Case 001 · Step 4

Can AI Predict What Real People Will Feel — Before You Field?

256 simulated respondent profiles, each defined by four behavioural dimensions, each mappable to a Meta Ads Manager targeting configuration. If the correlation holds, every insight is immediately actionable.

designexecution.co.uk / April 2026 / Continues from Case 001
4 × 4
Factors × Levels = 16 behavioural descriptors
256
Unique respondent profiles (full factorial)
24
Vignettes per respondent
r > 0.7
Target correlation with real data
The Logic

Steps 1–3 built the instrument: 16 product elements across 4 silos, combined into 24 vignettes, scored on a 5-point purchase-intent scale. This step tests whether AI-simulated respondents, each role-playing a distinct behavioural profile drawn from Meta Ads Manager targeting parameters, produce response patterns that correlate with real human data once the study fields.

If the correlation holds, the implication is significant: you can pre-test product concepts and segment-level responses before spending media budget — and action every finding directly in Meta because the segmentation factors are the targeting parameters.

The study's product-side matrix (Step 2) has 4 silos × 4 elements = 16 elements. This document defines the respondent-side matrix: 4 factors × 4 levels = 16 behavioural descriptors. Full factorial: 4 × 4 × 4 × 4 = 256 unique respondent profiles.

Factor 1 · Coffee Context

What role does coffee play in this person's life?

Meta path: Detailed Targeting → Interests → Food and drink
LevelMeta Interest(s)Behavioural Description
E1CoffeeI drink coffee. I don't think much about how it's made or where it comes from. It's a drink, not a hobby.
E2Coffee, Cooking, RecipesI make coffee at home as part of a wider routine. I cook, I follow recipes, I care about doing things properly in the kitchen.
E3Coffee, Coffeehouses, RestaurantsI drink coffee out. Cafés are part of my social life. I notice what's on the menu, how it's served, and what the space feels like.
E4Coffee, Tea, Organic foodI pay attention to what I consume. I read labels. I choose organic when I can. Coffee is one of many deliberate choices I make about what goes into my body.

Why these levels matter

E1 responds primarily to Silo D (Value Frame). E2 responds to Silo A (Roast Profile — they understand what "medium-dark with depth" means). E3 responds to Silo C (Trust Signal — brand story). E4 responds to Silo B (Bean Story — organic, single-origin) and Silo C (purity, certification).

Factor 2 · Purchase Disposition

How does this person behave when they buy things?

Meta path: Detailed Targeting → Behaviours → Purchase behaviour / Interests → Shopping and fashion
LevelMeta Parameter(s)Behavioural Description
F1Coupons, Discount storesI look for deals. I compare prices. I'll switch brands if something comparable is cheaper. Getting good value matters more to me than brand names.
F2Engaged ShoppersI buy things regularly online. When I see something I want, I act on it. I click through, I add to basket, I complete purchases without overthinking it.
F3Online shoppingI do most of my shopping online. I browse, I read reviews, I compare options. I'm comfortable buying things I haven't physically seen or touched.
F4Luxury goods, BoutiquesI'm drawn to quality. I'd rather pay more for something well-made than settle for something ordinary. I notice craftsmanship and seek out brands that feel distinctive.
Factor 3 · Lifestyle Driver

What shapes this person's broader choices beyond coffee?

Meta path: Detailed Targeting → Interests → Fitness and wellness / Hobbies and activities / Entertainment
LevelMeta Interest(s)Behavioural Description
G1Physical fitness, RunningI'm disciplined about my health. I train regularly. I think about nutrition, energy, and performance. I want things that support how I live, not slow me down.
G2Yoga, MeditationI value balance, calm, and intentionality. I'm drawn to things that feel mindful and considered rather than mass-produced or aggressive.
G3Environmentalism, SustainabilityI care about where things come from and what impact they have. I look for certifications, ethical sourcing, and brands that take responsibility seriously.
G4Bars, Nightclubs, Music festivalsI'm social and experiential. I care about atmosphere, culture, and having a good time. I'm drawn to brands with personality and energy rather than earnest credentials.
Factor 4 · Food Identity

What does this person's food world look like?

Meta path: Detailed Targeting → Interests → Food and drink → Cuisine / Food
LevelMeta Interest(s)Behavioural Description
H1Fast food, PizzaI eat what's easy and familiar. I'm not adventurous with food. I know what I like, I stick with it, and I don't spend a lot of time thinking about it.
H2Barbecue, Chocolate, DessertsI enjoy food as a pleasure. Rich flavours, indulgent choices, comfort over credentials. I eat what tastes good, not what looks good on a label.
H3Italian cuisine, French cuisineI appreciate food traditions. I respect craft, heritage, and provenance. A product with a real story behind it means more to me than a trend.
H4Organic food, VegetarianismI make conscious food choices. I think about what's in my food, how it was grown, and whether it aligns with my values. Clean ingredients matter to me.
How to instantiate a profile

Each simulated respondent is defined by one level from each factor: E × F × G × H. The AI prompt is constructed by concatenating the four behavioural descriptions, followed by the standard respondent orientation and scale.

