Skip to content

Why

Why tells you what is happening and what is causing it, for the selected run. It answers the run’s question first, then shows every check FACE ran on the rows the run read, what your records and public pages say about each counterparty, the market & diligence analyses, and the business rules behind it all. Open it with Why, under Simulate in the rail (/simulate?tab=knowledge).

Every figure on Why is measured on the rows the selected run read. A figure that could not be measured says so and why. It is never shown as zero.

The answer

The page opens with one sentence that answers the run’s question, from the check that answers it. A stockout question, for example, is answered by the stock checks. The answer states the money the failing rows put at stake ({amount} at stake), the cause in plain words, and the worst rows first (Worst first). Two ways on follow: See options opens Options on that decision, and Act in Actionable Twins opens the decisions.

When no check answers the question directly, the answer says No check this run measured answers the question directly. The largest thing needing attention: and names the thing with the most money on it: a failing check, the deductions open more than 60 days, or the claims whose window closes within 14 days. Nothing on this run needs attention. means the run measured nothing that needs saying.

The money and the checks

Under the answer is the money line, for the selected run. It is on Why only, not on Ask (owner, 2026-11-02: the money tiles “are under Why page not Ask”). It has seven tiles:

TileWhat it countsOpens
At riskShipments past their promise, or with a temperature excursion no claim namesWhy
Owed to youOverbilling and duplicate invoices, plus every open claim at its claimed amountWhy
Claim windows closing in 14 daysOf that, what is claimed on open claims whose window closes within 14 daysWhy
Waiting for your approvalThe run’s decisions not yet decided, and the money they stateActionable Twins
Recovered this quarterWhat your team recorded as recoveredActionable Twins
Shipments at riskThe value of the shipments at risk, with their countWhy
OTIF finesWhat the retailers’ on-time-in-full terms fine(does not open a page)

A board tile shows the change since the last successful run, for example ▲ {amount} since the last run, or Same as the last run. A figure that was not read or not stated says Not measured with its reason. It is never shown as zero. Decisions that state no amount are counted apart ({count} state no amount).

The persona picked under Who this run is for on Ask decides the order (its Reads first, under Settings › Personas): one that reads what is owed first leads with Owed to you and Recovered this quarter; one that reads what is at risk first leads with At risk and Claim windows closing in 14 days. The rest follow in the order of the table. With the Generalist view, At risk leads. Only the order changes, never a figure.

After the money line, a line counts the checks, {total} checks: {failing} failing, {holding} holding, {notMeasured} not measured.

The operations board follows, with four figures measured on the run’s rows:

FigureDetail
On-time deliveryHow many of the delivered shipments were on time, and how many are overdue
Open claimsHow many close within 14 days, and how many were filed after the window
Invoices over contractHow many of the invoices checked bill above the contract, and by how much
Stock below reorder pointHow many positions are below their reorder point, and how many below safety stock

Each figure states its basis, for example over every row read, or Sample figure, not a total when the run read a sample. Overdue and days left are measured as of the data’s latest row read, not the time of the run.

What FACE checked

What FACE checked lists every check run on the rows the selected run read, each once: failing checks first, the most money then the most rows. Holding ({count}) and Not measured ({count}) are folded below them.

A failing check says what does not hold, for example Deliveries late against their promise, and how many rows: {failing} of {checked} rows checked do not hold, with the money its rows put at risk when they state any. Open it to see its rows. A check that held only in a sample says a sample, not a pass.

The checks cover deliveries and promises, claims and their windows, deductions, invoices against the contract, rate contracts, stock, bills of lading and proofs of delivery, customs, insurance and cover, and deadlines and signatures. Each check carries the Twins category it belongs to.

For: narrows the list to an audience: Everyone, Paralegals (bills of lading, deadlines, signatures and the claim file’s documents) or Brokers & underwriters (customs and cover). A deadline here is an input to a person, not legal advice.

Above the list, anything that blocks a decision is said first: rules that disagree with what your systems do, rules that could not be read, and values read from the run’s documents that fail a check (Correct them where the run shows what it was given.).

