−41%
partners below the minimum price in 10 days
It was 59. Now it is under 35. The brand manager worked from the AI agent’s list.
Case study · Brand and reseller price control
Brand control on marketplaces: price violators down 41% in 10 days, 250 third-party offers removed from the brand’s listings
We built a system of AI agents for a DIY goods manufacturer with its own brand. The brand sells through its three own stores on Allegro, on eMAG and Kaufland, and through a network of B2B partners. Partners used repricers, split packs into single pieces and pushed the retail price below the wholesale price. Every 2 hours, the AI agents check 7,481 offers for 1,085 brand products. They find sellers below the minimum price and third-party goods on the brand’s listings. The brand manager talks to the partners. A person always makes the decision.
Key figures
250 of 387
third-party offers removed from the brand’s listings in a month
The marketplace unlinked and ended them after complaints.
179 → 9
offers below the minimum at one of the main violators
In the 8 days after price control started.
7,481
offers from all sellers in each check
1,085 brand products. A check every 2 hours on our server.
Amounts, sales volumes, products, sellers and their names are changed under NDA. Percentages, time frames and offer counts are real. How we counted
Summary in 30 seconds
With AI agents, the brand
- Sees the price of each of its products at each seller on Allegro. A check every 2 hours
- Compares prices with the minimum price in the manufacturer’s price list, for packs and for single pieces
- Finds partners below the minimum price and gets an Excel dossier on each one at once
- Finds third-party goods on its listings and handles each complaint until it gets a result
- Checks its own stores by the same rules
- Sees how sales change when partners hold or break the price
- Gets alerts in the CRM and Telegram and does not search for violations by hand
Why price dumping is dangerous
One partner’s price dumping breaks the price for everyone
The brand sells in two ways: through its own stores in retail and through partners in B2B. When one partner cuts the price, both channels suffer. Then the partner suffers too: its margin also falls.
- 1
A partner uses a repricer
The software automatically sets a price slightly below the next seller. A second partner does the same. On one item, the price fell by 46%.
- 2
Retail is cheaper than wholesale
A buyer on the marketplace pays less than a distributor pays the manufacturer. For new B2B clients, it is cheaper to buy on the marketplace.
- 3
Brand ads work for the price dumper
On a listing, the marketplace gives the sale to the lowest price. The brand pays for ads and content, and the cheapest seller takes the sale.
- 4
Packs are sold as single pieces
Partners take goods out of the pack and sell one piece with the pack barcode. Advertising such products makes no sense: the marketplace shows them as the cheapest.
- 5
Reputation suffers
Third-party goods are sold on the brand’s listing. The brand gets the bad reviews. And the marketplace treats the brand store’s prices as too high.
Retail on the marketplace is cheaper than our price for distributors. For new clients, it is better to buy there than from us.
I never agreed to sell the product this cheap.
The brand stores’ repricer followed the cheapest partner. The low price started with the partners.
Background
For more than a year, nobody held the minimum price
The brand tried to introduce a minimum price policy before. Without facts on each seller, the agreements did not work.
-
No price policy
Marketplace review: almost all partners sell below the brand’s retail price. The minimum price policy does not work, even though the brand already tried to build price control.
-
The marketplace against the brand
In a big sale event, the marketplace treats the brand store’s prices as too high. The reason: partners sell cheaper.
-
Three scenarios for the year
We showed the brand three growth scenarios. In the best one, partners do not dump prices, and the same revenue needs less ad spend.
-
First check of all sellers
The AI price agent went live in the CRM. First check: 6,700 offers and 2,122 violations.
-
Repricer race
Partners use repricers. On one item, the price fell by 46%. About 40 packs are sold as single pieces. Brand ads work for price dumpers.
-
Turning point
The B2B sales manager sees that retail is cheaper than wholesale and sends the minimum price list. The same day, the AI agents compare the price list with the brand stores: 403 of 914 products sell below the minimum.
-
Control starts
The brand stores raise prices to the minimum. Every day, the brand manager works with partners from the AI agent’s list.
Since 19 October, the manager’s day starts with a list: who sells below the minimum price, where and by how much.
- Violators found
- Excel dossier
- Third-party offers on listings
- Alert every 2 hours
People keep the talks: calls, emails and agreements.
