Conjoint.app, an educational tool by Subconscious.ai

Causal experiments for business decisions

Start with one action. Measure one behavior. Decide what the evidence supports.

Start with the decision

Suppose a coffee company is deciding whether to add “Roasted this week” to a bag. A prediction tells the team what to test. It does not settle the decision.

Random assignment creates a fair comparison

Assign 400 illustrative shoppers equally to two groups. Keep the beans, price, and placement the same. Add the new sentence to only one bag. The difference between groups estimates the causal effect.

Turn the result into a decision receipt

In this made-up example, choice rises from 32% to 43%. The estimated effect is +11 points and the confidence interval is +1.6 to +20.4 points. A decision threshold shows whether the evidence is strong enough to act.

Subconscious applies the method to buyer choices

Subconscious randomizes products, prices, packages, and messages in simulations of target markets. Every read reports the action, behavior, estimated effect, possible range, guardrail, evidence status, and next action.

The public validation record includes 43 design-filtered studies across nine domains, with failures included.

Inspect the public evidence record