Home · Solutions · Supply chain

Solution · Supply chain

Every carrier invoice line re-priced against your contract before it is paid

Freight invoice audit against rate cards and shipments

Shipments from the TMS, rate cards on SharePoint and carrier invoices in any format are matched line by line, so overbilling becomes a dispute before payment, not a discovery a year later.

DepartmentalMicrosoft TeamsHuman in the loopAI where it earns its place
14,000shipments a month are billed to this illustrative retailer on 220 carrier invoices. Logistics checks a sample of the lines; the rest is paid as invoiced.

Executive summary

Challenge

Carrier invoices are checked on a sample and paid in full; the overbilled lines are the ones nobody opened.

What changes

Mientha designs and delivers a freight audit that treats the rate card as the contract and the TMS as the source of truth.

Business value

Every line of every carrier invoice is compared with the shipment and the contract; overbilling is disputed before payment.

Systems involved

ERP (SAP S/4HANA in the modelled case); SharePoint evidence archive and dispute register; Power BI

Business problem

Logistics cost control

Freight is bought on a contract but billed on a document. The rate card says what a pallet to zone 3 costs on a two-day service; the invoice says what the carrier's billing system produced, with the fuel surcharge of the month, a waiting-time charge and, occasionally, a shipment already billed in May. The two only meet if somebody compares them line by line.

Nobody has time for that at 14,000 shipments a month. Logistics checks a sample and approves the rest, because the invoice is due and the carrier relationship matters. Finance sees a total per carrier and can say whether it is within budget, not whether it is right. Procurement renegotiates rates with the carrier's own statistics.

The damage is quiet: a fuel percentage from last month, a weight break rounded upward, an accessorial charged twice. Each is a few euros, none shows in a total, and together they compound over hundreds of thousands of lines a year.

How it works today

What we find before automation, whatever the TMS.

  1. PersonCarrier invoices arrive as PDF, Excel or CSV attachments in the logistics mailbox and are filed by carrier and month
  2. PersonA freight analyst picks twenty or thirty lines per invoice and looks them up in the TMS and the carrier's rate-card spreadsheet
  3. WaitingQuestions about unfamiliar surcharges go to the carrier by email; a reply takes days, the due date does not move
  4. PersonThe invoice is signed off, coded to the freight cost centre and posted in full by accounts payable
  5. Risk of errorLines outside the sample, shipments billed twice and fuel surcharges at the wrong percentage are paid as invoiced
  6. SystemMonth-end freight reporting is built in Excel from ERP totals per carrier, with no link to shipments
  7. Risk of errorAnnual rate negotiations rely on the carrier's own volume and service reports
PersonWaitingRisk of errorSystem

Why the current process costs more than it appears

The bill that never reaches the budget.

  • Two hours per invoice buys a sample, not an audit. The lines nobody opens are exactly where a systematic error hides.
  • Overbilling that is caught is caught late: a dispute raised six weeks after delivery meets a closed period and is usually written off to protect the relationship.
  • Each rate-card change or new surcharge code opens a period in which the reference itself is out of date; nobody can say later which version was applied.
  • Peak season doubles the lines but not the checking hours, so the sample shrinks just when carriers add temporary surcharges.
  • Knowledge of what each carrier tends to get wrong sits with one or two analysts; when they leave, it leaves with them.

Cost of inaction

Twelve months of sampling 220 invoices by hand≈ €142,560
Three peak seasons checked at today's coverage≈ €427,680
If a fourth distribution centre brings the volume to 320 invoices a month (per year)≈ €207,360

What a sample cannot see dwarfs the checking cost, and round numbers illustrate it: 14,000 shipments a month at an average of €60 per shipment is about €10 million a year paid to carriers. If 1% of that is billed above the contract, that is €100,000; if today's sample catches a third, about €67,000 a year is paid and never recovered. The figures show the mechanism, not a benchmark; the real rate is what a re-audit will show.

Left as it is, every rate-card renewal is negotiated on the carrier's numbers, every peak season lowers the coverage further, and the next analyst to leave takes the knowledge of which carrier bills waiting time twice.

Illustrative scenario

A plausible organisation with realistic proportions. The figures are there to be recalculated on your data; they are not a client result.

Organisation

A European non-food retailer: 3 distribution centres, around 400 stores, an online channel; a TMS, SAP S/4HANA, Microsoft 365 E3 with Power BI Pro; two logistics analysts check carrier invoices.

Volume

14,000 shipments a month across 9 carriers; 220 carrier invoices a month averaging 64 lines, from a dozen to over 1,200; six carriers invoice as PDF, three as Excel or CSV.

Current process

About two hours per invoice: sampling lines against the TMS export and the rate card, emailing the carrier about surcharges, signing off; accounts payable posts the full amount.

Bottleneck

440 hours a month on checks that cover roughly a third of the lines; disputes leave weeks after delivery and are often abandoned; in the fourth quarter the lines double while the checking capacity does not.

Solution

PDFs are read by UiPath Document Understanding, spreadsheets are parsed; robots match each line to its shipment, re-price it from the versioned rate card on SharePoint, dispute the differences and post the approved amount; analysts see only disputed lines in Teams; Power BI shows savings and a scorecard per carrier.

Potential outcome

In the modelled case, coverage rises from a sample to every line, checking time falls to the minutes spent on exceptions, disputes reach the carrier within days, and 440 hours a month move to carrier management. The figures are a model, not a measurement.

