Marketplace monitoring for loss prevention
Find your stolen products online. Then prove it.
onmark watches Wallapop, Vinted, eBay and Milanuncios. It matches every listing against your catalog, then turns suspicious sellers into evidence-backed cases.
The problem
Resellers move faster than any team can search by hand.
Stolen stock is listed under new names, in new cities, by sellers who come back. Manual searches and scattered screenshots leave analysts a step behind.
How it works
From marketplace listing to case file in four steps.
Monitor
Wallapop, Vinted, eBay and Milanuncios are scanned continuously for the brands and products you protect.
Match
Every listing is scored against your catalog using AI-generated aliases, price, location and image similarity.
Alert
Strong matches become prioritised findings and alerts, delivered to your team by email or webhook.
Investigate
Link listings, sellers and notes into a case with the full evidence chain, ready to share.
Explainable matching
Every match shows its work.
Analysts see why a listing was flagged, so they can verify the evidence instead of trusting a score.
- Name and alias matchBrand, model and known aliases compared with your catalog product.
- Price against the marketListings priced far below the market median are flagged.
- Location and coverageListings near the stores and regions you protect weigh more.
- Image similarityListing photos compared with your reference images.
Recurring sellers
Spot the seller behind the pattern.
onmark recognises the same seller across products, locations and marketplaces, then ranks them by risk so repeat offenders reach the top of the queue.
Cases
One case file.
Every piece of evidence.
Link alerts, listings and sellers to a case, add investigation notes, and keep a clear chain from the product in your catalog to the seller who listed it.
Missing Ray-Ban Wayfarer RB2140
Acme Inc. · Barcelona
- 01 / Product
Ray-Ban Wayfarer RB2140
Catalog
- 02 / Listing
Ray-Ban Wayfarer RB2140
166,67 €Wallapop
- 03 / Seller
urbanvault_es
Wallapop11 alerts
Reviewed linked findings against location inventory. Evidence is consistent, escalating for further review.
AI and agents
AI that investigates.
Analysts who decide.
onmark applies AI where it saves analyst time: understanding how sellers describe products, and running open-ended research across your data. Every conclusion stays traceable to evidence, and people make the final call.
Aliases that understand sellers
AI reads listing titles and descriptions to extract brand and model aliases, and generates aliases for your catalog, so listings match even when sellers name products differently.
Reports on demand
Ask for an investigation and the agent chooses the period, pages through all the evidence and delivers a written analysis with cited sources.
Reports you can hand over
Every agent report is saved in the platform and can be downloaded as a PDF, ready to share with colleagues and stakeholders.
Seller dossiers, assembled
For each seller worth a closer look, the agent gathers their profile, every linked finding and any trust signals into one write-up, with the evidence attached.
Agent run / Tenant reportRead-onlyExecutive summary
Example report214 findings were created in the period and none in the 30 days before.
The jump matches alerts created on 18 to 21 Sep, so it reflects wider coverage, not a market trend.
No profile data for 9 seller accounts. Missing data is not treated as a warning sign.
Every step, traceable.
tenant_data_coveragePicks the analysis period from real coverage dates
finding_changesReads every page: 3 pages, 214 findings
vendor_signalsFinds 4 sellers recurring across 3 locations
seller_profileDeep-dives urbanvault_es: findings and trust evidence
finding_evidenceCites 12 listings by their original IDs and URLs
People stay in charge.
Factual tools, not black-box scores
The agent queries tenant-scoped, read-only tools that return facts and evidence. It never scores a seller or declares fraud.
Facts, inferences and gaps kept apart
Every report separates what was observed, what is reasonable to infer, and what data is missing.
Traceable by design
Each run stores its prompt version, token use, every tool call and the evidence IDs behind each claim.
People stay in charge
Reports are read-only. Through MCP, the only write an agent can make is dismissing named findings, with a reason and explicit confirmation, logged as agent and authorised by an administrator or investigator.
Frequently asked questions
A clearer picture before you start.
How monitoring, matching and investigation fit together in onmark.
Explore onmark with your teamWhich marketplaces does onmark monitor?
onmark monitors listings on Wallapop, Vinted, eBay and Milanuncios for the brands and products you protect.
How are listings matched to my catalog?
The matching process considers brand and model names, relevant aliases, price, location and image similarity.
Does a match mean a product was stolen?
No. A match is an indication to investigate, not proof that a product or seller is involved in theft. The findings and related cases are available for manual review.
Can my team identify recurring sellers?
Yes. onmark groups related listings, sellers and products into cases that can be reviewed manually.
What can I bring together in a case?
A case brings related listings, sellers and products together for manual review.
What does AI do, and who makes the decisions?
AI identifies relevant product details and aliases, reviews related findings and builds cases. Analysts review the results before deciding what to do next.
See what is being resold
before it is gone.
Book a demo and we will walk through onmark with your own catalog and locations.
Built for loss-prevention teams, brand-protection analysts and investigators.