Causal experiments for business decisions

Test the decision.
Then make it.

Subconscious is the causal AI company. Compare a price, product, or message in a causal experiment before you commit capital.

Test a decision See proof
// Users

Used across category leaders.

// The say-do gap

What people say is not
what they do.

StatedRevealed
Correlation to a survey.Replication of published trials.

Where launches fail.

Most platforms validate against surveys. A survey can be wrong.

We validate against published randomized trials: the choices people actually made.

// The method menu

Start with the decision.
The method follows.

Question. Method. What lands on the table.

What should we charge?Discrete choice with priceWillingness to pay, demand curve, revenue-optimal price
Which features and claims earn their keep?MaxDiff, best-worst scalingRanked priorities, by segment
What should the product or pack be?Build-your-own choiceThe winning configuration and its cost to serve
How many will they buy, and how often?Volumetric choiceUnit forecasts with basket and repeat effects
Which message wins, and why?Concept test with complete rankingRanked concepts with the reason behind each
What if a competitor cuts price next quarter?Market simulator on the fitted modelShare and margin under any scenario you type

Every row runs the same engine: a designed experiment, properly estimated.

// Decisions

Three decisions. One test.

01

Price

Compare price and package options before revenue is at risk.

02

Product

Test which feature, offer, or assortment deserves capital.

03

Message

Measure which claim changes customer choice before launch.

// Causal decision API

Test the decision before the market sees it.

Launch, price, bundle, and message bets pressure-tested before real money moves.

Launch the story patients, prescribers, and payers actually move on.PharmaFind the causal message behind peak share before launch money is gone.Stop guessing what deserves shelf, spend, and space.CPGModel the behavior behind trial, trade, pricing, and repeat before the category decides for you.Grow ARPU without triggering churn.Telecom / MediaSee which pricing, bundle, and audience moves create lift instead of backlash.Win the room before the opposition defines it.PolicyKnow which message changes minds, turnout, and trust while there is still time to act.

// Method

From question to action.

// Price decision

  1. 01

    Define

    Name the options and the outcome.

  2. 02

    Connect

    Bring the evidence that defines the market.

  3. 03

    Test

    Randomize the intervention.

  4. 04

    Decide

    Compare effect and uncertainty.

Options
$39 / $49 / $59
Outcome
Customer choice
Effect
Estimated lift
Range
Confidence interval
Decision
Act, revise, or stop

// Customers

Customer evidence.

A space tourism MVP shaped against a segment of fewer than 30,000 people worldwide, saving $500M in aerospace research.

“Subconscious.ai augmented the [client's] space tourism MVP in minutes. No wasted iterations. Good press coverage — not price — is the primary driver in consumer preference for [the client].”

Space tourism engagement, 2026 | A commercial spaceflight leader

River City Federal Credit Union lifted auto-loan ad Result Rate 42%, from 0.52 to 0.74.

“We’ve been looking for ways to utilize data better to accomplish our mission of serving our community and members and are thrilled to be one of a select few organizations in the world that have access to this revolutionary technology. It’s exciting to see how harnessing the power of AI with a data-driven approach can improve campaign metrics.”

Michael Quintanilla, CFO, River City Federal Credit Union | River City Federal Credit Union

New Age Floral avoided $650K in traditional research and set the OshunPouch® price before launch, not after.

“This research is instrumental in shaping the final pieces of our marketing strategy—messaging, positioning, and creative direction.”

Aaron Ziemann, Fractional CMO of New Age Floral | New Age Floral

// Evidence

Proof you can inspect.

  • AI assistants

    How do product, primary capabilities, key functionalities, safety and trust features, personality and tone, and update frequency jointly drive consumer choice of a primary AI chatbot assistant?

    400 respondents · 2,000 choice tasks · United States · 2026

  • Prediction markets

    How do payout speed, regulatory legitimacy, fees, market liquidity, and market range drive frequent bettors' choice among Kalshi, Polymarket, DraftKings, FanDuel, and Robinhood?

    400 respondents · 2,400 choice tasks · United States · 2026

01

87% of the human ceiling

87% is how close our best configuration gets to the human ceiling on one study: 0.832 rank correlation against the published human result, where two independent samples of real humans reach 0.959. Across all 43 studies that pass design filters our mean is 0.73, which is 76% of that ceiling. Measured on the Hainmueller immigration conjoint, best configuration. Two independent human samples agree at 0.959 on the same study, so the ceiling is not 100%. It is a validation result, not a guarantee for a new market.

Read the paper
02

Public leaderboard

Current results and failure modes.

Open the leaderboard
03

Customer evidence

See decisions, outcomes, and source notes.

Read case studies
04

Security is reviewable

Enterprise security documentation is available in the Trust Center.

Open Trust Center

// Next

Make the decision with evidence.

Bring the decision.

Test a decision

The rest of the constellation

Docs

Agents can now perform research.

docs.subconscious.ai ↗

Market View

Simulate your world.

world.subconscious.ai ↗

Newsletter

Replicated findings, in your inbox. No noise.

Subscribe