Skip to content
AnrilX

Finance

A figure that is not a field

Margin is not stored anywhere. It is what you charged minus what you paid, and those two numbers live in different modules with different keys, different grains and often different currencies. Finance spends its month-end assembling figures that do not exist as columns.

What gets in the way

The number you need is a calculation, not a column

Margin, variance against plan, effective discount rate, cost of a delay; none of these are fields. They are arithmetic across modules, which is why they end up in a spreadsheet maintained by one person.

Every figure needs a basis, and the basis is in a document

Your reporting standard says which allowances apply and which figures may circulate. That document is real, and it is not in the system computing the numbers, so compliance depends on whoever built the report remembering it.

By the time it is assembled, it is history

A close pack assembled from extracts is a photograph. Ask a follow-up question and the whole assembly has to happen again.

Where you start, and what you get

Questions you can ask on day one

  • “Which cost centres are over budget, and what drove it?”
  • “Where did margin go this quarter?”
  • “What is still open past terms?”
  • “How does this quarter's revenue compare with the same quarter last year?”
  • “Which customers account for the largest share of overdue value?”
  • “What did we bill against what we shipped last month?”

What you get back

  • Cross-module figures reconciled, with each side naming the entity it came from
  • Your reporting standard applied automatically, and the document named in the answer
  • The unadjusted figure marked, not quietly replaced
  • A live report you can page and sort rather than an extract that ages

Reconciliation, done as reconciliation

“What did we bill against what we shipped” is two figures from two different entity sets, and the useful part of the answer is whether they agree. AnrilX compiles each side as its own query, runs them through the same executor, and computes the verdict arithmetically, because a verdict a model asserts is indistinguishable from a verdict a model invented.

If one side cannot be read, it is excluded from the verdict and said so. It is never counted as agreement, which is the failure mode that makes a reconciliation worse than no reconciliation.

Your basis, applied without being asked

Upload your reporting standard once. AnrilX derives which figures that document governs (which entity sets, which kinds of measure) and applies it when those figures are reported, naming the document in the answer.

The rule matters as much as the application: a standard about order counts is structurally prevented from adjusting a currency figure. A wrong adjustment that arrives with a citation attached is the most dangerous output this kind of system can produce, because it looks checked.

Forecasting that can be argued with

Forecasts are trained on your own SAP history and backtested on whole periods held out of training. Three rules are constants in the product rather than intentions: a model that cannot be trusted shows no figures at all, error is stated as error, and a forecast is never presented as an actual.

One of these questions, answered

Which cost centres are over budget, and what drove it?

Seven cost centres, 8.4m SAR over plan year to date. Maintenance 4100 is most of it at 3.1m, and the driver is not maintenance: 2.6m of that is an internal activity allocation from plant 1000, posted monthly since March. The cost centre is carrying it rather than causing it.

Cost centre Actual · plan
4100 · Maintenance 11.9m · 8.8m SAR
2200 · Logistics 6.4m · 5.1m SAR
3300 · Quality 2.9m · 2.4m SAR
7 cost centres +8,412,770.15 SAR

Source: A_CostCenter and I_ActualPlanJournalEntryItem · actual and plan by CostCenter and CostElement · allocation lines separated by their origin object · 2026 to date

Assumed: budget means the plan version currently active, not last year's actuals. Allocations are reported against the receiving cost centre and the sending one is named, because charging the variance to whoever received it explains nothing.

Before a pilot

What it will not do for this role

The limits that bear on these questions specifically.

  • It answers from what is posted, so a question asked during close moves as the close does. Two runs an hour apart can legitimately disagree, which is why the answer carries the time it read at.

  • It does not know your management adjustments. A figure that lives in a spreadsheet between SAP and the board pack is outside it, and a variance it cannot see is one it will not explain.

  • It will not tell you a number is wrong. It shows what produced the figure so you can judge it; whether the posting behind it was right stays a question for you.

Questions people ask us

Can it apply our reporting basis automatically?

Yes. Upload the standard you already have and AnrilX works out which figures it governs, applies it when those are reported, and names the document in the answer. The unadjusted figure is shown alongside and marked as unadjusted.

How do we know a reconciliation is right?

Each side is compiled as its own query and run through the same executor, and the verdict is arithmetic rather than a model's claim. Each side also names the entity it came from, so both halves can be checked independently.

Is this safe to use during close?

It reads live SAP rather than an extract, so there is no second source of truth to reconcile against your system. Every figure carries its source, which is what makes it defensible in a review rather than something to re-derive.

Bring the question your reports cannot answer

Thirty minutes against a live SAP system we provide: no access to yours, nothing to set up. If it cannot answer, you find that out in half an hour rather than three months into a pilot.