HOW WE WORK

We are a fully transparent company and do not believe in black boxes. Everything we do, we will detail, explain, and review, if needed. We are analytically rigorous and brand savvy. We use proven methods that have been validated in academic research.

All of our work is led and conducted by senior leaders (we do not hand off work to junior researchers). We love innovation and constantly look for ways to do things more efficiently, and all-around better.

We use AI in two different ways

ANALYZE OPEN-ENDED TEXT
AND RESPONSES

We use AI to analyze open-ended text responses (such as we use in our brand density analysis). In analyzing open-ended responses, we start with a manual human coding step. Once we develop a human-coding-based dictionary of relevant meaningful terms, we use models such as the BERT model to score the entire dataset.

GENERATE SYNTHETIC
RESPONSES AND RESPONDENTS

We use AI to generate synthetic responses & respondents. This can be done in two ways:

We can generate synthetic respondents for a subset of variables using techniques such as SMOTE. See our paper on how we use these to improve segmentation typing tools.
We can generate synthetic responses to open-ended questions such as our brand density question. Our BD AI solution is based on this. Here we use a variety of AI tools such as ChatGPT, Llama3, and BERT.