Sign in
Set up for Supply chain

AI for insights and analytics

AI analytics for supply chain

A site reports forecast accuracy of 82 percent while thirty SKUs are forecast eighteen percent high every month and hold most of the site's aged stock. AI analytics reads the forecast, order, receipt and inventory files, and answers the supply chain director: where is bias hiding, which suppliers are deteriorating, and what is causing misses.

Ask it in your own words

These are the questions supply chain teams ask. Each one maps to a measure our tools compute from your files. The AI model chooses the measure and explains the result; the arithmetic is done by our code, and every total is checked.

The questionThe measure behind itWhat comes back
Where is the forecast biased, not just inaccurate?Forecast biasSigned error by SKU and site over the trailing months, with the items consistently high or low.
Which suppliers are deteriorating against their own record?Supplier deterioration watchOTIF and lead time against each supplier's own baseline, with the ones sliding.
What stock is ageing, and which forecasts caused it?Inventory ageingStock over 180 days by SKU and site, joined to the forecast bias on the same items.
Where are we exposed to a single source?Single-source exposureSpend and volume on items with one qualified supplier, by category and site.
Why did we miss customer OTIF?Cause of service missesMisses under the customer's rule by cause: stock, transport, order error, with the count and value.
What are expedites costing, and why?Expedite cost by causeExpedite spend by cause and site, against the forecast bias and supplier misses behind it.

A conversation, with the figures

Every figure below was computed by a tool from the rows and checked before it was shown. The lines sum; the percentages match; each line opens to the rows that make it.

Where is aged stock coming from?

Stock aged over 180 days totals $2.2m across 3 sites, and most of it sits on items that were forecast high for months. Northgate plant holds $1.1m, 50% of it.

Northgate plant$1.1m50% of the total
Southport depot$700k32% of the total
Eastfield plant$400k18% of the total
The 3 lines sum to$2.2m0 unexplained

Which Northgate items?

3 item groups account for $640k of the $1.1m aged at Northgate plant, 58% of it. Each was forecast an average of 18 percent high every month at this site and nowhere else.

Fastener range$380k30 SKUs; bias +18% for 9 months
Seal kits$190kbias +14%; two suppliers over-delivering
Coatings$70kbias +11%; shelf-life risk
These 3 are$640k58% of Northgate plant

Who fixes the forecast?

The site planner, this cycle, on the thirty items with the bias figures in hand. Purchasing pauses replenishment on the aged items, and the director reviews bias by site monthly beside accuracy.

Northgate plannerCorrect the forecast on the 30 biased SKUsThis cycle
PurchasingPause orders on items with more than 180 days of coverThis week
Supply chain directorReport forecast bias beside accuracy by siteMonthly
Where is the forecast biased, not just inaccurate?Which suppliers are deteriorating against their own record?What stock is ageing, and which forecasts caused it?Where are we exposed to a single source?

The measures behind the answers

Each measure has one formula, one source and one meaning. They are computed per site and supplier and in total, and every one carries an identity that must hold before it is shown.

MeasureFormulaFromWhat it tells you
Supplier OTIF against own baselineReceipts on time and in full ÷ receipts due, per supplier and site, against the supplier's trailing 12 monthsPurchase order and receipt filesSuppliers that have changed
Lead time variabilityInterquartile range, and 90th percentile, of actual lead time, per supplier and sitePurchase order and receipt datesThe number behind the safety stock
Customer OTIF, customer's ruleOrders on time and in full by the customer's definition ÷ orders dueOrder and shipment files; customer termsThe score the customer sees
Forecast biasMean of (forecast − actual) ÷ actual, by SKU and siteForecast snapshots; actualsPlans that are always high or low in one place
Inventory ageingStock value with no movement in 90, 180, 365 days, by siteInventory file; movement historyCash in stock that is not moving
Single-source exposureSpend, and parts, with one qualified supplier ÷ total, by siteSupplier master; purchase ordersWhere one failure stops a line
Days of cover against targetStock ÷ average daily demand, by SKU class, against targetInventory; demandToo much and too little, at once
Expedite cost by causePremium freight and expedite fees, by cause: supplier late, forecast miss, order changeFreight ledger; reason codesWhat unreliability costs
Supplier deterioration watchSuppliers worse than own baseline by more than the margin for two consecutive months, ranked by spendReceipts; quality fileThe call list for procurement
Cause of service missesCustomer OTIF misses by cause: stock-out, pick, carrier, order data, customer dateShipment and warehouse filesWho owns each fix

Each measure is worked through, with the export it comes from and what to drop, in KPIs for supply chain teams.

What the AI model does, and what our tools do

The AI model chooses the measure

From your question and the measures declared for supply chain, the model picks the one that answers it, and the period and comparison the question implies.

Our tools do the arithmetic

Deterministic code reads the rows, computes the measure, and checks the identities below. The same question on the same data gives the same answer, every time.

The AI model explains, and cites

The AI model writes the sentence around the result, naming the site or account behind it. It states no figure that is not in the result, and every figure links to its rows.

The identities that must hold

ReceiptsReceipts due = on time and in full + late only + short only + both
Customer OTIFOrders = OTIF + late only + short only + both
InventoryOpening stock + receipts − issues ± adjustments = closing stock; aged bands sum to total
ExpediteExpedite cost by cause sums to the freight ledger premium lines

What it reads

The exports supply chain teams already produce. Column names are mapped once and the mapping is reused. A file is the way in; scheduled delivery and connections to your systems come with the plan, and every source is listed here.

  • Purchase order and receipt files
  • Purchase order and receipt dates
  • Order and shipment files
  • Customer terms
  • Forecast snapshots
  • Actuals
  • Inventory file
  • Movement history
  • Supplier master
  • Purchase orders
  • Demand
  • Freight ledger
  • Reason codes
  • Receipts
  • Quality file
  • Shipment and warehouse files

Who owns each answer

An answer is a list with an owner and a cadence, or it is a chart nobody works.

MeasuresOwnerCadence
Supplier deterioration watch; supplier OTIFProcurement and supplier qualityMonthly
Customer OTIF; cause of misses; expedite costOperations director; site headsWeekly to monthly
Forecast bias; days of cover; ageingDemand and supply planningMonthly
Lead time variability; single-source exposureChief operating officerQuarterly

Questions supply chain teams ask about AI analytics

What is the difference between accuracy and bias?

Accuracy averages the size of the errors; bias keeps their sign. A site can be 82 percent accurate with thirty items forecast high every month and others low, because the errors cancel. Bias by SKU finds the thirty; accuracy never will. The tool reports both.

How is a supplier's own baseline set?

From its own history in the receipt file: OTIF and lead time over the trailing period. Deterioration is a fall against that, so a supplier that was always at 80 percent is not flagged while one that fell from 96 to 88 is. Fixed thresholds miss the second and punish the first.

Whose OTIF rule does it use?

The customer's, when the contract file carries it, because that is the one the customer measures you on. The internal rule is shown beside it. Misses are then listed by cause from the order and shipment files, so the meeting is about causes rather than the definition.

What files does a supply chain team need?

Forecast and actual demand by SKU and site, purchase orders and receipts, inventory by age, and shipments against customer orders. The supplier master with qualified sources adds single-source exposure. Each is an ERP export you already run.

See it on your data

Bring a few thousand rows. The data map opens next, every column mapped once, and the first question is answered in minutes. Free, in your browser, no account.