Financial & operational intelligence
The expensive problems never show up as one big number.
They show up as a small gap on every line, a pricing exception that keeps repeating, or a manual workflow that consumes hours every week. Probatus finds them in your transaction data and business workflows, puts a figure on what they cost, and builds the smallest system that closes the gap.
- 01Find the problem
- 02Prove the impact
- 03Fix what's worth fixing
20-minute conversation · No ERP integration required
| Customer | SKU | Expected | Actual | Variance |
|---|---|---|---|---|
| ABC Supply | 48392 | $117.40 | $109.85 | -$7.55 |
| XYZ Electric | 39201 | $84.20 | $78.90 | -$5.30 |
| Norwood Bearing | 77104 | $231.50 | $231.50 | — |
| Coastal Fasteners | 12038 | $56.75 | $51.30 | -$5.45 |
| Delta Industrial | 60217 | $148.00 | $141.30 | -$6.70 |
Sample output — pricing variance by line. Constructed example, not a client engagement. See what we don’t claim.
Nobody decided to give away the margin.
Price gets set in a contract, adjusted by a rep, overridden at a branch, then quietly outdated by a supplier cost increase. Every step is defensible on its own.
The gap only appears when you line up what should have been charged against what was actually invoiced — one line at a time, across a year of orders.
A $5 gap on one transaction is noise. The same gap repeated across thousands of transactions, customers and SKUs is a pattern worth investigating.
Distributor Margin Audit
Your ERP records what happened. We investigate what should have happened.
We analyze historical transaction data to identify potential pricing and margin anomalies, quantify their potential impact, and trace every finding back to the underlying transactions.
- Scope
- Fixed scope, fixed-fee engagement
- Timeline
- Typically 2–3 weeks
- Access
- No ERP integration, installation or system access
How an audit runs
- Historical transaction export12–24 months of order and invoice lines. A flat file is fine.
- NormalizeReconcile formats, units, customer and item identifiers across the export.
- Reconstruct expected pricingRebuild what each line should have been priced at, from contract terms, price files, cost records and quantity breaks.
- Detect anomaliesIsolate the lines that don't reconcile with the price that should have applied.
- Quantify potential impactGroup by cause, count, and size each pattern so the largest surface first.
- Trace to transactionsAttach the underlying invoices, customers and items behind every finding.
- InvestigateYour team confirms or dismisses each finding against their own records.
Eight ways margin leaves without a decision being made.
Pricing leakage rarely appears as one obvious error. It appears as a set of small, individually defensible patterns spread across customers, SKUs and branches — each one invisible in a margin report that aggregates by product line.
Not every anomaly represents actual leakage — see what we don’t claim.
Signals we look for
- Supplier cost increases never passed through to price
- Contract and SPA pricing not applied at order entry
- Discounts granted below the customer's agreed floor
- Quantity breaks applied at the wrong tier
- The same item priced differently across branches
- Freight and handling charges not recovered
- Rebate-eligible purchases never claimed
- Margin drift on long-running repeat accounts
Example of how an audit might quantify findings.
The figures below are constructed to show the shape of an audit deliverable — how findings are grouped, counted and sized. They are not drawn from a real customer engagement.
Illustrative potential impact
$427,300
Illustrative example only. This is not a client result. Potential impact requires customer investigation and validation.
| Finding | Transactions | Potential impact |
|---|---|---|
| Cost increase never passed through to price | 421 | $126.2K |
| Contract price not applied at order entry | 842 | $173.4K |
| Discounts below the customer's agreed floor | 317 | $74.6K |
| Same item priced differently across branches | 198 | $44.5K |
Supporting transactions
| Customer | SKU | Expected | Actual | Variance | Why flagged |
|---|---|---|---|---|---|
| Harlow Pipe & Supply | 50214 | $342.10 | $318.40 | -$23.70 | Cost +7% in March, price file unchanged |
| Kestrel Industrial | 88732 | $96.75 | $96.75 | — | Within contract terms — not flagged |
| Bridgeton Electric | 22109 | $154.60 | $138.90 | -$15.70 | Contract price not applied at order entry |
| Anchor Fastening Co. | 67450 | $41.20 | $34.85 | -$6.35 | Discount 15% below agreed floor |
An anomaly is not automatically an error. Findings are a starting point for investigation, not a conclusion about recoverable money.