What the AI agents had to do
- See the price of each brand product at each seller main goal
- Find minimum price violations for packs and for single pieces main goal
- Give the manager facts on each partner for the talk very important
- Find third-party goods on the brand’s listings and remove them through complaints
- Check the brand’s own stores by the same rules
- Show how partner prices affect the brand’s sales
Solution
Brand AI agents instead of manual searches on the marketplace
Nobody can check 7,481 offers by hand. The AI agents do it every 2 hours. They collect the prices of all sellers, compare them with the price list and report each violation. The brand manager calls and emails the partners.
Where the AI agents get data
Prices of all sellers
TradeWatch: each offer for each brand EAN, seller NIP (tax ID), stock
Brand stores
Allegro API: own prices every hour, the Allegro Ceny program, ended offers
Manufacturer’s price list
Minimum price per piece with VAT, 1,081 items
Partner list
Who is a brand client and who is an outside seller
Marketplace mail
Replies to complaints, every hour
Brand AI agents
- server collection agent
- price agent
- listings agent
- complaints agent
- task agent
What the team gets
Violations in the CRM
Who, which product, how far below the minimum
Seller dossier
An Excel file for each partner: all its offers and prices
Alerts
CRM notification bell and Telegram: new violations and failures
Complaints
A case for each third-party offer, from the complaint to the marketplace reply
Brand manager
Calls and emails partners. The AI agents send nothing to partners
Five AI agents and what they do
AI collection agent
Runs on our Ubuntu server. Every 2 hours, it opens the TradeWatch web panel on its own and builds a report for 1,085 EANs in batches of 499. It sends the report to the CRM. A full pass takes about a minute.
AI price agent
Compares the price of each offer with the minimum price from the price list. For a pack, it multiplies the minimum by the number of pieces. It shows single-piece sales from packs separately. It checks the brand stores by the same rules.
AI listings agent
Looks for third-party goods on the brand’s listings. Flags: the offer title contains the product code of another brand product, or the price is 50% or more outside the normal range.
AI complaints agent
Keeps a case for each third-party offer. Every hour, it reads the marketplace replies in the mailbox and sorts the cases into groups. A person checks the text and then enables complaint filing for a group.
AI task agent
Tracks the managers’ tasks in Telegram. It reminds them of deadlines and does not close a task without a result: a file or a link.
How it works every day
- STEP 01
Collect prices
Every 2 hours, the AI collection agent gets the prices of all sellers for all brand EANs.
- STEP 02
Find violations
The AI price agent checks each offer against the price list and marks new violations.
- STEP 03
Prepare facts
For each partner, the CRM builds an Excel dossier: products, the partner’s prices, the minimum, the difference.
- STEP 04
The manager talks to the partner
A call and an email with prices. The manager posts the result of the day in the team’s shared chat.
- STEP 05
Check the result
The next check shows if the partner raised its prices. If not, the talk moves to the next level.
What counts as a violation
Eight rules that keep price control honest
On a marketplace, a simple “price below the minimum” rule gives hundreds of false alarms. The AI agents check how each product is sold.
Minimum per piece
The minimum price equals the price list price × 1.23 (VAT) for one piece.
Pack
The AI agent checks a pack of 10 pcs against the minimum × 10. Otherwise the pack looks 90% cheaper.
Single-piece sales
Sellers sell a piece from a pack with the pack barcode. The AI agent shows such offers in a separate filter.
Brand stores too
The AI agents check the brand’s own stores by the same rules. The brand’s repricer can also go below the minimum.
A violation closes by itself
When a partner raises the price or ends the offer, the violation leaves the list at the next check.
The marketplace cuts the price itself
The Allegro Ceny program sometimes gives the buyer a price below the minimum. The AI agent sees such offers, and you can exclude them from the program.
Third-party goods on a listing
The product code of another brand product in the offer title, or a price 50% or more outside the normal range.
A manual price wins
If a product has a special minimum price, the manager sets it in the CRM. The CRM recalculates at once.
Each rule is simple on its own. Together, they separate real price dumping from data errors. So the manager calls a partner with facts, not with suspicions.
How the managers work
From an alert to an agreement with the partner
The AI agents find and calculate. A person talks to the partners. Every day, the manager takes the list of violators, calls, emails and posts the result in the team’s shared chat.
