The engine's error was 2.2× smaller than yours

Order accuracy, head to head

165 orders with enough sales evidence to compare. These are invented figures chosen to show the report’s layout; the demo site’s own nine-week scoreboard is the more conservative illustration of the same comparison.

• Engine closer on 86. You on 34. Level on 45.
• Warehouse top-ups are included: a suggested order that was too low still counts as an error.
• Left out: empty shelves, expired stock, true stockouts.
• “Gap” is the difference from the order needed to cover sales. Lower is better. Grouped by product, biggest total error first.

Where its smaller order was closer

On these lines the engine asked for less than you did, and its number sat nearer what the shelf needed. Stock saved is the difference between the two orders, not a claim that every one of those units was waste.

The biggest misses

These suggestions were too low and less accurate than the actual orders.

• Worst: Whole milk family pack. It said 23, 64 sold, 16 short.
• These are the first lines to improve.
• “Engine short by” is the extra stock needed to cover recorded sales.

Stock left over: checks and limits

Only supplier orders are compared. Warehouse and cash-and-carry purchases are excluded, so a shelf that was topped up by hand still counts the supplier order as short.

• 0 of 5 skip-calls would have run dry: on a skip-call the engine asked for nothing at all.
• On the 160 orders the engine sized itself, it ran 167 units heavier than you net (41 of those still would not have covered).
• Split by shelf life: on chilled it ran 148 units heavier; on ambient 19 units heavier.

Did it keep the shelves covered?

Follow the engine and the shelf stays covered on 75% of these 165 orders.

• Your own orders covered 58% of them, so on cover alone the engine is ahead of your own. Its lead is in sizing, not in cover.
• Read this next to the stock saved above, not on its own. Where a shelf ran empty we cannot count the shoppers who walked. The real picture is both together: less stock sitting, shelves still covered.

Would it cut your warehouse and cash-and-carry runs?

Yes, on this sample.

• To meet sales, your buying needed 677 units fetched from outside these orders. The engine, 249.
• On 60 orders it bought bigger where it mattered, saving 590 units of fetching.
• On 30 orders it skipped or ordered light, costing 162 units more.
• Net: 428 fewer units to fetch. We track it weekly.