Synthetic Users
Research-grounded user profiles — enriched by your own data. Run structured interviews across any segment before you talk to a single real person.
Synthetic Users is a structured research instrument built on published segmentation frameworks — Nielsen/Claritas PRIZM, GfK Consumer Life, Experian Mosaic, Bain & Company, McKinsey, Euromonitor, Ipsos, and the B2B Buyer Persona Institute. Each profile is grounded in documented behavioural and attitudinal data drawn from public research, not invented from whole cloth. The profiles are living: upload your own market data or test results and the tool will update them accordingly.
Synthetic users are a thinking tool, not a substitute for primary research. Use them to sharpen questions, stress-test assumptions, and explore before you recruit.
Select users and add questions to begin
The comprehensive guide
Synthetic Users runs structured AI interviews across 21 research-grounded user profiles plus any tuned variants you create. It exists to let you pressure-test ideas, briefs and products against a range of authentic-feeling perspectives — before committing to fieldwork, not instead of it.
What this tool is — and what it is not
What it is. A research instrument that puts 21 carefully constructed synthetic users at your disposal, each grounded in published segmentation research or anonymised authentic interviews. You select users, write open questions, and the tool runs Claude-powered interviews in their voices, then synthesises themes across them.
Good for. Early-stage exploration. Pressure-testing a brief, value proposition or design decision against a range of perspectives. Generating hypotheses before recruiting real participants. Stakeholder alignment around a shared cast of users. Workshop activation when real users are unavailable.
Two kinds of synthetic user
The tool distinguishes between two kinds of synthetic user, so you can see at a glance how much of a profile’s feedback is grounded in real data and how much in psychometric design.
Class A · Research-derived
Built from real-world research. Either grounded in published segmentation research (Nielsen, GfK, Bain, Ipsos, McKinsey, Euromonitor, Circana — 17 profiles), or built from anonymised authentic user-interview data (4 profiles, marked with a green Authentic data badge). The narrative, decision style, pains, values and statistics carry real evidence. Treat the feedback as a defensible perspective.
Class B · Designed (psychometric-first)
Built from psychometric definitions; narrative generated to fit. Any tuned variant you create with the ⚙ Tune a variation feature is a Class B user — you set the personality (Big Five) and decision metrics, you choose the cultural calibration (Hofstede), and the system inherits the behavioural narrative from the base. Class B users are useful for exploring edges and sharpening a profile for a specific scenario — but the feedback is psychometrically authentic, not empirically grounded.
Every card shows the user’s class and an Authenticity Index score so it is unambiguous which class you are working with.
How much of this user is based on real data?
There is no established industry-standard score for synthetic-user authenticity. Adjacent frameworks exist — Cooper’s provisional · prototype · validated persona taxonomy; AAPOR’s Transparency Initiative for survey-research methodology disclosure; ESOMAR and MRS codes of conduct; ISO 20252 quality standards for market and opinion research — but none publishes a single 0–100 score per synthetic persona.
The McKay Authenticity Index is our transparent internal heuristic. It is computed live for every profile from three components:
class_base: Authentic-interview = 88 · Research-grounded = 70 · Designed = 40
profile_confidence: 0–100, how well the segment is documented in published research
boosts: +5 if Enhanced (data uploaded this session) · +10 if Calibrated (against a real transcript)
Clamped to 0–100.
An authentic-interview profile with confidence 91 lands around 89. A research-grounded profile with confidence 82 lands around 75. A tuned variant inherits its base’s confidence but uses the designed base, landing around 59.
Who is in the library
Every profile is a fully-realised person: name, age, role, environment, decision style, values, pains, communication preferences, plus the 10-metric psychometric profile and the six Hofstede cultural dimensions.
