MaveraMavera Docs
Operator's Guide
Checking…
Use or share this pageSend the page itself, print it, or open its complete context in your AI workspace.

Running studies

Structured surveys with numbers you can defend.

2 min readDocs build Updated Jul 26, 2026
Interactive product walkthrough

Build and run a study

Set the question, audience, method, sample, then read the result.

Simulation onlyNothing is saved or charged
1
Step 1 of 8Open Studies

Choose Studies from the sidebar.

Demo workspace / Mavera workspace
Local simulation
Mavera workspace

Good afternoon

Choose a surface from the sidebar to begin.

Demo workspace0 actions completed
WorkspaceDemo companySafe local simulation
Current surfacestudyFollow the highlighted action
Recent workspace activityStatus
Audience comparisonReady to review
Customer interview evidenceGrounded
Campaign pre-flightDraft
  1. 1
    Open StudiesChoose Studies from the sidebar.
  2. 2
    Start a studyCreate a new structured study.
  3. 3
    Write the decision questionState the decision the study should resolve.
  4. 4
    Choose the audienceSelect the population that should answer.
  5. 5
    Choose the methodSelect a message test, pricing read, MaxDiff, or survey.
  6. 6
    Set the sampleMove the sample-size control and see the estimate update.
  7. 7
    Run the studyApprove the local plan and start the simulation.
  8. 8
    Read the evidenceOpen the result, uncertainty, and source-lineage panels.

When a study is the right tool

Use a study when someone will ask "what percent?" — message tests, pricing reads, concept ranking, feature priorities. If the question is "why do they feel this way," start with a focus group or interview instead.

Setting one up

From Studies (or by asking Mave):

  1. Pick an audience from the dropdown — or leave it blank and describe one; Mavera builds it on the fly.
  2. Write your question(s). Types available: single choice, multi select, scale, NPS, ranking, numeric, open text.
  3. Panel size: use the platform minimum for a more stable comparison within the modeled panel. A larger synthetic sample does not create population representativeness.
  4. Run it. Typical time: 1–3 minutes.

Reading results

  • Shares are computed from the actual synthetic responses — counted in code.
  • honesty.sharesAre on every result says what the numbers are: measured-synthetic (counted from responses), partially-grounded (some real data behind it), or estimate.
  • Open-text themes come from what respondents actually wrote, with quotes.

A/B message tests

Give Mave (or the study form) two variants. The panel splits automatically, each half sees one variant, and you get share-per-variant with the gap called out. Ask Mave "is this gap meaningful at this sample size?" and she'll tell you straight.

What studies can't do

A synthetic panel can't tell you a real market share or predict a launch number. It is designed for relative comparisons such as which message lands harder, which price triggers more resistance, or which objection appears first. Treat exact effect sizes as directional unless independently validated.

Talking to your audience

Chat, focus groups, interviews, and debates — when and how.

Workflows & Programs

Guided recipes, the natural-language builder, and long-running research.

On this page

When a study is the right toolSetting one upReading resultsA/B message tests