kaimersa
IN DEVELOPMENT JOIN WAITLIST ↓
§00

We don’t recommend
decisions. We prove them.

Kaimersa applies machine learning to the four decisions that set a retailer’s margin: what things cost, where they go, how they’re described, and what gets promoted. One demand model underneath all four.

Every recommendation ships with
a.A held-back control group, so lift is measured and not modelled
b.Performance read while the decision is still live
c.A decision at the end of it: scale across the estate, or drop
§01
METHOD

Four decisions, one loop. Each one hands its evidence to the next.

01 Price What each item should cost and when it should move, tested against a held-back control rather than a forecast.
02 Allocation Where the units go, informed by the price that was actually set and the demand it actually produced.
03 Content Copy and attributes written for what is in stock where, and rewritten when the allocation changes.
04 Campaign Spend pointed at the products, ranges and prices the loop has already shown can carry it.
§02
PROOF

Most models report the number they predicted. We report the measured difference between where a recommendation was applied and where it was deliberately held back.

MARGIN INDEX, WEEKS 1–12 CONTROL TREATED ILLUSTRATIVE
W1W4W8W12

Nothing ships without a control. The split is agreed before the model runs, not chosen afterwards to flatter the result.

Read in-flight, weekly, against that control — not in a post-mortem a quarter after the season closed.

Then it scales or it stops. We will tell you when a recommendation didn’t work, because you can see it anyway.

§03
STACK

This is not a rip-and-replace. We read from your POS PIM PLANNING and PROMOTIONS systems and write recommendations back into them. Nothing is migrated, replatformed or retired to make room for us.

§04
ACCESS

Watch this space

We’re working with a small number of retailers ahead of launch. Leave your details and we’ll be in touch when we open up.

TYPICAL REPLY — WITHIN TWO WORKING DAYS

Used only to contact you about early access. No newsletter, no list sharing.