Bots are changing the market research landscape, in the way we think about what questions can be asked, of whom, and how often.
Tracking studies are often the cornerstone of large insights departments – critical to measuring how a brand is performing in relation to the competition against a pre-determined set of metrics.
At the other end of the scale, when it comes to understanding attitudes and behaviours which don’t neatly fit into a pre-determined set of closed questions, an ad hoc qual approach is often in order. This type of traditional qual is great at delving deep into an issue, but it does this in a way which delivers information that doesn’t have much of a shelf life. Some organisations respond to this by running the same set of groups with the same discussion guide every year or so to check if things have changed, but I suspect that this is quite rare. And of course, it’s highly subjective and difficult to measure subtle changes with small sample sizes. In reality, most of the time once a qual project is finished, decisions are informed and the report itself, if it can be found, has little ongoing relevance to the business.
Chat-bot based research is slowly carving out a niche for altogether new types of research which can sit somewhere in between these two extremes – the always-on qualitative questions.
In many categories, issues can emerge which have the potential to change how people view the category, and how they respond to brands. A move to drinking less. A trend towards better-for-you snacking. A rejection of meat, concern around factory farming. If I were the owner of a large established brand who knew that they would not be able to make nimble adjustments to their products or brand assets, I would want to know about consumer issue as they emerged. I would want to understand them so I could consider how to respond from a product, brand, advertising or distribution perspective. An annual dipstick in and out of such issues may not be the best approach. So what is?
For the cost of conducting maybe four focus groups per year, you could set up and run a short, engaging chat bot using qual and quant to monitor issues with around 200 of your target audience per month. For example:
To what extent do you agree with the statement “I am consciously trying to drink less alcohol nowadays.”
(If agree). Can you tell me why you are trying to drink less nowadays?
Can you give me some examples of what you are doing to drink less alcohol?
What are you drinking more of?
What are you drinking less of?
Within each of these responses, we can use natural language processing to measure sentiment in these comments, and identify, brands, products, places and people. We can even measure the sentiment of each of these things. So we can measure, for example, how mentions of low alcohol wines are increasing or decreasing, we can also measure the sentiment around low alcohol wine – even when it is included in a comment which meanders through a range of things.
Just say someone says:
I do feel like I should be drinking less to be healthy but I still feel like I want to have a drink. I recently tried Matura Lighter Sauvignon Blanc, it was really nice. I think I will buy it again. I have noticed people drinking Carlton Zero but I can’t see the point of non-alcoholic beer, I’d rather have a soft drink.
From this we can count mentions of two brands, we can extract positive sentiment around the wine but negative around the beer. We can also measure reference to three categories – low alcohol wine, zero alcohol beer and soft drink. And we can also measure an overall sentiment score around the topic before we dive into the next question.
Over time, we can even develop a simple machine learning model to ask follow up questions.
Importantly you can see how each of these instances increase or decrease over time.
Overlayed with demographics, location, and perhaps linked into shopper history from a loyalty program, this becomes a valuable resource across the business. The power lies not just in the research itself, but the fact that the dashboard can be self-served by anyone in the business to see how trends are changing over time, which audiences are responding in which way, and deep dive into individual verbatims for greater insight.
Such a research programme could prove invaluable in generating thinking around new products, messaging, claims, reasons to believe, new product ideas, new brand development and communications territories.
Let us know if you would like to have an obligation free discussion about what we could design for your business.