AUTHOR

GREG SAMPSON
Head of Quantitative Research

The exciting element of the democratisation of this technology is that it demonstrates the power of a well-trained Natural Language Processing AI. Particularly its ability to easily and quickly summarise unstructured data at scale.

This technology and approach is not entirely new to us here at T garage. For some time, we have been building this capability into our InsightIQ platform and deploying it in the ongoing conversations we have with Australians. We are already seeing the benefits not only in summarising and making sense of written responses but also with richer inputs such as video responses. We can approach unstructured, open feedback in a much more efficient way than we ever have before.

This technology is like catnip for me and the team here at T garage. What we have been able to do with it so far has been amazing and I can’t wait to see where it goes next. However, for all the benefits, I am also reminded of that trap that so many of us can fall into. The trap of market research as a rear-view mirror.

If we don’t approach this technology in the right way, we are in danger of simply creating a much better rear-view mirror.

A much better and more efficient way of describing what people are thinking and feeling now. We need to do more than this.

The work we do is (or very much should be) to answer the one underlying question that clients have: what should I do next? All the interviews, screening, weighting, coding, modelling analysis and report writing should be driving toward answering that question. Advances in AI are giving us ever greater scope to efficiently mine, question and summarise large swathes of unstructured data to support this. It’s really up to us though to turn this into an answer to the question – what next?

Try asking the AI, no matter what data it has to work with; should we launch this line extension?, which of these packs best represents the brand?, which opportunity areas offer the most sustainable growth opportunity for my brand?, how do I make my category easier to shop? Interestingly when you ask ChatGPT these questions, it tells you that you that it cannot give you specific direction. It does, however, guide you on how you should go about making your decision and in almost all cases recommends market research. Well done ChatGPT!

So, AI has a role to play – mostly in making the researcher’s job more efficient, but effectiveness – providing the necessary direction – that requires more. From our experience at T garage, here are some things to consider when using AI approaches to your data.

1. Get to know the data yourself

It’s one thing to read the AI summary or coding but it’s quite another to read the responses or watch the videos yourself. Maybe one day machine learning will get to a level of understanding of humans in the cultural context that humans can. For now, though, it really does require a human to fully appreciate the subtlety and meaning of a human. This approach allows the researcher to put the AI- generated summaries in terms that make sense to other people and provides direction on what to do next.

2. Add the context to the findings

An AI summary of unstructured data is a very efficient way to understand the key themes from the specific question asked. This will always sit within a broader context. A skilled researcher looking at and interpreting the summary in the broader context will always add additional meaning and direction.

3. Make it real for the business question

Meaning and direction are dependent on your point of view. The AI summary of that data will be the same for the brand in question and the competitor. The true meaning though is determined by your point of view. This is again where the researcher comes in. Use the summary as a basis for a clear point of view on what is next.

Let’s embrace AI, it is the way of the future. At the same time, let’s make sure it improves the whole car, not just the rear vision mirrors!