In short: Grey market leakage is an arithmetic outcome: wherever the gap between two of your own net prices exceeds the cost of moving goods between those two points, somebody will move them. Most of that gap comes from your own volume tiers and stale currency conversions rather than from anyone behaving badly. Fixing it with a resale price instruction is restricted in most jurisdictions, so the working instruments are differentiated ranges, functional discounts tied to things you can verify, and terms with a traceable serial trail behind them. Build the pairwise arbitrage width across your own price file before you build an enforcement policy.
The national distributor forwards a marketplace listing. Your part, correct part number, in stock, priced eight percent below what they pay you for it. They want to know who is dumping. Two weeks of tracing lot codes says the units left your plant on a legitimate order to an authorised dealer four countries away, one who had bought two quarters of stock in a single month to clear a volume threshold and then sold the surplus to a trader.
Nobody in that chain did anything unusual. The dealer bought at the price you offered, on the terms you wrote, and disposed of stock they no longer wanted. The listing that landed in your inbox is the output of a price structure working exactly as specified.
The width of the gap is the whole problem
Start with the calculation that tells you whether a leak is possible at all.
For any two points in your channel structure, take the net price at point A, subtract the net price at point B, and subtract the cost of moving a unit from B to A. Freight, any duty, repackaging or relabelling, and a risk allowance for warranty and returns. What remains is the arbitrage width, and it is the gross margin available to anybody willing to do the moving.
Take a mid sized industrial part. Net price to a stocking distributor in one market is 100. Net price to a dealer in a neighbouring market inside the same customs union is 78, because that dealer bought at a tier your first distributor cannot reach. Freight between the two is 6 per unit at pallet quantities, relabelling for local language is 2, and there is no duty. The arbitrage width is 14 per unit, or 14 percent of the higher price.
Fourteen points is a good business. It will be found, it will be worked, and it will be worked by people who carry none of the cost of demand creation, technical support or warranty handling. If the width had been three points, the same trader would have looked at the freight risk and gone elsewhere.
That reframes the question usefully. Enforcement is a way of raising the cost of the arbitrage. Structure is a way of removing it. The second is cheaper and it is entirely within your control.
Where the gap comes from, in order of size
Run the width calculation across your whole price file, every channel against every other channel and every market against every other market, and the large positive numbers cluster in a small number of causes.
Volume tiers reachable by buying ahead. A tier that a customer can reach by pulling forward two quarters of purchases converts a timing decision into a permanent price advantage on stock they never needed. This is the same mechanism that PP2 covers on rebate structure, and it is the largest single feeder of grey stock in most portfolios.
Currency lag. A price list issued annually in a local currency and left alone through a fifteen percent move has manufactured an arbitrage nobody decided to create. Markets with pegged or managed currencies hide this for a while and then release it all at once.
Launch and clearance pricing. End of life stock discounted to clear in one market is a fresh supply for another market where the product is still current. The discount was priced against local disposal value and is being realised against global list.
Tender and project pricing. A price granted for a specific installation, with volumes sized for that installation, becomes general trading stock when the project is descoped. Terms that tie the price to the named end use are the defence, and they only work if somebody checks.
Antia, Bergen and Dutta (2004), writing in MIT Sloan Management Review, made the observation that surprises most manufacturers when they first trace lot codes: grey market goods overwhelmingly originate inside the authorised channel rather than outside it. The distribution agreement is the supply route.
What you are allowed to do about the resale price
Before designing an intervention, be clear which levers are actually available, because the obvious one is restricted almost everywhere.
In the United States, minimum resale price maintenance was moved from per se illegality to a rule of reason analysis under federal antitrust law by the Supreme Court in Leegin Creative Leather Products v. PSKS (2007), and several states have taken a stricter line under their own statutes. In the European Union, resale price maintenance remains a hardcore restriction under Regulation (EU) 2022/720, the vertical block exemption regulation in force since June 2022, which also treats restrictions on the territories or customers into which a buyer may sell as hardcore, subject to specific carve outs for selective and exclusive distribution.
