An inventory number on a dashboard feels like a fact about the present: “we have 420 units.” In practice it is a statement about a specific moment, drawn from a report that was produced some time after that moment and read some time after that. Most inventory mistakes come from forgetting one of those gaps, or from treating different kinds of stock as one pile. This guide covers the vocabulary, the timing, and a simple coverage estimate with its assumptions in plain view.
Snapshots and flows
Inventory data comes in two shapes.
A snapshot is a count at a point in time: how many units sat in each state at, say, 06:00 on Tuesday. It is like a bank balance.
A flow is a movement during a period: units received, shipped to customers, returned, removed, transferred or adjusted as lost, found or damaged. Flows are like the transactions on a bank statement.
The two connect only when every flow is recorded:
closing stock = opening stock + receipts + sellable returns − shipments − removals ± adjustments
If any flow is missing, the equation stops balancing, and the honest thing to show is the unexplained difference rather than a number forced to fit.
Four quantities that are easy to confuse
- Available: units that can be sold and fulfilled now.
- Inbound: units in shipments that have been created, sent or are being received. They are not sellable until they are received.
- Reserved: units held for pending customer orders, transfers between fulfillment centers, or processing. They are not part of available stock, and some will be released back.
- Unfulfillable: units that cannot be sold in their current condition, for example damaged, defective or expired items.
| Quantity (fictional SKU, one snapshot) | Units | Sellable now? |
|---|---|---|
| Available | 420 | Yes |
| Inbound | 600 | No, not until received |
| Reserved | 35 | No, held for orders or processing |
| Unfulfillable | 12 | No, cannot be sold as is |
| Sum of all four | 1,067 | Not a useful “stock” figure on its own |
The 1,067 total looks reassuring, but a shopper can buy only from the 420. Inbound stock might arrive in two days or three weeks.
Why the data is always a little old
Many inventory reports are produced asynchronously. A report is requested, the source system processes it, and the result is downloaded when it is ready. Different reports refresh on different schedules, and two reports describing “today” can reflect different moments. Report availability also varies by account, marketplace and report type.
That is why every inventory view should carry an “as of” time, the moment the data describes, separately from when it was retrieved. A report describing 06:00 that was generated at 09:00 and viewed at 15:00 is already nine hours behind the shelf. On a fast-selling product, nine hours matters.
Estimated days of cover
Days of cover is a quick way to ask “how long would current stock last at the recent pace?”
estimated days of cover = available units ÷ average daily units ordered
KEYXE’s demo uses the average over the last 30 days of the selected period. With 420 available and an average of 14 units a day, the estimate is 30 days.
| Scenario (fictional) | Available | Avg daily units | Estimated days of cover |
|---|---|---|---|
| Steady month | 420 | 14.0 | 30 days |
| Promotion doubles demand | 420 | 28.0 | 15 days |
| 10 out-of-stock days in the window | 420 | 9.3 | 45 days (overstated) |
| No recent sales | 420 | 0 | N/A |
The assumptions behind the estimate
- Recent demand continues at the same average rate.
- Inbound shipments are not counted until received.
- Seasonality, promotions and price changes are not modeled.
- Zero recent sales gives N/A, not “infinite”.
The third row shows a quiet trap. If the product was out of stock for 10 of the last 30 days, it sold nothing on those days, the 30-day average falls, and the estimate inflates. One fix is to average over in-stock days only; whichever choice you make, state it next to the number.
Why past stock cannot be rebuilt from sales alone
It is tempting to work backward: 420 units today plus 300 sold this month means 720 units a month ago. That works only if nothing except sales moved the stock. Suppose, in this fictional example, a 500-unit shipment was received mid-month, 15 units came back as sellable returns and 8 were written off as damaged. Then 420 = opening + 500 + 15 − 300 − 8, so the opening stock was 213, not 720. Sales history alone cannot reveal receipts, removals, returns or adjustments. Where history is incomplete, a gap is more truthful than a guess.
Cautions
- Read the timestamp first. A number without an “as of” time cannot be judged.
- Gaps are not zeros. A missing day of data is unknown, not a day with zero stock.
- Partial periods distort averages. A period that ends today includes incomplete data.
- Estimates are not forecasts. Days of cover helps decide where to look; replenishment decisions still need lead times, supplier information and judgment.
- Expect revisions. Reserved units return to available, inbound shipments arrive late, and reports are sometimes reissued.
Try it
Open the inventory workspace of the synthetic demo to see available, inbound and reserved units, a stock movement bridge with its unexplained difference, and days-of-cover estimates. The Data Foundation page explains how “as of” times and lineage are designed to travel with each figure. Continue with Seller Sales vs Ad-Attributed Sales or How to Evaluate an AI Recommendation, and look up terms in the glossary.