Solutions / Price optimization

Price & promo optimization

Price elasticity estimated from your own sales, turned into a recommended price and promo depth per SKU and segment — recalculated daily and tested before it goes live everywhere.

Pricing Revenue ML
Daily recalculation
across 100% of the assortment
Elasticity per segment
not one rule for the catalogue
Tested first
on a subset, against current prices
Who it is for

Pricing, category and e-commerce managers with a catalogue large enough that price is set by habit rather than analysis.

Short answer

A model estimates how demand responds to price for each product and segment, using your own transaction history, competitor moves where you track them, and past promotions. It recommends a price and a promo depth within the guardrails you set — floors, MAP, brand rules — and every change is trialled on a subset before it touches the whole catalogue.

The problem

Prices are set by a markup rule and moved when a competitor moves, and nobody knows what either decision cost.

01

One markup for everything

A flat margin rule prices a product with loyal demand exactly like a commodity where buyers compare three tabs. You leave margin on one and volume on the other, every single day.

02

Promo depth is a guess

Thirty percent is chosen because thirty percent was chosen last time. Nobody separates the customers who would have bought anyway, so the discount is booked as a success and repeated.

03

Competitor moves trigger reflex cuts

A price drop somewhere gets matched within a day, across products where the competitor is not even relevant. The reflex is fast, undirected, and directly subtracted from margin.

How it works

From trigger to result, step by step.

01

Rebuild the price and demand history

Prices, promotions, stock, competitor prices where you collect them, and the periods that must be excluded. Elasticity estimated on dirty history is a confident number pointing the wrong way.

02

Estimate elasticity where it is estimable

Products with enough price variation get a real elasticity per segment; products without it get grouped by category and attributes. We label which is which rather than presenting both as equally solid.

03

Set the guardrails with you

Price floors, minimum margin, MAP and brand agreements, maximum change per period, rounding and psychological price points. The model can only recommend inside what your business actually allows.

04

Test before rolling out

New prices go live on a subset of products or stores while comparable ones stay as they are. Margin and volume are compared across the two groups before anything is applied broadly.

05

Recalculate daily and monitor

Recommendations refresh with new sales, stock and competitor data, and land in your PIM or e-commerce platform. Elasticity is re-estimated on a schedule, because demand curves move.

Before / after

What changes on the ground.

Today, by hand
×One markup rule across the catalogue
×Promo depth chosen by habit
×Competitor cuts matched reflexively
×Nobody knows the margin cost of a discount
With the automation running
Price and promo depth recommended per SKU and segment
Discounts sized against measured elasticity
Competitive response limited to where it matters
Margin and volume effects measured on a test group
What you get

Delivered, not demoed.

A cleaned price, promotion and demand history you keep
Elasticity estimates per segment, with explicit confidence
A guardrail layer for floors, MAP, rounding and change limits
A test design comparing new prices against a control group
Daily recommendations into your PIM or e-commerce platform
Built with

We build in your stack rather than moving you onto ours. The list below is what this solution most often connects to — other systems are a scoping question, not a blocker.

Python statsmodels PyMC LightGBM BigQuery dbt Airflow 1C Shopify Power BI
Time to production8–12 weeks
Build priceFixed quote
First stepFree mini-audit
Honest limits

When this is not the right solution.

·If your prices are fixed by contract or regulation, there is nothing to optimise and we will say so in the audit rather than after the invoice.
·If prices have barely moved in two years, there is no variation to learn elasticity from. The honest first step is a structured price test, not a model.
·If a few large customers negotiate every price individually, this is a sales-process project, not a pricing-model one.

Questions we get about this one

No. Elasticity is often low, and the model raises price where demand is insensitive at least as often as it cuts. Both directions are constrained by the guardrails you set.

Bring us the process that hurts.

The mini-audit is free: we take your version of this process apart and tell you plainly whether automating it pays. If it does, you get a scope and a fixed price.