Marketplace data does not need to be fraudulent to mislead you. A dashboard can be accurate within its own rules and still produce a bad decision when it is compared with another report using a different date range, transaction basis, or settlement status.
The most dangerous forecasting mistake is treating the first number on a dashboard as final. Before increasing a purchase order, lowering a price, or declaring a channel unprofitable, reconcile what the number actually represents.
A real weekly example
In one recent operating review, Amazon showed $691.19 in sales across 30 units and 29 orders. Walmart showed $1,201.17 in Sellerboard, while primary daily data reconstructed approximately $1,233 for what appeared to be the comparable period. eBay showed $120.82, but the number was still preliminary.
The Walmart difference was about $31.83, or roughly 2.6% of the reconstructed total. That gap was small enough to ignore emotionally and large enough to corrupt a repeated weekly forecast. The correct conclusion was not that one system was wrong. It was: report the gap – do not infer.
We also treated some Walmart fees and refund activity as subject to a 14-to-28-day reconciliation window. A sales spike or profit estimate inside that window was useful operationally, but it was not yet a closed financial result.
Reconcile five definitions before forecasting
1. Period
Confirm the same start date, end date, time zone, and order cutoff. A rolling seven-day dashboard is not the same as a Monday-through-Sunday report.
2. Transaction basis
Determine whether the report uses order date, shipment date, delivery date, settlement date, or payout date. These views answer different questions.
3. Revenue definition
Separate item sales, shipping income, tax, discounts, refunds, reimbursements, and promotional credits. Do not compare gross merchandise value in one system with net product sales in another.
4. Unit definition
Orders are not units. Multi-quantity orders and multipacks can create a material difference between order count, units sold, and consumer units.
5. Cost maturity
Mark fees, advertising, refunds, storage, inbound freight, and reimbursements as final, estimated, or missing. Forecasting from preliminary revenue can be appropriate; calling it final profit is not.
Separate replenishable and finite inventory
In our current resale operation, much of the catalog is discontinued or finite merchandise. Some SKUs remain replenishable only while store inventory is available. That inventory cannot be forecast like a product with a stable manufacturer lead time.
For true replenishable products
Basic reorder point: Average daily unit sales x replenishment lead time + safety stock
Adjust the average for seasonality, promotions, price changes, stockouts, and channel-specific growth. Lead time should include ordering, supplier processing, transportation, receiving, prep, and marketplace check-in.
For discontinued or store-depleting products
Use a depletion forecast instead of a reorder forecast. Track units remaining, recent unit velocity, estimated days of supply, price floor, and the likelihood of finding additional inventory.
Estimated depletion date: Current sellable units / adjusted average daily unit sales
A finite SKU can be profitable even though it is not a long-term replenishable. The decision is whether to hold price, accelerate sell-through, or replace the revenue with a new product before supply ends.
Build a forecast confidence label
- High confidence: matched dates, matched units, reconciled refunds and fees, stable inventory, and sufficient sales history.
- Medium confidence: sales and units reconciled, but recent fees, returns, or advertising remain preliminary.
- Low confidence: mismatched periods, missing costs, stockout distortion, or only a few days of demand.
A confidence label prevents a precise-looking spreadsheet from implying more certainty than the inputs support.
The practical takeaway
A useful forecast does not begin with a complicated algorithm. It begins with matching periods, definitions, units, and cost maturity. Reconcile the gap, label uncertainty, and use the correct model for replenishable versus finite inventory. Clean inputs will outperform false precision.
Talk to Doty Distribution about marketplace expansion
Contact info@dotydistribution.com to discuss the right channel, inventory model, and operating plan for your brand.
