Quantitative Methods
Quantitative research is run by respondents, light (mini) AI twins, and answers how many: percentages, distributions, statistically validated signals.
Surveys — Custom
Section titled “Surveys — Custom”What it is. Build your own survey from scratch. You write the objective, design the questions, and choose your respondents. The platform is agnostic to what you build: custom surveys are open-ended.
When to use. When your research question doesn’t fit a standard template, or when you have specific hypotheses to test that the standard surveys don’t cover.
Surveys — Standard
Section titled “Surveys — Standard”Standard surveys use locked question sets to ensure consistency and prevent misuse. There are three types:
Brand Track
Section titled “Brand Track”What it is. A share-of-voice and brand-funnel measurement survey. Questions are standardized, you cannot change them, to ensure that results are comparable across runs and that no user inadvertently skews the measurement by rewording a question.
What it measures:
- Share of voice: what percentage of category conversation is about your brand versus competitors
- Brand funnel: awareness → familiarity → consideration → preference → intent → endorsement
- 90-day purchase intent: how likely respondents are to purchase from each brand in the next 90 days
- Brand perception metrics: how the brand rates on dimensions like trust, innovation, value for money, premium feel, and customer support
The endorsement stage maps to the Net Promoter Score (NPS) framework: how likely a respondent is to recommend the brand. Scores of 9–10 indicate loyalists; 7–8 are neutral switchers; 1–6 are detractors.
When to use. When you need a baseline read on brand health, or when you want to compare your brand against specific competitors on a standardized scale. The locked questions are a feature, not a constraint: they exist because user-modified questions would make results incomparable and easy to game.
Segmentation
Section titled “Segmentation”What it is. Classify your respondent base into sub-segments based on behavior and attitude inside a defined context.
Real example. Walmart in-store shoppers were classified into three in-store segments: urban time-savers (come with a list, want to move fast), value families (high coupon usage, in-store-shopping-first), and conscious seniors (eco-conscious, deliberate). Each segment has a distinct profile, market size, behavior patterns, sentiment, surfaced from the segmentation survey.
The drill-down. Once segmentation produces a sub-segment, there is a “create another twin from here” button. That sub-segment becomes the seed for a new AI twin, which you can then use to run further research, including another segmentation, drilling deeper into the audience as many levels as you need.
When to use. When you have a defined base (a city’s Walmart shoppers, a brand’s existing customers) and want to understand the distinct mindsets within it rather than treating the base as homogeneous.
Media Consumption
Section titled “Media Consumption”What it is. A standard survey that maps what media channels the audience consumes, television, apps, websites, and builds a story around their media behavior.
Quick surveys / polls
Section titled “Quick surveys / polls”What it is. Simple, lightweight surveys. A poll gives respondents two or three options and asks them to choose. Quick surveys let you add your own answer-set questions without building a full survey structure.
When to use. When you have a fast, specific question that doesn’t need the depth of a full survey. Influencers run polls on social media to decide which restaurant to go to, and the mechanism is the same: pick the best option, tally the count.
Concept testing
Section titled “Concept testing”Concept testing covers three distinct use cases under one umbrella:
Creative testing
Section titled “Creative testing”What it is. Upload an image or ad copy and run it against respondents to measure reaction. Not reactions from one or two twins: statistically significant response from a respondent set.
The “$5K instead of live A/B” value framing. Traditional A/B testing means splitting real ad budget, say, $5,000 to ad variant A and $5,000 to ad variant B, running both live, and waiting for conversion data. The pitch here: give us $5,000 instead, and we will tell you which ad wins before you spend a dollar in market. And you can test more than two variants.
Message prioritization
Section titled “Message prioritization”What it is. Present several candidate messages (taglines, value propositions, headlines) to respondents and measure which one resonates most.
Real example. Three fried-chicken taglines:
- “Finger licking good”
- “Juicy through and through”
- “Crunchy on the outside, juicy on the inside”
Run a message prioritization survey. Get a statistically grounded answer about which message lands best with your target segment, before any production spend.
Concept / product testing
Section titled “Concept / product testing”What it is. Share a concept, a product idea, a new service, a positioning, with respondents and measure how it lands. Not qualitative exploration (that’s co-creation); this is a statistically measured read on how many people respond positively, what their hesitations are, and how the concept ranks on key dimensions.
When to use. When the concept is developed enough to evaluate, and you need numbers rather than exploration. Use co-creation to shape the concept; use concept testing to validate it.