Competitive commerce intelligence
MelQart watches the competitors that matter to your catalog, shows you the evidence, and recommends the next pricing move — nothing changes without your approval.
Evidence on every recommendationGuardrails before commitmentYou approve every change
The ten-field decision card — every recommendation ships in this exact anatomy · sample catalog data
How it works
Every recommendation is the product of the same five steps — and step four is always you. Nothing executes on its own at any point.
MelQart monitors the competitor pages that overlap your catalog and records each price and listing change as a timestamped observation.
Changes are weighed against your own revenue — a move on a line that pays your rent outranks noise on one that doesn't.
You get a decision card: the change, the evidence, the confidence, the guardrails, and the exact price to set.
Nothing touches your store without your verdict — and each verdict tunes what MelQart brings you next.
Approved moves are tracked against their expected range and reported as modelled figures, misses included.
Why MelQart
Most dashboards fill silence with estimates. MelQart renders what it observed — and states, in plain words, what it refuses to compute.
THE HONEST GAP
A field that cannot be supported renders a gap with its reason — never a blank, never a guess.
WITHHELD MONEY
Money is never silently converted. A total that needs an assumption is withheld — and the refusal is shown.
THE BLOCKED APPROVE
A breach arrives blocked. Overriding is a separate, named, logged act — never a silent approval.
The product
A morning queue, a review lane, and a ledger of what your approvals actually did.
Today — a ranked queue of decisions, not a wall of charts.
Match review — uncertain matches wait for your verdict, and your verdicts retrain the matcher.
Outcomes — every approved move is measured against what it promised, misses included.
Integrations
Price writes go through your platform's official API and only ever carry a decision you approved.
Read-only sources feed observations and comparisons only. Where a platform is marked read-only, MelQart says so — it does not simulate a capability it doesn't have.
Pricing
Every tier gets the same honesty. Smaller tiers see fewer sources — never softer evidence.
A first competitor set, the decision queue, and full evidence on every card.
More sources, match review, guardrails, and outcome measurement for a working store.
Multi-currency reporting, deeper history, and priority refresh across larger catalogs.
Multi-storefront, custom guardrails and floors, and a named contact.
Questions
Confidence is decomposed — match quality, price recency, and margin-data completeness are shown separately, never blended into one opaque score. Anything below your auto-accept line goes to a review queue where you rule on it, and measured precision is reported only on matches you have actually reviewed. MelQart does not claim a universal accuracy number, because it cannot verify one.
Every claim carries the age of its newest observation: under 24 hours reads as current, 24–96 hours is marked stale, and beyond 96 hours a claim is treated as dead — its approve action is disabled until evidence refreshes. When a source is blocked or quiet, the surface says so instead of padding the gap.
No. Your margin floors, daily move limits, and ceilings are checked before a recommendation is issued — a move that would breach them arrives blocked, with the breach stated. And when competitors move up or you hold the cheapest position, MelQart recommends raises, not just cuts. Overriding a guardrail is a separate, named, logged act.
Nothing. Every price change requires your explicit approval and applies through your store platform's official API, with a 30-day rollback. Automation beyond that is opt-in, off by default, and unlocked gradually as your review history builds — limited to small, reversible moves inside your guardrails.
The Outcomes surface tracks each approved move against the range it was expected to deliver, on your own sales data. Expected figures are always labelled modelled, misses are flagged rather than hidden, and mixed-currency results are reported per currency — never blended into one number MelQart cannot defend.