- 1
Alert
The AI agent found a violation. The manager sees the partner, the products and the difference from the minimum price.
- 2
Call and email
The manager calls the partner and sends an email with prices from the price list. The partner’s Excel dossier is at hand.
- 3
Check the next day
The AI agent shows if the partner raised its prices. Large partners usually fix prices the next day.
- 4
Stock repurchase
A partner may stop buying and clear its remaining stock at a low price. Then the brand offers to repurchase the stock and pay after it accepts the goods.
- 5
Supply decision
If a partner breaks the price again and again, the brand owner decides the next step. The owner approved strict measures, up to a review of supplies.
Today, fewer than 35 sellers undercut prices. Yesterday there were 45.
One of the main violators has 9 offers below the minimum left, out of 179.
Only 5% of the violations are left at the main violators. The other 95% are isolated offers.
What the numbers showed
A minimum price works only when everyone holds it
On 19 October, the brand stores raised prices to the minimum. The AI agents compared sales before and after by product group. Almost all of the drop came from products where a partner stayed cheaper.
- A partner is cheaper all the time 161 offers −42.6%
- Nobody is cheaper 103 offers −7.2%
- Control group price did not change −6.2%
Periods: 6–17 and 20–31 October 2026. Percentages are real.
59 19 Oct
43 23 Oct
45 28 Oct
under 35 29 Oct
The B2B sales manager’s report in the work chat, based on the AI agent’s list.
The lesson for any brand: if you hold the price alone, sales go to the sellers who do not. You need control over all sellers at once and every day.
Brand reputation
Third-party goods on the brand’s listings: 250 of 387 removed
An outside seller puts its own product on the brand’s listing because the barcode matches. The buyer gets the wrong product, and the brand gets the bad review. The marketplace does not remove such offers on its own.
- 1
Find
The AI listings agent finds a third-party offer: a foreign product code in the title or a price outside the normal range.
- 2
Complain
A complaint through the marketplace form: the offer is linked to the wrong product.
- 3
Read the reply
Every hour, the AI complaints agent reads the replies in the mailbox and moves the case to the right group.
- 4
Prove
If the marketplace refuses, the manager makes a test purchase and photographs the product.
- 5
Check again
A removed offer can return. The AI agent checks the listing and files a repeat complaint.
250 120 12 5
- 250 offers unlinked and ended
- 120 marketplace refusals next step: test purchase
- 12 no reply
- 5 “no differences found”
Where the brand is under control
Allegro, Kaufland, eMAG, Amazon and B2B partners
Each marketplace has its own data and its own limits. The pattern is the same: marketplace data, brand rules, an alert to the manager.
Allegro every 2 hours
The brand’s main market. Prices of all sellers for each EAN, third-party offers on listings, complaints, the Allegro Ceny program, ended offers.
- 1,085 EANs and 7,481 offers in each check
- An Excel dossier for each seller
- The AI agent handles complaints about third-party offers
Kaufland 7 storefronts
Sales go through BaseLinker to 7 storefronts: DE, AT, CZ, SK, IT, FR, PL. We compare the actual price of each sale with the catalog price.
- Price of each sale by storefront
- Of 187 pieces sold, only 11 sold at the catalog price
eMAG account health
Romania, Bulgaria, Hungary. Here, seller reputation means account health. eMAG closed the store in Romania because of missed emails about an account test.
- Sales through the API in the shared analytics
- Appeal for the Romanian store filed
- Lesson for all marketplaces: an agent must read marketplace emails
Amazon Brand Registry
The brand is in Brand Registry. We connected SP-API access on 11 EU marketplaces through our OAuth bridge.
- Competitor snapshot: prices, rating, sellers
- We studied the Transparency program requirements
B2B partners every day
The list of brand partners is in the CRM. For each partner, you see where it sells, at what price and how many violations it has.
- 59 partners with violations at the start
- The manager works from the list every day
The AI agents check Allegro most often: the brand has its main sales and the most resellers there. We connect a new marketplace with the same pattern.
Alerts
The AI agent writes first when someone breaks the price
New violations come to the CRM notification bell and to Telegram. The manager does not search for price dumping by hand. The day starts with a list.
