What is causal AI?
Causal AI estimates what changes when one factor is deliberately varied. It helps a team choose an action, rather than only describe a pattern in past data.
A choice story · by Subconscious.ai
Your survey said yes.Buyers said no.
People do not always choose what they say they want, so map one possible reason and use a fair test to learn what to change.
Use the map to form a starting idea, then test which change affects the choice.
Name who is choosing and the options they see
The launch
The answer sounded right. It was still a guess.
Imagine a watch company asking buyers what matters most. They name rarity in a survey, then choose accuracy at checkout, leaving the team with a report of what buyers said rather than evidence about what changed their choice.
Before launch · survey
I want a watch that feels rare.Sounds clear
At checkout · choice
I will take the more accurate watch.Story broken
What made the choice change?
The turn
A pattern is not a cause.
Prediction finds a pattern. A fair test finds the cause.
What happened togetherPeople who notice rarity often buy.
This is a useful clue, but it does not prove that rarity changed the choice.
What one change didWhen only rarity changes, do more people buy?
Compare similar buyers making the same choice after changing one detail.
Group A · Current message
Everything buyers see today.
Group B · New message
The same choice, with rarity added.
The answer
The answer should change the plan.
A useful result says what to do and how sure to be. A range keeps one lucky number from looking like a fact.
Choice changed+12
more buyers out of 100
0+8+12+16
Recommended actionUse the rarity message for first-time buyers.
What we still do not know
Whether returning buyers will respond the same way.
The method
How does Subconscious learn why buyers choose?
Subconscious runs fair choice experiments with simulated buyers to estimate how changing one factor affects a decision. Each result reports the estimated change, its likely range, and what the study cannot tell us.
What is a hybrid choice model?
It connects what people encounter to their perceptions and preferences, then to the choice they make. Survey answers measure those hidden factors, but do not cause the choice by themselves.
Why can surveys disagree with choices?
People can describe one preference and act on another when price, context, or tradeoffs become real. Choice experiments measure the decision under a defined set of options.
Is this map a study result?
CausalFlow does not estimate the effect. It drafts a possible mechanism and a testable question. A fair study is still required before the map can guide a decision.
The graph follows the behavioral framework in Figure 2 of Hybrid Choice Models: Progress and Challenges by Ben-Akiva, McFadden, Train, and colleagues.
Read the source paperPublished by Subconscious.ai · Updated August 29, 2026 · CausalFlow is an educational starting-point tool, not a fitted model or study result.