Imagine being able to talk to your customers, as if they were in the room with you.
Advances in large language models (LLMs) made this a realistic prospect, but out-of-the-box solutions were falling far short of being able to interpret market research deliverables.
At STRAT7, we spotted this opportunity early. We created strat7GPT last year – a generative AI layer that integrates our own custom document pre-processing, search retrieval and chat functionality – to make information dense materials (like PowerPoint, Word and PDF reports) easily accessible to an entire organisation, through natural language processing.
Given our heritage in segmentations – delivering over 2,000 of them since 2005 – we’ve taken the natural step to create a segmentation chatbot product, which lets users access segmentation data entirely through natural language.
This product directly uses the high quality research outputs associated with segmentation, such as a debrief, pen portraits, data tables, and qualitative interviews. It can be viewed like having a segmentation expert as part of your team, to disseminate segment knowledge across the business, or receive segment level feedback on propositions or campaigns.
The tool lets users pose a variety of questions:
Basic summaries – ‘Can you summarise the segments in one sentence, using simple language?’
Fact finding – ‘What are the biggest concerns that [Segment X] have about eating out?’
Opinions – ‘Based on all that you know about [Segment X] how would they respond to the new product concept below?’
Now, the obvious question: How is this different to out of the box Generative AI applications out there?
The main way lies in how we process the data that is fed into the tool. We have trained our AI models how to process traditional market research outputs (which are often messy) to provide both accurate and creative responses to lots of questions about segments. We have consciously engineered the chatbot for accuracy and generating user confidence – employing 1000s of tests for hallucinations to factual questions, adding functionality to share sources for its response, and making the default behaviour to reveal if it’s not sure of the answer – removing the black box, and putting users in control.
To date, our clients have benefited from the segmentation chatbot in three ways:
1. Democratising the segments
The chatbot opens up a new way for people to engage with segments, especially those outside the core insights team that are not as familiar with market research deliverables or the information available from the segmentation.
2. Saving time and freeing up resources
The chatbot reduces the volume of requests the core insight team fields from stakeholders on a daily basis – freeing up time to work on ‘higher value’ tasks. It also makes easy work of manual tasks, all in one place. Hours of qualitative interviews, hundreds of data tables, and lots of PowerPoint deliverables, are all accessible in seconds and engaged with via simple chat.
3. Enabling timely decision-making
The chatbot answers questions posed of the data and outputs from the segmentation, giving strong opinions to inform decisions and trade-offs about marketing, brand and innovation. Providing quick answers and opinions alongside the supporting evidence.
If you’re looking for a solution to chat to your segments ‘in the room’ please get in touch, or contact Hasdeep Sethi directly at hasdeep.sethi@bonamyfinch.com.
