For consumer-brand pricing, promo & category teams

Run any price or promo past your market — and prove it before you bet real money.

Thinking of changing a price, planning a promo, or launching a product? Check it first against a model of your own market — thousands of shoppers that behave like your real buyers. In minutes you see roughly what you'll sell and make, which ideas quietly lose money, and which few are actually worth testing for real. Less guessing, less waiting weeks on research — and we're honest about how sure we are.

That little market behind is the model working — each figure is one shopper; change a price and they switch brands. In the demo we run it on your category.

right 4 of 5×calls which way sales move when a real price changes — so it stops the money-losing moves
on your databuilt from what people actually bought, not what a survey said they'd do
honestevery answer comes with how sure we are — and where it stops holding
Frozen Pizza · live what-ifHELD-OUT ✓
DiGiorno Pepp.20.1%
Store-Brand16.4%
DiGiorno Supr.14.8%
Freschetta6.3%
Promo −25%+15% lift · 61% truly incremental · worth it
Raw simoff ~26×
Calibrated≈ real
Real salestruth
Accuracy vs real sales it never saw. Held-out R²=0.64 · calibration recovered share error −0.35 → +0.44.
01How it works
A working model of your market you can ask anything.

Grounded in your real sales, not a survey — and tuned until its numbers match reality.

01
Give us your sales

Your category's products, prices and units sold. Even a few weeks is enough to start — the rest can be public data.

02
We build your market

Thousands of simulated shoppers that weigh brand, price and features and choose — the way your real buyers do.

03
We tune it to reality

We fit the model to your actual sales, so its numbers match your real market — not just a plausible-looking guess.

04
You ask any what-if

Change a price, plan a promo, launch a SKU, react to a competitor — get a straight answer, and how sure we are.

02Proof · believable ≠ accurate
Everyone can simulate a customer. We prove ours predicts the real sale.

We tested it on real sales it had never seen — so the accuracy is real, not just fit to the past. That held-out proof is the one thing “synthetic shopper” tools don’t show.

◇ Simulation vs real sales (held-out)

Bar length = how close each lands to the true market share. Watch calibration snap onto reality.

Raw sim
off by ~26×
Calibrated
≈ real
Real sales
truth
Raw agent simulation looks believable but lands far from the truth. Adding our calibration layer moves share-recovery from −0.35 to +0.44 — the gap between “looks right” and “is right.”
Held-out accuracy · R² (0–1)
0.64
How well we predict real market shares on sales the model never saw. 1 = perfect; strong for a blind test — and something no “synthetic shopper” tool publishes.
−2.5%
simulated price sensitivity matched real, almost exactly
2
independent public supermarket datasets validated on
1 wk
even a few weeks of your data beats guessing
03Applications
What you can check before you commit.

The three decisions your team makes money (or loses it) on.

Pricing & promotion

Find the price that makes you the most profit. See if a promo actually pays — or just gives margin away. Know who you win shoppers from, and who wins them from you.

New products & assortment

Thinking of adding, dropping, or launching a product? See how much it'll sell and how much it eats into your own products — even with no sales history yet.

The shift to AI shopping

When shoppers (and their AI assistants) ask "what should I buy?", are you the answer? See what AI-driven shopping does to your demand.

04Get started

Request a demo

We'll run the live simulator on your own category, calibrated to your real sales — and send you the full reproducible benchmark report. Access to the simulator is granted after the walkthrough.

SENDING…
AVANTI · MARKET TOWN