The practical consequence is that instructing a reseller what to charge is a route that needs legal advice in every jurisdiction you operate in, and in several of them the answer will be no. Everything below is about the levers that remain, which are about what you supply, to whom, and on what terms.
Functional discounts, priced against what each channel does
The durable structure pays each channel for the functions it performs, at a rate you can defend by reference to the cost of performing them.
Unbundle the distributor discount. Suppose it currently sits at 22 points off list, granted as one number. Take it apart: stocking and availability worth 8, credit extension to end customers worth 5, application and technical support worth 5, local demand generation worth 4. Now every point in that 22 has a function attached, and the price to a channel that performs three of the four is 18 rather than 22.
That arithmetic answers the question every conflict conversation eventually arrives at, which is why one partner pays less than another. It also gives a pure fulfilment reseller a defensible price rather than an arbitrary one, and it means an online reseller who genuinely does hold stock and does answer technical questions earns the discount for it.
The condition attached to this is measurement. A stocking allowance paid to a channel that no longer stocks is a price cut you have stopped noticing, so the functions need annual verification: stock file, average days of cover, support call volume, warranty handling rate. Where the verification cannot be done, the function should not be in the discount.
Making the leak visible
You cannot manage a leakage rate you have never computed, and most companies have not computed one because they believe they lack the data. They usually have more than they think.
Warranty registrations carry serial or lot codes. Match those codes against your own shipment records and you get, for every market, the share of units registered there that were shipped somewhere else. A figure of three percent is a nuisance. A figure of eighteen percent in one market means the local price is fiction and your local partner already knows it.
Where product carries a datamatrix or serialised code, the same match runs at unit level and can be automated. Building lot to finished goods linkage well enough to run that match is the subject of QQ4, and the commercial use of it is usually the easiest business case that capability ever gets.
Set a tolerance rather than a target of zero. Some leakage is cheaper to accept than to chase, particularly at the tail of the range where the volumes are small and the enforcement cost per unit is high. What matters is that the number is measured, reviewed, and attached to the specific accounts feeding it, because the conversation with a partner who is supplying a trader goes very differently when you can name the shipment.
The commercial cost of leaving it alone
Leakage prices itself into the structure faster than most teams expect, because the leaked price becomes the reference price.
Once a buyer can see a marketplace listing at 92 while your distributor quotes 100, the distributor's quote has to be explained rather than accepted. Their salesperson discounts to close, so your realised price falls in a market where nothing changed. The distributor then sees their margin compress on volume they are still servicing, reduces stocking depth, and the availability advantage that justified their discount weakens. Each step is individually rational.
There is a version of this that ends with the partner buying grey themselves, because at some point buying your product from a trader at 92 is better than buying it from you at 100. When that happens, your official channel and your grey channel have merged and the price file has stopped describing anything.
Where this stops
Structure removes the incentive and it does not remove every route. A product with high value density, no serialisation and a global spot market will leak regardless of how carefully you price, because the arbitrage only needs to exist for a few weeks a year to be worth somebody's time.
Enforcement has a cost curve that flattens quickly. Tracing lot codes, terminating an agreement, and defending the termination consumes commercial and legal time that is worth spending on the accounts feeding fourteen points of width and worth spending on almost nothing else.
And some price differences are genuinely defensible and genuinely unstable. Where two markets have different costs to serve, different regulatory requirements and different competitive sets, the right prices differ. If the gap that follows is wider than the freight between them, transparency will eventually close it whatever your policy says, and the honest planning assumption is that geographic price differences converge toward the cost of moving goods.
Take your current net price file, pick the six market and channel pairs with the largest spread, and compute the arbitrage width net of freight and relabelling for each one. Any pair showing more than about five points is already being worked, whether or not it has reached your inbox.