Test the price before the market does.

Precheck simulates how your customers might respond to a price, a bundle or an offer, so you run only the real test worth running.

See it tested against real experiments

Precheck / Blind tests /Monthly subscription to an online job board

Price $19

Real36%

Precheck, blind31%

Tested blind against randomised price experiments published in

Full record
  • Journal of Political Economy
  • American Economic Review
  • Quarterly Journal of Economics
  • Marketing Science

Every price change is a bet placed with real customers. Precheck places it first with simulated ones, group by group, and hands you the one real test worth running.

Before the real test, not instead of it.

Teams choose prices by asking customers or by testing on them. Precheck sits between the two.

Ask in a survey

What you learn

What people say they would pay. It runs high: people say more than they spend.

Time to an answer

Days to weeks

What it risks

Little money. A confident answer that flatters the price.

Run it through Precheck

What you learn

Which prices and customer groups are worth a real test, who leaves first, and why.

Time to an answer

Same day

What it risks

No customer ever sees the price. It points to the real test; it does not replace it.

Test it live

What you learn

What people actually did. The ground truth, and the only one.

Time to an answer

Weeks of traffic

What it risks

Real revenue, and real customers who saw a price you may take back.

129 simulated customers. Each one is shown a single platform fee.

₹5₹10₹15₹20100%Metro professionals68%Quick-commerce crossovers65%Tier-2 value orderers47%Households with two apps25%Students and young sharers

A food delivery app: platform fee on a ₹350 dinner order. A real Precheck run, illustrative: there is no real outcome to compare with yet.

Every run ends in one real test

One decision, simulated across your customer groups: which price holds, who leaves first and why, and the one real test to run next.

Run report · A food delivery app

Platform fee on a ₹350 order

129 simulated customers · 4 fees · 2 model families

Share who complete the order

₹20

  • Everyone65%
  • Metro professionals100%
  • Households with two apps47%
  • Students25%

The drop is front-loaded. Most of the loss happens between ₹5 and ₹10, from students and from households that keep two apps and compare.

Who leaves first

Students: 64% → 25%

Why they walk, in their words

  • “There's no coupon tonight so between the platform fee and full price it's cheaper and tastier to just cook at home.”

    Households with two apps

  • “No discount code showing tonight so I'll just eat what's in my PG kitchen instead of paying extra.”

    Students

  • “I only order if there's a coupon, so I'm not paying extra fees without a discount.”

    Tier-2 value orderers

Next real test

A live price test: give real dinner orders one of the four fees at random, about 400 completed checkouts per fee over two weeks.

It would settle

  • Where the real drop-off sits between ₹10 and ₹15
  • Which groups are most price-sensitive for real
  • Whether ₹10 beats ₹5 once revenue is counted

Illustrative: a real Precheck run on a typical decision. There is no real outcome to compare it with yet.

Run your next price through Precheck first.

Waitlist members are invited as capacity opens, in the order they joined.

Who brings which decisions

Decisions they bring

01Launch the three-pack bundle, or keep selling singles?

02Move the subscription from monthly to every eight weeks?

03Lead the launch email with price, speed, or ingredients?

04What actually stops the first purchase?

Leaves with

A two-week bundle test on repeat buyers, at one price.

Inside a run

Between your question and the one test worth running, four things happen.

  1. 01Frame the decision

    The price, bundle or offer you are weighing, and what you would do differently depending on the answer.

  2. 02Build the customer groups

    From your own data when you share it, otherwise from public research. Numbers in your files are computed, never guessed by a model.

  3. 03Each person sees one price

    Every simulated customer meets one offer and takes it or walks away. Nobody is asked what they would pay: people say more than they spend.

  4. 04Leave with one real test

    Which price sells more, which groups leave first and why, and the cheapest real test that would settle it.

Precheck workspace / Runs /Platform fee on a ₹350 dinner order

Share who complete the order

129 people · Claude Sonnet 5 + Gemini 2.5 Flash

507510084%₹571%₹1069%₹1565%₹20

Band: 95% interval. Dotted lines: each model family on its own.

Consistent

All four checks pass. Consistency is not accuracy.

  • Two model families agree7 points apart on average, rank correlation 0.96
  • Demand never rises with the feeNo step moves the wrong way
  • Answers vary65% to 84% across the fees
  • Every group is coveredAt least 20 people in each
The Precheck workspace, in development. The screens show a real run, 21 September 2026, with brand names removed.

It knows which price sells more.

Tested blind against published experiments where real customers were randomly shown different prices. Price order right in 3 of 3 shown here.

published experiments run blind, misses included
6
prices compared, real beside simulated
25
held out: never used to tune the engine
2

  1. The answer is sealed

    Precheck never sees the result, the paper or the brand.

  2. Precheck runs blind

    Simulated customers from the same market meet the same offer at the same prices.

  3. Both are printed

    Real beside simulated, price by price.

SaaS paywall, ten prices

Monthly subscription to an online job board

United States, 2015 · Dubé & Misra, Journal of Political Economy, 2023 (Table 5)
0–50%

Right

Ranked all ten prices in order, within a few points at most of them.

App discount, three levels

Same-day cinema voucher sent to phones inside the mall

A large city in China, 2014 · Dubé, Fang, Fong & Luo, Marketing Science, 2017 (Appendix Table 15)
0–40%

Right

Ranked the discounts correctly. Held out: never used to tune the engine.

Rival's app discount, two levels

The same offer from a rival cinema a bus ride away

A large city in China, 2014 · Dubé, Fang, Fong & Luo, Marketing Science, 2017 (Appendix Table 13)
0–20%

Right

Ranked the discounts correctly. Held out.

Selected runs that look like pricing decisions. The share who buy is less reliable than the order; the full record, misses included, is public.

Full record of blind tests

Already made the decision? Test us on it.

Send a price you changed, a bundle you shipped or an offer you ran, and what happened. We run Precheck on it blind and show you how close it got, before you trust it with your next decision.

Share a past decision
  1. 01

    The decision

    What you launched, priced or bundled, and the options you weighed at the time.

  2. 02

    The audience

    Who it went to: the segment, the channel, the size of the group.

  3. 03

    What happened

    Sales, conversion or churn, whatever you measured, over whatever window you have.

Questions

Straight answers about what Precheck does, and what it does not do yet.

It simulates how different customer groups might respond, so you can see which price sells more, who leaves first and why, and pick the one real-world test worth running. Predictive accuracy has not been established yet; calibrating against real outcomes is how it will be.

Be first to test your next price.

Join the waitlist, or send a decision you already made and get it back scored. Invitations go out as capacity opens.

I want to