6 types of alerts
- New minimum price violations
- Third-party goods on a brand listing
- Offer back on a listing after removal
- Test purchase needed for a complaint
- Allegro Ceny cut the price below the minimum
- Collection skipped: data quota ran out
59 → 34 partners with violations in 10 days
393 third-party offer cases that the AI agent handles
2 h between checks
AI price agent bot
27 October
🔔 AI price agent: 14 new violations
🔴 Below the minimum price (11)
• Partner A: Facade scraper 350 mm
27.90 zł with a minimum of 34.44 zł, −19%
• Partner B: Brush set 5 pcs
41.50 zł with a minimum of 49.20 zł, −16%
🟡 To check (3)
• Single-piece sale: Plastering trowel
A single piece sells with the barcode of a pack of 6 pcs. Check the pack size.
🟢 Partner C raised prices
• 170 offers are now at or above the minimum price
9 of 179 offers are left. Dossier updated.
AI agent findings on screens
Brand control on four CRM screens
The AI agents work inside the CRM that we built for the brand. Products, sellers and amounts on the screens are replaced.
01 Partners
All sellers whose brand products are cheaper than the minimum price. Totals at the top, the list by number of violations below.
02 Seller dossier
All offers of one partner: product, partner price, minimum price and difference. The “Download Excel” button gives the same list for the email to the partner.
03 Hijackers
Offers from outside sellers on the brand’s listings. The AI listings agent marks the flag: a foreign product code in the title or a price outside the normal range.
04 AI complaints agent
A case for each third-party offer: complaint, marketplace reply, next step. The AI agent reads the mailbox every hour and moves the case to the right group on its own.
Results
Violators down 41% in 10 days, and third-party offers leave the brand’s listings
We compared 19 and 29 October 2026. The work continues: two large partners still keep some prices below the minimum.
| Metric | Before | After | Result |
|---|---|---|---|
| Partners below the minimum price | 59 | under 35 | −41% in 10 days |
| Offers below the minimum at one of the main violators | 179 | 9 | −95% in 8 days |
| Third-party offers on the brand’s listings | 387 complaints | 250 removed | 65% October 2026 |
| Brand store products below the minimum | 403 of 914 | prices raised | in 6 days from 13 to 19 October |
| From the price list to work with partners | Price list on 13 October, “Partners” section in the CRM on 15 October, first calls from the list on 19 October | 6 days | |
| Brand sales where everyone holds the price | 12 days after the price rise compared with 12 days before | −7.2% control group −6.2% | |
| Brand sales where a partner is cheaper | The same periods | −42.6% so control must cover all sellers | |
Is your brand losing its price on marketplaces? Start with a 60-minute consultation. We review your channels, partners and prices and show what to hand to AI agents first.
Book a consultationHow we counted
- Period: 19–29 October 2026. Number of partners with violations: the B2B sales manager’s daily report in the work chat, based on the AI agent’s list.
- Minimum price: the manufacturer’s price list, price × 1.23 (VAT) per piece. Pack: minimum × number of pieces.
- Prices of all sellers: TradeWatch reports for all brand EANs, a check every 2 hours.
- Complaints: marketplace replies to complaints through the “offer linked to the wrong product” form from 2 to 30 October 2026.
- Sales before and after: brand store offers with a price rise on 19 October. Periods: 6–17 and 20–31 October 2026.
- Sales amounts and volumes are changed under NDA. Products, sellers and their names are replaced. Percentages, time frames and offer counts are real.
- Other factors: the end of the renovation season, warehouse stock, price changes. After the price rise, brand store sales are still lower than before it. Work with two large partners continues.
How the project went
From the first price check to the AI complaints agent: 4 months
- JULY 2026
AI price agent
Prices of all sellers for the brand EANs through the TradeWatch API. First check: 6,700 offers and 2,122 violations.
- 30 SEPTEMBER
Hijackers
A CRM section for third-party goods on the brand’s listings. At the start: 371 offers from 53 sellers.
- 4 OCTOBER
Agent on our own server
Every 2 hours, the AI collection agent on the Ubuntu server builds TradeWatch reports for 1,085 EANs.
- 13–19 OCTOBER
Minimum prices
Manufacturer’s price list in the CRM, the “Partners” section, Excel dossiers. Brand stores at the minimum, the manager calls partners.
