AI-driven insights for innovative, never-before-seen products

"AI is just based on existing data. How can it generate useful insights for a new product? Surely this can’t work for my product; there isn’t anything like it in the market"

When we speak to clients and customers, we hear these kinds of concerns a lot. Frankly, this is a real problem for any statistical or AI approach – predicting the future is notoriously hard!

However, instead of learning from and making predictions for products as they are, our AI models work by finding and learning from patterns in data. This means breaking down products into multiple dimensions and retrieving relevant historic information for each dimension or combinations of them to make meaningful predictive analyses. Below is an example of how the Seer approach works in detail.

At Seer, we take a unique approach to predicting consumer attitudes to new and existing products. Instead of focusing on a product as a whole, we distinguish specific features, ingredients, and other attributes used in the development and marketing processes. (This is actually how Amazon’s search function works!) These features are compared with similar features of existing products, allowing us to pull in consumer perspectives that reflect on these. By combining insights across all relevant features of the new product and the perspectives that consumers have on them, we can create synthetic personas that can provide “feedback” on your innovations.

For example, let’s say it’s 2010 and you want to develop a new non-alcoholic beer. There really aren’t many players or equivalents out there. Here’s how you can use Seer Voice to get instantaneous insights:

1. Provide descriptions of the:
   1. Product: taste, alcohol level (i.e., <0.5% ABV), nutritional information, price, etc.
   2. Persona: who you aim to target (e.g., pregnant women or athletes looking for refreshing alternatives to beer).
2. Seer Voice extracts the key attributes of the product and persona to pull in the most relevant data, including specific use cases (e.g., your beer alternative)
3. We leverage proprietary Bain frameworks to create synthetic personas that learn and pull from this data.
4. User queries are passed to each synthetic persona, and these are summarized to provide an easy-to-view interface.
5. Relevant quotes representing the perspective of the synthetic persona can also be highlighted making it easier to understand why specific product attributes are desired by consumers.

Of course, any approach leveraging past data has its limits and we recognize this. With Seer, you can generate instantaneous insights to validate your hypothesis, reducing reliance on time-intensive, costly surveys. This can enable your teams to iterate faster and, if needed, run a small, quick survey (e.g., <100 people) to validate.