What if the future of marketing research passed through… synthetic personas created by artificial intelligence? That’s certainly the bet of Léger and Plus Compagnie, the parent company of the Cossette agency, which have just announced the deployment of Smart Persona, their new AI tool developed by Plus Compagnie and powered by Léger’s data, designed to interact in real time with custom-created synthetic personas. Interview with Sarah Mottet, Vice President of Transformation and AI at Léger and Paul Guité, Senior Vice President of Research and Insights at Plus Compagnie.
Can you remind us of the genesis of this technological rapprochement?
Sarah Mottet: Paul’s team at Plus Compagnie has been working on this tool for over a year. They approached us a few months ago to partner and take it even further. At Léger, we add our methodological rigor to the ingenuity and innovation of their AI platform. We ensure the quality of the results.
We actually tested it recently with our proprietary studies and the quality is more than sufficient to make marketing decisions. We achieved a relevance score between 85% and 90%!
This represents a real revolution in our fields. Previously, if a client wanted to test visual A or B to publish on Facebook, it could take time and a certain cost to recruit the right respondents. Whereas now, we can literally get an answer in real time.
Paul Guité: We are not the only ones considering this synthetic approach but we are among the first. And we can be proud that two Canadian organizations have the audacity to put this tool directly into clients’ hands.
How does Smart Persona work concretely?
Paul Guité: First, there is a notable difference compared to research in an LLM like ChatGPT—traceability. To create a synthetic population, we combine multiple public sources. Then, we will be able to segment it according to the desired targeting. For example, we can keep only cheese lovers or electric vehicle drivers, or even add nuances like an introverted or extroverted personality.
We can also add what I call watercooler conversations. Data like that from the Canadian Census doesn’t date from yesterday. So we need to be able to add a touch of currency—we give it a light Internet connection to account for current topics. And Léger’s science deepens our tool even further.
Sarah Mottet: Indeed, we add the segmentation studies we have already conducted in the past to enhance this synthetic population, which is ultimately just a representation of the Canadian or American population. This also allows us to confirm the tool’s results through real research and thus have confidence in its answers.
What are the use cases? It seems to unlock many possibilities?
Sarah Mottet: It’s true that this opens up a whole new field of possibilities in terms of marketing. However, I want to remind everyone that this does not replace traditional research when it comes to making high-stakes strategic decisions, like a brand repositioning.
But for everything concerning daily or moderately strategic decisions, the tool perfectly meets the need. It is now even possible to test elements that were not even imaginable before, such as a press release or variants of a Facebook post.
Paul Guité: We have clients who have already tested packaging on the tool, since we can also submit images. Others have tested the ergonomics of movement inside a store or even a film script.
Imagine having to do this during a group meeting. It takes forever, whereas here, we can get feedback in 30 seconds.
It is also possible to ask it to be creative with questions like: “You, what would you like?” or “How would you see this?”
Sarah Mottet: Yesterday, I tested it on one of our studies on eating habits. I asked it what it would like to improve in its home delivery experience from its grocery store. It answered that it was frustrated not being able to choose if a food was ripe enough, like a banana or an avocado to make guacamole. I really found that funny and ultimately interesting as an idea. A human could have had it.
I also think of a character who embodies a young man in Toronto. When asked where he would go when looking to eat a quick sandwich, he answers Chick-fil-A. Then, we ask him if he has heard about the past controversies toward the LGBTQ community. He answered that yes, he supports this community… but when it comes time to choose a sandwich, he will prioritize speed over his ethical values. He clearly exposes his contradictions like a human would. Or even better because some humans sometimes wouldn’t dare say certain things openly.
Paul Guité: One of the advantages of the tool is that it allows you to reach groups that traditionally respond little to surveys, like high earners. Or to target very specific groups of a few hundred individuals in Canada.
Aren’t you afraid this will cannibalize your traditional research?
Sarah Mottet: I don’t think this will make traditional marketing research disappear. On one hand because the tool is continuously fed by the latter, whether ours or Statistics Canada or Census Canada. It’s necessary.
Moreover, clients need to move faster so we must evolve our working methods. We still recommend starting with primary research so that we can subsequently create custom synthetic personas, adapted to our client’s market. We always need to start with tangible elements at the base.
Subsequently, we come to meet a need for instantaneity that opens new opportunities sometimes not addressed before. And, as I said, Smart Persona allows us to go from 20 hypotheses to 2 or 3, but we still recommend relying on traditional research to decide, especially if it’s a strategic decision. In summary, we have significantly reduced the time and cost of the ideation process!
training.isarta.com 
