Survey tools are great for asking people what they think.
But many marketing questions are about what people do:
which product they put in their cart,
what they'll actually pay,
how fast they associate a brand with "premium".
Answering those questions takes interactive, adaptive tasks,
and usually a custom build for every study.
oTree is a behavioral research platform with over
3,300 citations.
With oTree AI (OTAI),
you describe your study in plain English,
and an AI agent builds and tests it for you.
There is no coding.
What you can build
- Choice-based conjoint: load your own experimental design, with multiple versions assigned across respondents.
- MaxDiff: best/worst scaling of features, claims, or messages.
- Adaptive pricing: Gabor-Granger price ladders that move up or down based on each answer.
- Simulated shopping: a realistic online store with your product catalog, a cart, and a budget.
- Concept and content testing: monadic or sequential designs with rotated order.
- Implicit measures: reaction-time tasks such as the IAT, for brand associations that people won't or can't report.
- Incentive-compatible willingness to pay: auctions, price lists, and lotteries where choices have real consequences.
- AI shopping assistants: participants chat with an LLM-powered sales agent or chatbot
(see oTree for human–AI interaction experiments).
Combine them in one study
This is where oTree differs most from survey tools and single-method tools.
You can chain any of these tasks together,
randomize respondents into cells,
and branch on anything that happened earlier.
For example: show an ad, let the respondent shop,
then ask a pricing question about the exact product they picked.
If you run a different study for every client,
you no longer have to choose between a rigid template and weeks of custom development.
See it for yourself
Our live demo site
has examples you can try right now
(or import into your own OTAI project and customize):
choice-based conjoint, MaxDiff, Gabor-Granger pricing,
an online grocery store, monadic and sequential concept testing,
and the IAT.
Everything built in
- Hosting, with no software to install.
- Recruitment through Prolific, or any panel that sends respondents to a link.
- Precise reaction times, measured from when the stimulus actually appears on screen.
- Live monitoring of every respondent's progress and answers.
- Automated testing: bots run through every branch and cell of your study before launch.
- Data export of every choice and response, ready for analysis in R, Python, or Excel.
- Readable code that you own and can audit.