- OCTOBER
387 complaints
Complaints about third-party offers through the marketplace form. 250 offers unlinked and ended.
- 4 NOVEMBER
AI complaints agent
A case for each third-party offer and marketplace replies from the mailbox every hour. 393 cases.
What was hard
The marketplace does not share other sellers’ offers
The marketplace API shows only your own offers. The AI agents get the prices of all sellers from TradeWatch: through the API and through the web panel on our server.
Anti-bot protection blocks pages
The marketplace blocks data center servers, and the marketplace rules forbid scraping with bots. So the AI agents work only with permitted data sources.
Packs and pieces
Sellers sell a piece from a pack with the pack barcode. If you ignore the pack size, a pack looks 98% cheaper. The AI agent checks the pack size against the title and the price list.
A barcode does not prove the owner
The marketplace replies: the GTIN identifies the product, not the owner. A complaint needs a test purchase, trademark documents and the logo in the offer.
The brand’s own stores dumped prices too
The brand stores’ repricer followed the cheapest partner and set the price 0.4% lower. So the first step was to raise the brand’s own prices to the minimum.
What we need from the brand
- A price list with minimum prices per piece
- A list of partner clients
- A manager who talks to partners every day
- Trademark documents for complaints
Who it fits
Is brand control right for your business?
It fits if
- You have your own brand, and partners and distributors sell it
- You sell on marketplaces where many sellers share one listing
- Partners complain that retail is cheaper than wholesale
- Third-party goods appear on your brand’s listings
- You have 100+ products, and nobody can check the prices by hand
- You are ready to agree on price rules with partners
You don’t need it if
- You are the only seller of your product
- You have no partners and no B2B sales
- You have 10–20 products: a weekly check is enough
Questions
What brands usually ask
Can we require a minimum price from partners?
In the EU, a brand cannot strictly fix the resale price. A recommended price is allowed. So the brand checks the partner rules with its lawyer. The AI agents give the facts: who sells, where and at what price. The brand makes the decisions on contracts and supplies.
Does the AI agent write to partners on its own?
No. The AI agents find violations and prepare dossiers. The brand manager calls and emails partners. The AI agent files complaints with the marketplace only for groups that a person enabled.
How often do prices update?
Prices of all sellers update every 2 hours. The AI agents check the brand stores’ own prices every hour.
Where do the AI agents get other sellers’ prices?
From TradeWatch, a marketplace analytics service. The marketplace API does not share other sellers’ offers, and the marketplace rules forbid page scraping.
What if a partner sells a pack by the piece?
The AI agent sees the pack size. It checks a single piece against the minimum per piece and a pack against the minimum per pack. Single-piece sales with the pack barcode go to a separate filter.
Does this work on Amazon, Kaufland and eMAG?
The pattern is the same: marketplace data, brand rules, an alert to the manager. Each marketplace has its own set of data. We start with the marketplace where the brand has the most sales and resellers.
How long does the launch take?
In this project, price collection worked from July. The AI agents compared the price list with the marketplace on the day the price list arrived. The “Partners” section went live in the CRM after 2 days. After 6 days, the manager worked with partners from the list.
Do sales drop when a brand raises the price to the minimum?
Yes, if partners stay cheaper. Where a partner was cheaper, brand store sales fell by 42.6%. Where everyone held the price, they fell by 7.2%. So control must cover all sellers at once.
SellerUp · AI automation for e-commerce
A 60-minute consultation on your brand control
We review your channels, partners and prices and show what to hand to AI agents first. You get a written plan within 48 hours. The price is $249.
Where to start
A 60-minute consultation: how to regain control of your brand’s price
A working session on your brand. We look at your channels, partners, prices and listings. You get a written plan: what to automate first, in what order and what it costs in hours and money.
- 60 minutes, online
- Estimate in hours and money
- Written plan within 48 hours
Consultation · 60 minutes, online
$249 $399
Credited towards the next piece of work if it starts within 30 days
Book a consultation or leave a request belowConsultation request
What eats the most time right now — yours or your team’s?
- Reply within 1 business day
- We send a meeting time and a payment link
- Honest go / no-go answer
- No “boilerplate proposal”