Skip to content

JonbarIstanbul (UTC+3)

ENTR

[ Product ] Simulation engine v0

One decision. Several versions. The same market.

Jonbar forks your next price or campaign into 3-5 versions, runs each on your own sales history, and shows units, revenue and gross profit per version, with a range and the reasons behind the difference.

Join the waitlist

Run · Wireless headphones, Q4

Sample scenario · synthetic data
Winner · B+5.4% vs. A
Gross profit · 80% range
WorldValue80% range
A2,087,395baseline
B (Winner)2,200,9182,133,428 – 2,256,772
C2,071,0712,019,176 – 2,125,434
Weekly · 12 weeksGross profit · Weekly · 12 weeks
Engine SIM-CORE v0Seed 42Not yet backtested

[ 01 ] How it works

One loop: see, rehearse, choose, learn.

  1. See
    • 01 Connect
  2. Rehearse
    • 02 Fork
    • 03 Simulate
    • 04 Compare
  3. Choose
    • 05 Explain
  4. Learn
    • 06 Ship and measure
  1. [ 01 ]In pilot

    See

    Your sales, price and campaign history in one place.

  2. [ 02 ]In pilot

    Rehearse

    3-5 versions of one decision, run side by side in the same market.

  3. [ 03 ]In pilot

    Choose

    The range and the reason behind each result, before you pick.

  4. [ 04 ]Building

    Learn

    The forecast is locked when you decide, then checked against what really happened.

[ 02 ] Simulation engine

Classic demand models, run side by side.

No black box, and no AI playing shoppers. v0 uses demand models you can check, fitted to your own history.

  • Price elasticity

    From your own price changes, blended with a published prior when your history is thin.

  • Promo lift and pack framing

    Percent off vs. "2 for x" presentation of the same discount.

  • Pull-forward and post-promo dip

    Sales that were only moved, not created, and the weeks that pay for them.

  • Stockpiling

    Available as a parameter; off by default until your data supports it.

  • Cannibalization

    Units your other products in the same group lose.

  • Competitor reaction

    A simple rule: a share of a price cut matched after a lag.

  • Per-channel baselines

    Decide on one channel; the other channel's history stays clean.

Not modelled yet: free-shipping thresholds, shopper segments and baskets, cross-product bundles.

[ 03 ] Scenario panel

Every version, every metric, with its range.

Switch between gross profit, units and revenue. The winner can change with the metric; the panel says so.

Weekly gross profit by world · sample scenario
Weekly gross profit by world · sample scenario
A
Planned: 20% off
B
20% off as a pack, weeks 5-6
C
10% off, weeks 5-6
Units sold in each world, as shares · sample scenario

ABC

[ 04 ] Backtest

We publish the backtest before the claim.

Before we tell you how accurate Jonbar is, we test it blind on past pricing and campaign decisions whose outcomes are public, and publish the result next to what a simple rule scored on the same cases. That test is being built. Until it's published, every run says "not yet backtested".

Read the methodology

[ 05 ] Security & data

Your sales data, and nothing else.

  • We never ask for customer data. Product × day sales is enough.
  • Deleted on request, kept no longer than 90 days.
  • Each customer's data is isolated at the database level.
  • No AI model sees your data. Runs use statistical demand models, no LLM calls.

Pilot runs are operated by the Jonbar team; there is no self-serve upload yet.

[ 06 ] Decision-time check

Before you join the campaign, see what's left.

The platform suggests 15% off. After the discount, commission, shipping and product cost, how much stays with you per order? And how many more orders would you need to come out even?

Building

Decision-time checkSample scenario · assumed commission and shipping

Platform campaign: 15% off · Wireless headphones

  1. List priceTRY 1,000
  2. Discount−TRY 150
  3. Commission (assumed 18%)−TRY 153
  4. Shipping (assumed)−TRY 60
  5. Product cost−TRY 585
  6. Left per orderContribution marginTRY 52

Without campaign: TRY 175

Extra orders needed+237%3.4× the orders

We're building this check now. It will use the same economics layer as the rehearsal, so the check and the worlds will agree.

[ 07 ] Goal to decision

Write the goal. Get the three best plans.

Instead of drafting every version yourself, you state what you want and the limit you won't cross. Jonbar will search the levers it models and bring back the three plans that fit best, each with its range and the reason it ranks where it does.

Sample goalBuilding

Your goalRaise contribution margin by 8%, without losing more than 3% of volume.

585 candidate plans · 85 meet the goalSample scenario · seeded model, not engine output
  1. Plan 1Price +10% · 10% off for 1 wk
    Contribution
    +21.7%
    Volume
    -2%
  2. Plan 2Price +8% · 5% off for 1 wk
    Contribution
    +19.2%
    Volume
    -2.4%
  3. Plan 3Price +6% · 5% off for 1 wk
    Contribution
    +14.7%
    Volume
    -0.8%

[ 08 ] Roadmap

Now, next, later.

  1. Now0-3 months

    • Price and campaign decisions
    • Platform economics and the decision-moment check
    • Learn: the forecast locked at decision time, measured against the result
    • Shelf, level 1: one snapshot of the competing list at decision time
    • Goal to plan, narrow version
    • Blind backtest on past decisions

  2. Next3-9 months

    • Launches and bundles
    • Learning from similar products
    • First enterprise pilot

  3. Later9-24 months

    • Ranking algorithm, if ranking history is available
    • AI shopping agent world
    • Marketplace partnerships
    • New-market decisions, with data from Turkey, Germany and Saudi Arabia
    • Decision API and platform licence

See it on your own campaign.

Join the waitlist