Make human choice visible. See why people choose.
We build a digital twin of your market, your competitors and your customers. Then we run large-scale simulations on that market, so you can make important business decisions with confidence.
For price, launch, message and market-entry decisions.
“Right now it is straight vibes and it's 2026. We don't have to do this this way anymore.”
Your price: $10 to $12. 60 buyers switch. Revenue falls 19%.
Illustrative market: 600 simulated buyers, your offer and two competitors.
2.73×Annual price, $348 to $950“We validated customer feedback on pricing through Subconscious.ai’s survey insights, which gave us the confidence to adjust our pricing strategy successfully.”
0.79–0.88Jensen-Shannon similarity across Mintel’s attribute distributions“The best comparison we've done across all our work together.”
$60Modeled price from five iterative experiments“This research is instrumental in shaping the final pieces of our marketing strategy—messaging, positioning, and creative direction.”
02Cause
A survey asks. An experiment intervenes. We change one thing in the market and measure what changes in choice. Only one thing changed, so you can explain the result.
You are about to
What they say
“Price is not what I choose on.”
What they choose
0 of 600 buyers change their choice. Move the lever.
Illustrative market of 600 simulated buyers. The interval comes from 240 resampled markets.
03Scale
One experiment answers one question. The twin runs the same test on every lever you control, at every step of the customer journey. The two or three levers that matter rise to the top.
2,400 experiment arms run 8 levers × 5 steps × 6 levels × 10 groups
Test these first Pick a lever to find its experiments.
Illustrative design: 8 levers, 5 journey steps, 6 levels and 10 buyer groups make 2,400 experiment arms. Bars show the change in choice share, with a 95% interval.
04Proof
Customers bring a decision that matters. They leave with a result they can explain and act on. Behind each result is a method checked against published human studies, one study at a time.
“We've given you almost no direction and you've come back with things that work.”
“You do not have the resources and bandwidth to go test them all. Let's simulate them all.”
Checked against real people
Each mark is one published human study. Left to right: how closely the simulated effects rank like the human ones. Red: the direction of effects failed. Point at a mark to read its study.
Public study registry, read 2026-11-12. Download all 308 comparisons
July 2026 working paper, not peer reviewed: 0.73 mean Spearman rank correlation on estimated choice parameters across 43 design-filtered studies; 0.55 across roughly 300 replications. Open the evidence
Use the simulation to find the two or three levers that matter. Then confirm them with real people or a live test before you commit. When a question cannot be simulated reliably, we say so at the start.
Bring a study you already trust. We run the same design and show you both results, like for like.
PhD scientists on staff do the research. Scientific advisors: Dr. Pascal Wallisch, NYU clinical professor; Dr. Howard Moskowitz, Harvard-trained psychophysicist; Connor Joyce, Microsoft. Board advisors: Johny Mair, Ethic; Terri Hall, formerly Gartner; Lynn Girotto, Qualtrics. Patent claims allowed.
Meet the team