B2B
- A · Astrid Lund — R&D Data Analyst, Aarhus
- B · Søren Madsen — Production Team Leader, Nordborg
- C · Thomas Rasmussen — Gas Network Field Technician, Viborg
- K · Marcus Thompson — Digital Transformation Leader, Chicago
- L · Aiko Hasegawa — APAC Innovation Strategy Analyst, Singapore
- M · Fatima Qassim — Head of Organisational Development, Copenhagen
- N · Victor Santos — Junior Product Analyst, Copenhagen
- R · Ethan Caldwell — Medtech Technical BDR (CGT instruments), Boston · Authentic data
- S · Annika Vogel — Medtech Territory Account Manager (DACH) · Authentic data
- T · Naomi Brandt — Junior B2B Technical Sales Rep, Manchester · Authentic data
- U · Daniel Mercer — Senior B2B Technical Account Executive, New York · Authentic data
B2C
- D · Emma Johansen — Price-Conscious FMCG Shopper, Copenhagen
- E · Katrine Berg — Luxury & Performance Buyer, Charlottenlund
- F · Mette Christensen — Home Manager, Hellerup
- G · Oliver Pedersen — Budget Traveller, Copenhagen
- H · Alexandra Møller — High-End Traveller, Frederiksberg
- I · Jeppe Nielsen — Budget Entertainment Consumer, Valby
- J · Sofie Andersen — Premium Entertainment Consumer, Østerbro
- O · Amara Williams — Urban US Professional Consumer, Brooklyn
- P · Liang Wei — APAC Digital Consumer, Singapore
- Q · Grete Lindqvist — Experienced Nordic Senior Consumer, Aarhus
The 10 metrics — and the frameworks behind them
Every profile is scored 0–100 on 10 metrics from industry-standard psychometric frameworks. The Big Five / Five-Factor Model (Costa & McCrae) anchors personality; five decision and consumer-behaviour metrics each come from established, heavily-cited frameworks.
Big Five personality (Costa & McCrae)
Low = practical, conventional. High = curious, exploratory.
Low = spontaneous, flexible. High = organised, disciplined.
Low = reserved, reflective. High = outgoing, energised by people.
Low = sceptical, challenging. High = cooperative, trusting.
Positive pole of Neuroticism. Low = reactive. High = calm, resilient.
Decision & consumer behaviour
Low = risk-averse. High = comfortable acting under uncertainty.
Low = prefers simple cues. High = enjoys effortful thinking.
Low = late majority / laggard. High = innovator / early adopter.
Low = price-insensitive. High = highly price-driven.
Low = prevention (security, loss-avoidance). High = promotion (growth, aspiration).
Scores feed directly into the interview system prompt as natural-language descriptors. Drop Conscientiousness to 20 and the answers loosen and ramble. Raise Risk Tolerance to 85 and you get a faster, more aggressive decision-maker.
Geography as six dimensions
Each profile carries Geert Hofstede’s six national-culture dimensions, calibrated by default to the user’s home country. The data covers 30 countries across Europe (12), APAC (10) and the Americas (8), sourced from The Culture Factor Group / Hofstede Insights.
- Power Distance — flat & egalitarian ↔ hierarchical & deferential.
- Individualism — collectivist, group-first ↔ individualist, self-first.
- Motivation: Achievement vs Consensus — consensus, care, quality of life ↔ achievement, competition, success.
- Uncertainty Avoidance — comfortable with ambiguity ↔ needs rules and predictability.
- Long-Term Orientation — short-term, tradition, quick results ↔ long-term, pragmatic, persevering.
- Indulgence — restrained, duty-bound ↔ indulgent, leisure-positive.
Designing a Class B user from a Class A base
Open any profile and hit the ⚙ Tune a variation button. You will see one panel with three groups of controls: Big Five sliders, Decision-metric sliders, and Geographic calibration (country dropdown + 6 Hofstede sliders).
What changes. Slider values feed into the interview system prompt as natural-language descriptors — the Claude interview reads “you are highly conscientious, cautious, an early adopter, prevention-focused, from a high-Uncertainty-Avoidance culture” — and the answers shift accordingly. The behavioural narrative is inherited from the base.
From profiles to interviews
The Research Studio is a three-step workflow.
A confidence score for the whole report
Every report carries a Triangulation Confidence Index (TCI) — a 0–100 score that aggregates three signals about how well-grounded a multi-user session is. Caveat first: this is a heuristic, not statistical inference. Synthetic interviews have no significance level.
avg_profile_confidence: mean of profile-confidence scores of all interviewed users
sample_breadth: log-scaled function of n (1 → 25 · 3 → 62 · 5 → 76 · 10 → 90 · 13+ → 95)
segment_diversity: distinct base segments ÷ n × 100
Read it as a triangulation indicator: high TCI means broad, well-grounded, diverse sampling. Low TCI means narrow, weakly-grounded, or homogeneous.
Get more interview sessions
Click ↗ Refer & earn 10 in the top bar. Add a colleague’s first name, generate a personal referral link, copy it or send via email. When you confirm you have sent it, your account is credited with 10 more sessions (honor-system, up to 3 referrals per browser). Anyone arriving via your link gets their own 10 free sessions and sees a friendly welcome banner.
Frequently asked
You’ve completed your 10 free research sessions
Book a free conversation with Michael to discuss your research needs and get continued access — including unlimited sessions, custom segments, and report downloads.
Book a free call →Already on a plan? Contact michael@mckayconsulting.dk for access.