Wholesale is measured in accounts, not in orders. That single sentence is the whole shift, and it explains why a seller who can recite their advertising cost of sales to two decimal places often has no idea whether their wholesale channel is working.
On the marketplace you have thousands of small events, so averages are meaningful and a dashboard updates daily. In wholesale you have a few dozen large events, so averages mislead, one account can distort a quarter, and nothing updates unless you write it down. The measurement has to be built for that shape: fewer numbers, checked monthly, with the account as the unit.
The seven numbers worth keeping
Accounts opened. How many companies placed a first order in the period. This is the top-line measure of whether the channel is growing.
Accounts contacted per account opened. The cost of acquisition expressed in effort. If it takes 90 contacted companies to open one account, either your list or your targeting needs work before you send more email.
Revenue per account. Total wholesale revenue divided by accounts that ordered. Watch the distribution rather than the average, because one chain among thirty independents makes the average meaningless.
Days to first reorder. The single most predictive number in the channel. A reorder at week six says the product sells through. Nothing at week twenty says it is sitting on a shelf regardless of what the buyer tells you.
Reorder rate. The share of accounts that ordered a second time. Below half, the problem is almost always sell-through rather than your sales effort.
Contribution per case. Wholesale price minus landed cost minus freight minus allowances. Not gross margin, contribution, because allowances and freight are the costs that quietly eat a wholesale channel.
Sell-through per store per week, wherever a retailer will share it. This is the number the buyer uses to decide your future, so it should be a number you track too.
Seven is enough. The temptation is to build twenty, and the twentieth is never the one that changes a decision.
Where the data actually comes from
None of this exists in Seller Central, and there is no single system that produces it. It comes from four places you already have.
Your invoices give revenue, units, dates and the account. If nothing else is in place, the invoice history alone produces five of the seven numbers.
Your account record gives the outreach side: contacted, replied, sampled, quoted, ordered. This is why the record has to be account-shaped rather than order-shaped, an argument made in full in tools for wholesale.
Retailer reports, where they exist. Larger chains share sell-through through a supplier portal. Smaller ones will tell you if you ask in the right way, which is to ask the buyer how the item is performing per store rather than asking for a report.
Your own cost data for contribution: the landed cost, the freight per case, and every allowance you agreed. The bookkeeping treatment is covered in wholesale accounting for Amazon sellers.
A monthly hour with those four sources produces a truer picture than any live dashboard, because the underlying events are monthly anyway.
Build it by cohort, not by total
The most useful structure is a cohort table: accounts grouped by the month they first ordered, with what they have bought since.
| First ordered | Accounts | First order | Since |
|---|---|---|---|
| January | 6 | $9,400 | $21,800 |
| February | 4 | $5,100 | $7,300 |
| March | 9 | $12,600 | $12,600 |
Read across a row and you learn whether accounts from that month came back. Read down the last column and you learn whether the channel compounds. In the made-up table above, January accounts have more than doubled their first order, February accounts have barely moved, and March is too recent to judge. That difference is a question worth answering: different account type, different product, different follow-up.
This structure also fixes the most common misreading of a wholesale channel, which is judging a month by its revenue. A month with one large opening order looks better than a month that opened six accounts which will reorder for three years. The cohort view shows which month was actually better.
Separate the channels in your numbers
Keep wholesale revenue, costs and inventory visible separately from marketplace revenue. Netting them hides the thing you most need to know, which is whether each channel earns on its own.
Three specific separations matter. Revenue, obviously. Inventory, because a wholesale order consumes stock in a lump and a forecasting model built on daily marketplace sales will treat it as noise, then under-order for the reorder. And contribution, because the wholesale price is roughly half the marketplace price while the cost of goods is identical, so a blended margin number describes neither channel.
The comparison between them is worth doing deliberately once a quarter rather than continuously. Marketplace contribution per unit after fees and advertising, against wholesale contribution per unit after freight and allowances, with the hours each consumed noted beside them. The framework is in Amazon seller unit economics, and the inventory side in inventory split between Amazon and wholesale.
Measuring the lift a shelf gives your listing
Retail placement raises marketplace demand. Someone sees the product in a store, photographs it, and buys it later where their payment details are saved. That lift is real and most brands never measure it, which makes wholesale look less profitable than it is.
There is no attribution link between a shelf and a listing, so you measure it by geography and timing. Before the retail launch, record your marketplace units per month for the states the retailer operates in, and for the rest of the country as a control. After the launch, compare the two. If the retailer's states grow faster than the rest, the difference is your lift.
It is not a perfect measurement and it does not have to be. A rough number that is directionally right changes decisions: it tells you whether to pay a slotting fee, whether to fund a promotion, and how to value a chain that orders modestly but sits in high-traffic locations. The way that arithmetic feeds a fee decision is in slotting fees.
Watch branded search terms in your own reports too. A rise in people searching your brand name rather than the category is the clearest signal that the shelf is doing work beyond the units it sells.
The outreach funnel, measured honestly
The front end of the channel has its own four numbers: contacted, replied, in conversation, ordered. Track them by account type, because independents, specialty chains and distributors behave nothing alike and a blended reply rate tells you nothing.
When the numbers are poor, diagnose in this order. A low reply rate is usually the list, not the copy, because a good message to a company that does not stock your category still produces silence. A good reply rate with few conversations means the offer is unclear, most often the price or the case pack. Conversations that do not become orders usually mean a readiness gap: packaging, terms, minimums or lead time. And orders that never reorder mean sell-through, which is a product and placement problem rather than a sales one.
That diagnostic order saves months, because the instinct is always to rewrite the email first. If the list is the weak link, paste your listing into WholesalePilot and the preview shows which distributors and retailers would plausibly stock your product, which is the cheapest fix available to the top of the funnel.
Questions Amazon sellers ask about wholesale analytics
Do I need a dashboard? No. A spreadsheet updated monthly is the right tool for a channel with monthly events. Build the dashboard when someone other than you needs to read it.
How long before the numbers mean anything? Two quarters. Days to first reorder needs one full buying cycle, and cohorts need at least three months to compare.
What if a retailer will not share sell-through? Estimate it from their reorder pattern. An account that reorders every eight weeks at a known quantity across a known store count gives you units per store per week within a useful range.
Should I count samples as a cost? Yes, as an acquisition cost against accounts opened. Samples are one of the larger hidden costs of a first wholesale year.
Which number would you keep if you could only keep one? Days to first reorder. It predicts the health of the channel earlier and more reliably than revenue does.