Pricing · Smart Pricing

What price should this offer have today?

Smart Pricing reads the product card behind your offer — every competing seller of the same item — keeps the offers that actually sell and recommends a price: competitive, balanced or above your margin floor. Every figure it used is printed next to the number. It reads; it never touches your Allegro.

  • One link, no account — a public offer link is all it needs; no Allegro login, no API key, no password.
  • The same item only — sets, other sizes and other products on the card are left out, and it says which.
  • The offers that sell — the median and the anchor come from offers with buyers; parked listings are counted, not priced against.
  • Three strategies and a floor — Competitive, Balanced, Margin; the price never goes below cost plus your margin.
Recorded in the cabinet on a demo account — real Allegro data
platform.marzi.ai
Data source
0credentials — a public link is the whole input
100offers in one run, pasted or from a CSV
3strategies: Competitive · Balanced · Margin
Read-onlyMarzi recommends; the price on Allegro stays yours to set
Why you need it

“Am I priced against the market — or against listings nobody buys from?”

Without it

  • A product card with fifty offers: two-pan sets, a 24 cm pan, a pancake pan and thirty listings without a single buyer — and the “market median” comes out at 209 zł.
  • The buy box sells 786 units at 110 zł while you sit at 199 zł, sure you are “below the median”.
  • A repricer that undercuts the cheapest offer by a grosz follows a typo and a dumping seller alike.

With Smart Pricing

  • The competitors that are the same item, one per seller, sets and other sizes out — with the reason for every exclusion.
  • The median and the anchor from the offers that sell; the demand-weighted median printed beside them.
  • A recommended price with its confidence, the reasons in plain words, and a warning when the move is large.

The price the buyers are paying, not the one the parked listings show

On a real Gerlach pan card, 786 of 901 buyers paid 110.44 zł while the median of all 51 offers said 209 zł. A seller at 199 zł was not “slightly under the market” — they were 80 % above the price that sells. Smart Pricing measures the market where the money goes and says so.

How it works

Three steps, the whole product

1

Paste the offer

One allegro.pl offer link — or up to a hundred, pasted one per line or uploaded as a CSV with a price and a cost column. Marzi reads the public offer page and the product card behind it. Nothing is asked of your Allegro account, because nothing is written to it.

  • A single link, a list, or a CSV with a preview before the run
  • Only allegro.pl offers; the field says so before the button lights
  • Every offer analyzed on its own — one bad link never stops the rest
Smart Pricing — 1: Paste the offer
2

Marzi keeps the same item and the offers that sell

On the product card, every seller’s offer is read with its title. Sets and multipacks, other sizes and other products are left out by rules, then by a judge model on what the rules cannot decide. One seller counts once. When enough competitors have buyers, they are the market; parked listings are counted and named, not priced against.

  • Sets, sizes and other products out, each with a reason
  • The buy box, the median, the demand-weighted median
  • A wide spread flagged: one card can hide two markets
Smart Pricing — 2: Marzi keeps the same item and the offers that sell
3

Read the price and why

The recommended price sits at the top with its confidence. Beside it your price and the median; under it the reasons — how many competitors are below you, where the buy box is, what the trend did. Switch to Competitive or Margin, type a cost, simulate any price: the same market, re-priced without another read.

  • Competitive · Balanced · Margin with a floor from cost and margin
  • What-if simulation that never touches Allegro
  • A warning when the calculated price differs a lot from yours
Smart Pricing — 3: Read the price and why
What is inside

Built from what Allegro actually shows

Every figure on the screen has the page it was taken from and the method printed next to it. These are the parts.

No credentials

A public offer link is the whole input. No Allegro login, no API key, no OAuth — Marzi reads what any shopper can see.

The same item only

Sets, multipacks, other sizes and other products attached to the card are left out — by rules for what words and numbers can tell, by a judge model for the rest — and shown in a list with the reason.

The offers that sell

When enough competitors have buyers in the last 30 days, the median and the anchor are theirs; the parked listings are counted, not priced against. The demand-weighted median shows where the money goes.

Three strategies

Competitive sits just under the leading offer; Balanced weighs it against the median, the trend and the competition; Margin is Balanced, never below cost plus your minimum margin.

Reasons, not a black box

Every statement comes from a figure that went into the calculation: competitors below you, the buy box, the median, the trend, what was left out. Confidence is scored from six readings and shown as a number.

Read-only, with a warning

Nothing is written to Allegro. A recommendation that moves your price by more than the threshold carries a warning; insufficient data is said, never guessed.

What the numbers mean

Measured, estimated, and not yet

Marzi says which is which on the screen itself. Here is the short version for Smart Pricing.

Measured

Competitor prices are read from the product card at the time of the analysis, with the time shown.
The recommended price is arithmetic over those readings — decimal, to the grosz, reproducible on the same signals.
Every exclusion is listed with its reason, so what was left out is yours to check.

Estimated or limited

An offer alone on its card has no market of its own: similar products from the index stand in, at lower confidence, or the analysis says insufficient data.
The judge model is a model — on a few titles per hundred it errs; the rules it sits behind are measured at 97 % precision.
The price history is real readings only: the chart appears from the second reading a week apart, never from a drawn line.
Who it's for

Three ways to use Smart Pricing

Starting out

Before the first listing goes up, paste a competitor’s offer and see the price the product actually sells at — not the one the parked listings show.

Already selling

A hundred SKUs in one CSV, one recommended price each with the reasons, and a re-run whenever the shelf moves.

Brands and agencies

The price the market clears at, per product, with the resellers who sell and the ones who only list — for a pricing policy built on readings.

FAQ

questions about Smart Pricing

Does Marzi need access to my Allegro account?

No. The input is a public offer link. Marzi reads the offer page and the product card the way a shopper would. There is no login, no API key and no write of any kind — the price on Allegro stays yours to set.

Which offers count as competitors?

The other sellers of the same product card, one per seller, the same single item only: a two-pan set on a pan card, a 24 cm pan on a 28 cm card, a pancake pan or a bundle with a case are left out and listed with the reason. When enough of the rest have buyers in the last 30 days, those are the market.

What do the three strategies do?

Competitive sits just under the leading offer — the buy box or the cheapest selling competitor. Balanced weighs that point against the median, pulled towards it when the market is falling or crowded. Margin is Balanced, never below your cost plus the minimum margin you enter.

What is confidence?

A score from six readings: how the product was matched, how many relevant competitors were priced, how complete and fresh the data is, how tight the spread is, and whether there is history. It is not a probability of a sale, and it is shown as a number, not only a colour.

What happens when there is not enough data?

The analysis says so and names the reason — product not matched, fewer than two relevant competitors, stale market data — instead of showing a price. Nothing is computed artificially.

Can I analyze many offers at once?

Up to 100 in one run: pasted one per line, or a CSV with offer_url and optional current_price and cost columns, previewed row by row before the run. Duplicates are analyzed once; a link that fails never stops the others.

Start with a free trial

7 days, no card required. The whole index, your trackers and the MCP endpoint are live from your first sign-in.