Unauthorised sellers and minimum advertised price violations arrive as two different complaints, usually from two different teams. Sales notices that channel partners are unhappy about being undercut; legal notices that products are being sold by parties with no agreement. Both are the same monitoring problem underneath: you need to know who is selling your products, in which markets, at what advertised price, and how that compares to what you authorised.
Build the collection once and both programmes run off it. Here is how.
Start from your own authorised list
The prerequisite is unglamorous and it is where most programmes stall: a maintained record of who is allowed to sell what, where, and under what pricing terms.
Without it, every finding is ambiguous. A seller you cannot classify is not a violation, it is a question for someone in sales, and a queue full of questions gets abandoned. The list needs seller identities as they appear on each marketplace, which is not the same as their legal entity name, plus the markets each is authorised for and the MAP applicable per product per region.
Keep it current. The most common cause of false positives in this work is an authorised partner who opened a new storefront that nobody added to the list.
What to collect
For each product and market, you want the complete offer set rather than just the buy-box winner, because the violations frequently sit below the winning offer where nobody looks.
Per offer, capture the seller name and identifier, the advertised price and currency, any strike-through or promotional pricing, shipping cost where it affects effective advertised price, the marketplace and storefront, stock status, and a timestamp with the market the check ran from.
Two collection details matter more than they seem. Prices are localised, so an offer must be read from inside the market it applies to, which is the same requirement as any price monitoring work. And promotional mechanics differ: some marketplaces show a discount only in the cart, which means a naive price capture understates violations and a thorough one has to follow the offer further.
The MAP comparison is not a simple threshold
Comparing an advertised price to a MAP number sounds trivial and produces enormous noise if implemented that way. Four adjustments cut most of it.
Compare like with like. MAP is usually about advertised price, so bundle listings, multi-packs, and refurbished or open-box items need separate treatment or exclusion rather than being flagged against the single-unit floor.
Handle currency and tax deliberately. A price in another currency needs a conversion rule and a tolerance, and whether tax is displayed inclusively differs by market, so the comparison has to be defined per region rather than globally.
Require persistence. A price that appears below MAP in one check and not the next is often a pricing error, a flash promotion, or a scraping artefact. Requiring two or three consecutive observations before flagging removes a large share of false positives at almost no cost in detection time.
Set a materiality floor. A violation of a fraction of a percent is usually not worth a partner conversation, and flagging it damages the credibility of the ones that are.
Classifying sellers
For the unauthorised-seller side, the output should be a classification rather than an alert.
Match each observed seller against your authorised list, and treat unmatched ones as candidates rather than violations. Then enrich: how long the storefront has existed, what else it sells, whether it appears across multiple marketplaces, whether shipping origin is consistent with legitimate distribution, and whether the same operator appears under several names. A single unauthorised seller is a compliance matter; the same operator across six storefronts is a diversion problem worth a coordinated response.
The overlap with counterfeit detection is real, and the two should share a queue, since an unauthorised seller offering product at an implausible price is frequently selling something that is not yours, which is covered in detecting counterfeit listings.
Evidence, because both paths end in a conversation
Whether the next step is a partner discussion or a legal notice, a finding needs to be demonstrable after the listing changes.
Capture a full-page screenshot showing the offer and price together, the listing and seller URLs, the marketplace identifiers, the price with currency, the market the check ran from, and a timestamp. Keep the raw capture rather than only the derived verdict, since disputes are usually about interpretation.
For MAP specifically, a time series matters as much as any single capture. A partner who was briefly below MAP and corrected it is a different conversation from one who has been under it for three weeks, and only continuous monitoring can tell them apart.
Turning findings into action
Route by type. MAP violations by an authorised partner go to the account owner with the evidence attached and a defined escalation ladder. Unauthorised sellers go to marketplace enforcement or legal depending on your policy. Repeat operators go to whoever owns your diversion investigations, because the useful question there is where the product is leaking from rather than which listing to remove.
Measure outcomes: time from violation to correction, recurrence by partner, and the proportion of unauthorised sellers successfully removed. Volume of flags is an activity metric and tells you nothing about whether the programme works.
One note on how MAP is used: this is monitoring of publicly advertised prices for enforcement of your own agreements. It is not a mechanism for coordinating prices between independent sellers, which is a different thing entirely with real legal exposure, and that boundary is worth having counsel define for your programme rather than inferring it.
Keeping the collection reliable
Marketplaces are defended and your data quality depends on collection integrity. Pace politely and spread sweeps, validate that a captured page is a real offer page rather than a challenge, and monitor per-market success rates so a silent failure is not read as compliance. The instrumentation is the same as any ongoing collection, per monitoring a scraping pipeline, and the failure mode to fear is a market that quietly stopped reporting while everyone assumes it is clean.
The bottom line
Unauthorised sellers and MAP violations are one collection problem and two workflows. Start with a maintained authorised-seller and MAP record, because without it every finding is a question rather than a violation. Collect the full offer set per product per market, read from inside each market since prices are localised. Make the MAP comparison careful rather than literal: compare like with like, define currency and tax rules per region, require persistence across checks, and set a materiality floor. Classify sellers instead of alerting on them, and enrich candidates before escalating. Capture evidence that survives the listing changing, keep a time series for MAP, and route findings to the team that owns the response. Then measure corrections and removals rather than flags.
The collection layer is brand protection with residential proxies, running on residential proxies with country and city targeting so every price is read from the market it applies to, billed per GB.