Chapter 4 of 10
Your first study
Build a small study, preview it, and launch. Start with Survey or 5-Second Test so the loop stays short.
A study is what participants take: a welcome screen, your tasks and questions, and a thank‑you at the end. Drafting is not launching. You choose a method, configure tasks and screens, decide how you will recruit, then review before anything spends credits or reaches real people.
Pick a small first method
Survey when you have questions and maybe a design to show beside them. 5‑Second Test when you want to know what people notice and remember after a short look at a screen. Either one teaches you the builder without a long prototype or tree.
You can combine methods later in one study. On day one, one method and one clear task is enough.
One clear task
Write the task the way you would say it to a busy person. “Where would you tap to pay?” beats “Please evaluate the efficacy of the primary call to action.” If the study shows a screen, make sure the image or prototype link is the one you mean to test. For Prototype Test later, you can paste a public Figma prototype link without connecting an account, or connect Figma for a fuller import. You do not need that for Survey or 5‑Second.
Walk the builder
The builder has four stages: Method, Configure, Recruit, and Review. Method is where you choose Survey, 5‑Second, or another type. Configure is questions, timing, images, and task copy. Recruit is where you will later choose AI Panel, Your Panel, GenPop, or a mix of paths that still stay labeled separately. Review is your last look before launch.
Optional AI assists can help draft questions (1 credit). Writing by hand is free.
Preview before you launch
Preview as if you were the participant. Fix wording that only makes sense inside your team. If the study will go to real people, remember: you cannot edit a study after it has been sent to real participants. When you need a variation, use Clone study to make a fresh draft with its own copy of every image. That keeps in‑progress results intact.
Launch order that saves pain
- Preview.
- Run an AI Panel first and treat that pass as a hypothesis (chapter 5).
- Only then open Your Panel or GenPop when you need evidence (chapter 6).
New studies created from an AI tool over MCP also land as drafts for you to review. Anything that spends credits or wallet funds should ask you to confirm first.
Keep it small
Every method has build limits, and the builder shows them as you go. A survey of a few dozen questions at most, a handful of tasks, a few variants: that is the spirit. If the builder stops you, follow the limit it shows.