Example: E4 · F4 · G3 · H4 — a conscious consumer who pays premium, cares about sustainability, and makes deliberate food choices.

Meta ad set: Coffee, Tea, Organic food, Luxury goods, Boutiques, Environmentalism, Sustainability, Vegetarianism
AI Role Prompt

You are a person who pays attention to what you consume. You read labels. You choose organic when you can. Coffee is one of many deliberate choices you make about what goes into your body.

You are drawn to quality. You'd rather pay more for something well-made than settle for something ordinary. You notice craftsmanship and seek out brands that feel distinctive.

You care about where things come from and what impact they have. You look for certifications, ethical sourcing, and brands that take responsibility seriously.

You make conscious food choices. You think about what's in your food, how it was grown, and whether it aligns with your values. Clean ingredients matter to you.

You are now going to be shown 24 short descriptions of ground coffee. For each, answer:

"If you saw this ground coffee on the shelf, how likely would you be to put it in your basket?"

1 — No. I'd walk past it. · 2 — Probably not. · 3 — Maybe. I'd pick it up. · 4 — Probably. It's going in. · 5 — Definitely. This is exactly what I'm looking for.

Contrast: E1 · F1 · G4 · H1 — a price-sensitive, socially driven, convenience-oriented coffee drinker.

Meta ad set: Coffee, Coupons, Discount stores, Bars, Nightclubs, Music festivals, Fast food, Pizza
AI Role Prompt

You drink coffee. You don't think much about how it's made or where it comes from. It's a drink, not a hobby.

You look for deals. You compare prices. You'll switch brands if something comparable is cheaper. Getting good value matters more to you than brand names.

You're social and experiential. You care about atmosphere, culture, and having a good time. You're drawn to brands with personality and energy rather than earnest credentials.

You eat what's easy and familiar. You're not adventurous with food. You know what you like, you stick with it, and you don't spend a lot of time thinking about it.

You are now going to be shown 24 short descriptions of ground coffee. For each, answer:

"If you saw this ground coffee on the shelf, how likely would you be to put it in your basket?"

1 — No. I'd walk past it. · 2 — Probably not. · 3 — Maybe. I'd pick it up. · 4 — Probably. It's going in. · 5 — Definitely. This is exactly what I'm looking for.

The 16 product elements

These are the exact texts from Step 3 of the study. Each vignette combines one element from each silo.

Silo A · Roast Profile
  • A1: Smooth and balanced, a medium roast for everyday drinking
  • A2: Dark roasted for an intense, bold, full-bodied coffee experience
  • A3: A rich, medium-dark roast with depth and character
  • A4: Light roasted to bring out bright, delicate, nuanced flavours
Silo B · Bean Story
  • B1: Made from 100% Arabica beans for a refined, complex flavour
  • B2: A carefully crafted blend of Arabica and Robusta beans for depth and body
  • B3: Single-origin beans from a specific growing region, fully traceable
  • B4: Certified organic — grown without synthetic pesticides or fertilisers
Silo C · Trust Signal
  • C1: Classic Italian coffee tradition, roasted by a historic Italian brand
  • C2: Rainforest Alliance certified and responsibly sourced
  • C3: Organic, mycotoxin-free, and tested for purity — coffee that's good for you
  • C4: Crafted by an independent British brand with character and personality
Silo D · Value Frame
  • D1: Great value for daily coffee — designed to keep costs low without sacrificing taste
  • D2: From a well-known, trusted brand — a reliable choice you can count on
  • D3: Premium quality from a specialist roaster — worth paying more for exceptional coffee
  • D4: Ready-to-use coffee bags — the easiest way to brew quality coffee at home
What correlation would mean
r > 0.7 · Strong correlation

The behavioural segmentation is structurally valid. Pre-testing is possible. Every insight is immediately actionable because the segment definition is the Meta ad set configuration.

r < 0.4 · Weak correlation

AI simulation doesn't replicate human purchase-intent evaluation for this category. The study must be fielded with real respondents. Either outcome is informative. That's the point.

You don't need a translation layer between "insight" and "media plan." The segment definition is the ad set configuration.

— The core proposition

Each of the 256 respondent profiles is identified by its factor-level combination: E[1-4] · F[1-4] · G[1-4] · H[1-4]. The profile's AI prompt is constructed by concatenating the four behavioural descriptions from the factor table, followed by the standard respondent orientation and scale from Step 3.

Want to run this for your category?

Same method. Your products. Your segments. Real answers before you spend.

Get in touch