How to Spot-Check a Reorder Suggestion in Thirty Seconds

Read it as four lines, not one number

Every suggestion in the Smart Buying Table is computed from your sales velocity, lead times, and incoming stock — then rounded to your case sizes and capped at your maximum stock levels. That's not a black box; it's an arithmetic sentence with four lines and a total. Which means the useful question is never "is 84 right?" It's "are these four lines right?"

So the audit is a fixed reading order, not a gut check: velocity → cover period → incoming → on hand → suggested. You never evaluate the total directly. You evaluate four claims about your business, each of which you already know something about, and the total inherits whatever confidence they earn. Four claims you can check against memory take seconds each.

suggested order = demand over the cover period − incoming from open POs − on hand

Four lines, three terms: the first two multiply — velocity across the cover period — to give the demand you're covering. And the buffer isn't a fourth term bolted on the end; it lives inside the cover period, expressed as days of cover rather than as a separate quantity.

Lines one and two: is this the demand and the window I'd have picked?

Velocity claims this is how fast the product sells. Check it against what you actually remember of the last few weeks — the season you're in, a promotion that ran, a channel that turned on. The usual suspects are two skews. A stockout inside the lookback window depresses velocity, because days you couldn't sell look identical to days nobody wanted it; Assisty's forecasting lets you exclude days you were out of stock, so the forecast reflects real demand rather than your empty shelf. A one-off bulk or wholesale order does the opposite, inflating an average that no ordinary week will reproduce.

Cover period claims this is how long the order has to last. In periodic review, that window is lead time plus the review period — so check it against two things you know: how long the supplier really takes lately, and how often you actually place orders. Lead time is the classic stale parameter; it's typically entered once at setup and rarely revisited, so it quietly encodes the supplier's performance from whenever someone last typed it. In Assisty each supplier's Lead Time, Reorder Cycle, Min Stock Cover Days, and order minimums and maximums are set once and the buying table's suggested quantities use them automatically.

Lines three and four: is this the stock I think I have?

These two are the most likely to be wrong, and they go wrong for non-mathematical reasons. The arithmetic is subtraction. The failure is bookkeeping.

Incoming from open POs claims this much is already on its way. Check it against the orders you actually placed. A shipment that exists in an email thread but not in the system is invisible to the subtraction, and you buy it twice.

On hand claims this much is in the building. It is a record, not a fact — and the gap between those is well documented: DeHoratius & Raman (2004) found discrepancies in 65% of nearly 370,000 inventory records at a large retail chain, with 15% off by eight units or more. That's brick-and-mortar research from a different channel, so treat it as directional rather than a DTC statistic. But the direction is the point: shrink, misplacement, a mis-scanned receipt, or a 3PL count that drifted all produce a confident number attached to stock nobody can pick. You can't verify this from a screen — you verify it by knowing which SKUs your counts have been unreliable on, and treating those rows as suspect.

The four lines and what to check each against

When a line disagrees, fix the input — never the total

The tempting move is to overwrite the suggested quantity. Sometimes that's correct: for a genuine one-off, Assisty lets you tell Atlas how reality differs from the data and the whole table recomputes, with a was→now chip and per-row undo on every change, and nothing committed until you create the order. Cheap to try, cheap to reverse.

But an edited total is consumed once. It fixes this week's order and nothing else. A corrected parameter changes every subsequent run — so if the disagreement is a standing fact rather than a one-off, route it upstream: a lead time that genuinely changed gets corrected where it's stored; a stockout that distorted the history gets excluded from the forecast; a shipment nobody recorded gets recorded. Same thirty seconds of work; one is a patch, the other a repair.

The thirty-second pass — and the rows that deserve longer

The pass, in order: does the velocity match the weeks I remember, allowing for the season? Does the cover period match the lead time I'd quote from memory plus my real ordering rhythm? Does incoming match the POs I know are open? Is on hand a number I'd bet on for this SKU? Four yeses and you move on: the total is only as good as its lines, and you've checked them.

Thirty seconds is a screen, not a review. Its job is to catch loud disagreements, and most rows pass because most rows are ordinary. Some don't get to be ordinary: high-value rows where a wrong quantity is real money, long-lead items where the correction arrives with the container, anything in its first season with thin history behind it, and — the most reliable trigger of all — any row that surprised you. Assisty's findings are ranked by money at stake, so the rows carrying the biggest dollar exposure float to the top on their own. The rest of that list doesn't sort itself: a long-lead, first-season, or merely surprising row can sit anywhere in the table, and pulling those out is still your judgment. Then you select the rows you're ready to buy and create the purchase orders, with Atlas splitting the selection into one PO per supplier.