Section 1 · what it does
Discrete Choice Analytics Without the Overhead
Conjoint Studio takes a discrete-choice study from experimental design to final deliverable. Generate a constraint-aware, D-efficient design for choice-based conjoint, MaxDiff, or ratings-based conjoint; field it on any survey platform; then turn the responses into the outputs that normally require specialist software and a dedicated analyst: part-worth utilities, market-share simulators, attribute importance, and item rankings. The output is on par with what Sawtooth-grade tools produce, without the license or the statistician. Already have fielded data? Skip the design step and upload it directly.
The Problem It Solves
Discrete-choice studies are the most rigorous way to learn what customers actually value, because respondents reveal preferences through trade-offs instead of rating everything a 9. But the analysis — hierarchical Bayesian estimation, individual-level utilities, share simulation — has stayed locked behind expensive licenses and the expertise to drive them. A founder validating a price point, or an MBA student running a course project, ends up either overpaying or guessing.
This tool closes that gap. Upload a dataset, pick a model, and get publication-ready output you can download and defend.
who it’s for
Small businesses and individual researchers who collect conjoint or MaxDiff data but don’t have the budget for licensed software or a dedicated statistician.
Where It Shows Up
CPG & Food
A regional snack brand tests pack size, flavor, and shelf price to find the configuration that wins the most share against incumbents.
SaaS & Pricing
An early-stage startup runs CBC on plan tiers and features to set pricing and decide which features justify a higher tier.
Health & Services
A clinic uses MaxDiff to rank which service attributes — wait time, cost, location, telehealth — patients care about most.
What the Tool Does for You
01
Explains the Model — Math and Intuition
Section 2 lays out the statistical foundations of consumer-choice modeling: random utility theory, part-worth utilities, the multinomial and exploded logit, and the hierarchical Bayesian structure underneath all three study types — with both the equations and the plain-language reading of each.
02
Designs the Experiment — Constraints Included
Generate D-efficient CBC designs, balanced MaxDiff sets, and profile designs directly in the app. Client constraints — prohibited combinations, conditional rules, minimal overlap — are enforced by the optimizer, and a live readout shows exactly what each constraint costs in design efficiency. Reachability and estimability diagnostics flag problems by name before you field a single interview. Export the design as a CSV any survey platform can consume.
03
Estimates with Modern Methods
Hierarchical Bayesian estimation via MCMC (PyMC), with a multivariate-normal population prior so individual-level utilities borrow strength from the whole sample. A fast variational / aggregate path gives instant previews; the full chain is the deliverable. Reproducibility — seed, model spec, posterior summaries — is stored with every run.
04
Generates the Deliverables That Matter
Depending on the study: market-share simulation and optimization, attribute drivers, consumer segmentation, item rankings, TURF/reach, and willingness-to-pay — all downloadable as tables and visualizations.
05
Keeps Your Work
Registered users store datasets, runs, reports, and saved scenarios across projects — so a simulation you built last week is still there next week.
how to move around
The axis at the top is your map of the workflow. Each section is a point; the gold point is where you are. Points are reachable once you’ve done enough to use them — later steps stay dimmed until then. You can always jump back to revisit the method or your data.
Next: The Method → — the mathematics and intuition behind the three study types, before you create an account and run your first study.