog- and let’s face it, who doesn’t these days? – this ability to map unexpected associations is what gives text analytics much of its strength in providing rich, actionable and unexpected insights.

In our recent work doing text analytics with a major UK retailer, we have been given a lot of unstructured, free-form data (i.e. what people said). But we’ve also been given a lot of traditional, structured data (i.e. which boxes they ticked). We found out a lot by looking at the free-form data. But then we looked at how it matched up with the structured data, and that’s where it got really interesting. And weird.

At Verbal Identity, we’d expected things like the under-35s and the over-70s being the cohorts who comment most about the cleanliness of a store. But other things, like price and tidiness, were also influencing perceptions of service.

When we’re looking for insights, the greater the number of connections that can be drawn, the more value the data has. If we were to combine the connections text analytics can draw from unstructured data with the connections that a structured data system like Opigram has already made…

Well, we’ll allow you to make that last connection yourself.


If you’d like to hear more about how text analytics (with a linguistics overlay) is providing actionable insights for marketing, customer experience and other commercial interest, please prove you’re not a ‘bot by answering the question in the comments box below, or drop us a line. We’ll read your enquiry very carefully.