An anomaly is not automatically an error.
Legitimate commercial agreements and business decisions create unusual patterns all the time. A variance is a reason to look, not a conclusion about recoverable money.
Probatus identifies where investigation is likely to be valuable. Your team determines whether a finding represents actual leakage.
Things that can explain a flagged line
- A negotiated exception nobody recorded centrally
- Promotional or campaign pricing
- A customer agreement with terms outside the standard file
- Special commercial terms for a strategic account
- A data quality issue in the export itself
- A deliberate business decision made for good reasons
We do not claim that every anomaly is recoverable revenue, and we do not report findings as if they were.
Evidence first.
Most analytics stop at “something looks unusual.” That is where the work actually starts.
A finding is only useful if somebody can act on it. Every pattern we report answers six questions before it reaches you — and if it cannot answer them, it does not go in the report.
- What happened?
- The specific pattern, described in business terms.
- Where?
- Which customers, items, branches and periods it touches.
- How often?
- The number of transactions involved, not a percentage.
- What is the potential impact?
- Sized in money at your volumes, and ranked against every other finding.
- What evidence supports it?
- The underlying invoice lines, so you can look each one up in your own system.
- What should someone investigate next?
- The control that appears to have failed, and who is best placed to confirm it.
Find → Prove → Fix
Most software projects start with a decision to build and work backwards to a justification. We work in the other direction.
- 01
Find
Identify anomalies, bottlenecks or expensive manual work — in transaction data, or in the process as it actually runs today.
- 02
Prove
Quantify the impact and trace the finding back to evidence: how often it happens, what it costs, and the transactions or steps behind it.
- 03
Fix
Implement the smallest system, workflow or control that addresses it — only when the number says it is worth fixing.
Don’t automate what you haven’t understood.
Automating a process nobody examined faithfully reproduces it, including the parts that should not have existed. If we cannot tell you what something costs today, we will not quote to build a system around it.
From finding the problem to fixing it.
An audit ends with a ranked list of things that are costing money. What happens next depends entirely on what the list says.
Sometimes the answer is a pricing control. Sometimes it is a better workflow. Sometimes it is automation. Sometimes it is a small internal application. Sometimes the problem is the gap between systems.
When the answer is software, we build it — not a platform, and not a replacement for your ERP.
Operational Systems
When your existing software doesn’t handle the workflow.
- RFQ QuotationEnquiries arriving as PDFs and email, re-keyed and priced by hand. See how we build this →
- PDF / email Structured dataInvoices, purchase orders and price lists read by a person and typed into a system.
- Excel ReportingThe same workbook rebuilt every week from three exports.
- ERP Internal workflowThe step your ERP has no opinion about, handled in a spreadsheet beside it.
- Multiple systems One processWork that crosses four tools, where every crossing is done by hand.
We build these around the systems you already run. We do not do ERP implementations, websites, mobile apps, or projects where the value is a matter of opinion — if we cannot estimate what the current process costs, we are not the right people for it.
What we can show you, and what it is.
Probatus is early. Rather than blur the line between what we have built, what we have tested and what a customer has paid for, we label each one and keep them separate.
Client Work
Nothing published yet.Real customer engagements, published with permission.
When there is client work to show, it will be named or described precisely enough to be verifiable.
Demonstrations
Margin Audit demo.Working software shown on synthetic or sample data.
Labelled as synthetic wherever it appears, including in every screenshot and figure.
View →Labs
3 experiments in progress.Experiments and prototypes built to answer a specific technical question.
Published with the approach, the architecture and the failure modes — results when they have been measured.
View →Research
Nothing published yet.Analysis and investigations we conducted ourselves.
When published, sample size and method will be stated, and facts kept separate from hypotheses.
We do not publish customer logos, testimonials, savings figures or case studies we do not have. Synthetic examples are never labelled as case studies.
Have a problem worth investigating?
Tell us how the workflow works today, where it hurts, and what you have already tried. We will tell you whether it is the kind of problem worth measuring, and what that would take.
20-minute conversation · No ERP integration required