Counterparty reviews

Counterparty reviews covers the carriers, shippers, buyers, sellers, distribution centres and logistics providers your data names. For each it shows what your own records say (shipments, claims, invoices over contract) and what public web pages say, with each red flag citing its page.

  • No cited red flag in the pages found. That clears no one.
  • The searches returned no pages. That clears no one.
  • Public reviews not searched: the run’s rows state no name for this party, and FACE does not search the web by an internal id.

This is not a background check.

Market & diligence

Market & diligence holds three analyses as cards: Run when you ask, over your own records and public sources; not part of the selected run. They were a page of their own until 2026-10-31 (owner); an old /market_diligence link opens the matching card here.

CardWhat it gives
Customer analysisEach account’s orders, revenue, cost to serve, margin, OTIF, claims and days to pay, from your own records.
Counterparty background checkA business checked by its name or identifiers (for example USDOT number, MC number, LEI (Legal Entity Identifier)) against public registries and sanctions lists, with its history in your records. Possible sanctions matches need human review.
Freight marketYour lanes’ rates against their own baseline and the same window last year, beside public diesel and freight index figures.

Closed, each card says where it stands: Not run yet, Running, Could not run, or what it found. Open one to choose what to analyse, then select Analyse (or Run check for the counterparty check). Every public source is listed with its state (live, cached, stale or not checked), and a figure the server did not send reads not known, never 0. Recommended actions (drafts for a person to decide) go to Actionable Twins with Send for approval, where a named person approves or rejects them.

What this run was given

What this run was given opens the run’s own page, where the files and captures attached to it are listed with how each was read. See Open a past run.

How FACE knows this

How FACE knows this (The rows behind each figure, the rules FACE read and their policies) is folded at the foot of the page. It holds:

  • The rows behind the figures: the board’s views over the rows it read, per carrier, claim, invoice and stock position;
  • the rules FACE read from your systems, counted apart from the checks: RULES HOLDING IN THE DATA and RULES BROKEN;
  • Rules your operations run on, described below.

Rules your operations run on

Rules your operations run on (Your business rules, checked against what your systems actually do and against the policy documents you add) lists the business rules FACE found and the ones you added. All domains ({count}) and the domain pills narrow the list. These are your business’s rules, not FACE’s own guard rules, which are on Admin.

Every rule carries one verdict, each with its own glyph and word, so it never depends on colour alone:

VerdictMeaning
AlignedAligned with policy.
DriftDrifted: needs reconciliation.
ShadowUndocumented.
MissingMissing: needs reconciliation.
UnratedNo verdict was returned yet.

Set a rule’s enforcement

Open the policy editor

Select the tune button on a rule (Open the policy editor). The Enforcement policy drawer opens below it.

Decide the enforcement

Choose an Enforcement action: Audit only, Review required or Block. Optionally fill in Tolerance (in the rule’s own units), Policy reference (the clause you cite) and Rationale (why, for the reviewer).

Save policy

Save policy stays disabled until you choose an action. The drawer records who set it and when. The policy was not saved keeps your entries so you can try again. Discard drops your changes.

Saving a rule’s enforcement validates it. A validated rule goes to the simulations on Options by itself, and the twins are re-run for your instance. Include this rule in Simulate takes a rule in or out of that selection.

The drawer also holds the Reconcile inspector: a read-only comparison of Actual — implementation against Intended — policy. Nothing to compare yet means FACE captured neither for the rule.

In an open rule’s drawer, Edit & recheck opens the rule (Rule Name, Logic (Plain English), then Save Rule), and Ignore sets it aside. Unsaved marks a rule you edited or added here until you save a policy on it. Under the list, Create New Business Rule… adds a rule in plain English.

RPCs behind this page

  • FetchService.GetOperationsBoard, for the money, the board and the checks.
  • RulesService.AnalyzeCode, UpdateRulePolicy.
  • PostureService.GetHealthScore.
  • MarketDiligenceService.AnalyzeCustomers, CheckCounterparty, AnalyzeFreightMarket, QueueRecommendedAction, for the market & diligence cards.

See the API reference.