Anything can produce a reorder quantity for an item that has moved twice. The number will be confident, arbitrary, and indistinguishable on screen from one built on two years of history — which is how a planner learns to stop trusting the whole column. Below a floor you set, this engine declines to answer, and says it is declining.
Fourteen days is a default, not a law. It is the horizon over which the engine asks whether an item runs out, and it is the first thing most workspaces change.
Running out and being overstocked look like opposites and are frequently the same workspace at the same time, on different items. The card is arranged so both are visible at once, because a business that fixes only the first ends up with a warehouse full of the second.
Not "below reorder level" — that is the low-stock signal and it is a different question. This asks whether an item runs out within the next fourteen days at its recent rate of movement, given its lead time and the safety stock you want kept. An item comfortably above its reorder point and moving fast is a stockout; an item below reorder that nobody has touched in a year is not.
Quantity against the level you set for that item. It is kept as its own signal rather than folded into the risk number because the two disagree usefully, and a planner needs to be able to see which of them is firing.
Items holding value and not moving. This is the signal that pays for the module in most warehouses, because dead stock is invisible by nature — nothing goes wrong, no alert fires in any other system, and the capital sits there until somebody counts it.
Cover far beyond what the movement rate justifies. Distinct from dead stock: an overstocked item is selling, just far more slowly than it was bought. It is a purchasing conversation rather than a write-off conversation.
Items whose selling price and cost have drifted into a relationship nobody would have agreed to deliberately. Cost moves quietly — a supplier price rise, a landed cost, an exchange rate — and price does not move with it unless somebody notices.
An item with fewer than three demand observations is not forecast at all. It is not given a cautious estimate or a wide range — it is left out, because a placeholder that looks like a forecast is worse than a gap that looks like a gap. The floor is a setting, so a workspace with dense data can lower it.
The forecast signal ships reading customer demand — what your customers actually took — rather than every movement type in the warehouse. Internal consumption, transfers between your own locations and adjustments are movements, but they are not demand, and treating them as such inflates a forecast in exactly the workspaces that move stock around most. The sources are a setting, so a workspace whose internal issues genuinely are demand can say so.
A single enormous order distorts an average for months. Observations beyond a multiple of the item's own typical demand — three times, by default — are capped rather than taken at face value. The one-off is still visible in your movement history; it just does not get to set your reorder quantity for the next quarter on its own.
Seven days of lead time and three days of safety stock ship as defaults and are yours to set. They matter more than the forecasting arithmetic: an item that takes six weeks to arrive and one that arrives tomorrow need completely different warnings from identical demand, and no amount of cleverness in the model substitutes for knowing which is which.
A sensitivity setting sits over all of it — balanced by default — for workspaces that would rather be warned early and occasionally wrongly, or late and rarely wrongly. That is a business preference about the cost of a stockout versus the cost of holding, and it is not a decision a vendor should be making on your behalf.
This is the single most important thing to understand about Inventory Insights, and it is the thing most “AI analytics” pages are careful not to say. Every figure on the page — every count, every total, every flagged record — is computed by an ordinary database query against your own data, with thresholds you control. A language model is used for exactly one job: turning that computed result into a paragraph a person can read quickly.
Counts, totals, ages and comparisons, run against your workspace under the same tenant scoping as every other screen. This step is deterministic: run it twice on unchanged data and you get the same answer twice. If the engine says nine invoices are overdue, nine rows matched.
What makes a request stale, a cost a spike, an entry large or a balance overdue is a number in your settings, not a judgement in a model. The defaults are listed below. Change one and the flags change with it, immediately and predictably.
Only then is a short plain-language summary generated from those facts. It is written on a background job so the page never waits for it, and it fails soft: if no provider is configured or the request fails, the summary is simply not there and every number on the page is unaffected.
The summary is stored with a fingerprint of the exact figures it was generated from. When those figures move, the fingerprint no longer matches and the summary is regenerated rather than served stale. This closes the failure that makes cached AI commentary dangerous: a confident paragraph describing last week, sitting directly above this week's numbers, with nothing on the screen indicating the two disagree.
Results are cached per workspace for a window you set — fifteen minutes by default — so the page is fast without being out of date, and a refresh action clears it when you want the current picture immediately.
These are the actual settings behind the flags on this page. They are per workspace, they are yours to change, and the defaults are chosen to be defensible rather than dramatic — an engine tuned to flag everything on day one is an engine nobody reads by day three.
| Setting | Ships as | What it decides |
|---|---|---|
| inventory_stockout_days | 14 | The horizon the stockout question is asked over. Shorten it for fast-moving consumables; lengthen it where lead times are long. |
| inventory_default_lead_time_days | 7 | How long replenishment takes when an item does not carry its own lead time. Usually the first setting worth correcting. |
| inventory_default_safety_stock_days | 3 | The buffer you want left when stock is planned to run down, expressed in days of cover rather than units. |
| inventory_min_demand_points | 3 | How many demand observations an item needs before it is forecast at all. Below this the engine declines rather than estimating. |
| inventory_outlier_cap_multiplier | 3.0 | How far above an item's own typical demand a single observation is allowed to count, so one freak order does not set the next quarter. |
| inventory_forecast_sensitivity | balanced | Whether to warn early and sometimes wrongly, or late and rarely wrongly — a judgement about the cost of a stockout against the cost of holding. |
| inventory_forecast_sources | customer_demand | Which movements count as demand. Ships as customer demand only, so internal issues and inter-location transfers do not inflate the signal. |
Every flag carries the way to act on it. A finding is not a number on a dashboard you then go hunting for — it is rendered with a label and a link to the screen where the work is done. An insight you cannot act on from where you are reading it is a report, and you already have reports.
What AWRA OpsHub does today
More we can add to your workspace
Where we point you to a specialist
Seasonality and automatic reorder proposals are the two most asked for, and both sit on top of a forecast layer, a settings layer and a purchase-request workflow that already exist. Tell us which one your planners would use.
Cash flow split three ways, entries that are large for you, and anomalies.
Revenue momentum, overdue money with no hidden grace period, days to pay.
Late orders, stale requests, cost spikes and how concentrated your supply is.
All four modules on one card, and the workspace-level narrative across them.
The module underneath — items, stock, movements and locations.
Acting on the overstock and stockout signals across locations.
Help Center
Run inventory, procurement, assets, sales, and field work with approved AWRA guidance for setup, migration, integrations, security, pricing, and support.