Proposed solution

Mientha designs and delivers a freight audit that treats the rate card as the contract and the TMS as the source of truth, on the UiPath Platform and the Microsoft 365 tenant you already run. Rate cards, accessorial price lists and fuel surcharge tables become versioned tables on SharePoint with validity dates, maintained in Excel by logistics procurement. Shipments come from the TMS by API or scheduled export with the attributes that drive price.

Invoices in every format go through one queue. PDFs are read by UiPath Document Understanding, using the pre-trained Invoices Shipping model refined on your carriers' layouts through the validations your team makes anyway; Excel and CSV invoices are parsed without machine learning. From there the process is deterministic: robots match each line to a shipment, recompute the charge from the rate-card version valid on the shipment date, apply the fuel percentage of the month and the accessorials the TMS records, and compare. Every difference gets a reason code.

Lines within tolerance are approved. Larger differences become a dispute note per carrier and invoice, sent from the logistics mailbox with the shipment evidence and tracked in a register on SharePoint until the credit note arrives. The approved amount is posted to the ERP; the treatment of the disputed remainder is a rule you set per carrier. Analysts see only what needs a judgement, as Action Center tasks in Microsoft Teams; the freight budget owner approves write-offs in the Teams Approvals app. Power BI reports savings, dispute ageing and a scorecard per carrier.

Native capabilities used

UiPath Document Understanding pre-trained Invoices Shipping and Invoices models with Validation Station; UiPath Orchestrator queues, triggers and audit; UiPath Integration Service connectors for Microsoft Outlook 365 and Microsoft OneDrive & SharePoint (files, lists, Excel Online); UiPath Action Center actionable notifications in Microsoft Teams; Microsoft Teams Approvals app; Power BI

What we build

The versioned rate-card data model, invoice intake for PDF, Excel and CSV, the matching and re-pricing engine with reason codes and tolerances, dispute templates, register and reminders, ERP posting, the Power BI savings report and carrier scorecard, the runbook

Custom integration

TMS shipment extract through the TMS API or a scheduled export; ERP posting through UiPath SAP activities (BAPI) in the modelled case, or another ERP's interface; carrier portal downloads where needed

How the automated process works

  1. AutomationA new carrier invoice in the logistics mailbox is queued in Orchestrator; PDFs go to Document Understanding, spreadsheets are parsed, repeated invoice numbers are stopped at intake
  2. SystemRobots pull the period's shipments from the TMS and the rate card and fuel percentage valid for those dates from SharePoint
  3. AutomationEvery line is matched to a shipment by reference (or by date, destination and weight) and re-priced from the contract
  4. AutomationLines within tolerance are approved and the amount posted to the ERP; invoice, extraction and result are archived on SharePoint
  5. PersonDisputed lines above threshold, unmatched shipments and low-confidence extractions arrive as Action Center tasks in Teams; the analyst confirms, corrects or accepts
  6. AutomationDispute notes are emailed per carrier and invoice and followed up until a credit note is booked in the register
  7. AutomationPower BI refreshes savings, dispute ageing and the scorecard; a monthly summary goes to the logistics channel in Teams
AutomationSystemPerson

Human-in-the-loop model

Automation handles

  • Capture and extraction of every invoice, whatever the format
  • Matching, re-pricing and reason-coded classification of every line
  • Dispute notes, the register, carrier reminders and posting of approved amounts
  • Savings, dispute and scorecard reporting

People decide

  • Whether a disputed surcharge stands when the carrier sends evidence, such as a signed waiting-time record
  • Tolerances, dispute thresholds and the treatment of the disputed remainder per carrier
  • Write-offs and settlements above the threshold, approved in Teams by the freight budget owner
  • Rate-card and fuel-table changes, entered by logistics procurement and versioned before robots use them

Before and after

BeforeAfter
Share of invoice lines checked against the contracta sample, roughly a third100%
Checking time per invoice~2 hminutes, for disputed lines only
Time from invoice to disputeweeks, if at alldays, with evidence
Freight cost visibilityERP totals per carriercost per shipment by lane, carrier and service in Power BI

Systems and integrations

Where a rule suffices we do not use a model. Where judgement is needed, a person decides.

Inputs

  • carrier invoices as PDF, Excel and CSV in the logistics mailbox or from carrier portals
  • shipment extract from the TMS
  • rate cards and fuel tables on SharePoint

Automation layer

  • UiPath Orchestrator
  • UiPath Robots
  • UiPath Document Understanding
  • UiPath Integration Service
  • UiPath Action Center

Target systems

  • ERP (SAP S/4HANA in the modelled case)
  • SharePoint evidence archive and dispute register
  • Power BI

Human touchpoints: Action Center tasks in Teams; Teams Approvals for write-offs; monthly scorecard in a Teams channel

carrier invoices as PDFUiPath OrchestratorUiPath RobotsERPAction Center tasks in Teams

Technologies used

UiPath Document Understanding (IXP)

pre-trained Invoices Shipping and Invoices models read PDF invoice headers and lines; Validation Station for low-confidence fields

A
UiPath Robots + Orchestrator

queue every invoice, run matching and re-pricing, post approved amounts, log and audit

A
UiPath Integration Service (Microsoft Outlook 365, Microsoft OneDrive & SharePoint connectors)

mailbox intake; rate cards, dispute register and evidence on SharePoint

A
UiPath Action Center in Microsoft Teams

dispute review and validation tasks completed in Teams

A
Microsoft Teams (Approvals app)

write-offs and settlements approved by the budget owner

A
Microsoft Excel and SharePoint

versioned rate cards, accessorial lists and fuel tables

A
Power BI

savings, dispute ageing and the carrier scorecard

A
TMS and ERP interfaces (custom integration)

shipment extract by API or export; posting through UiPath SAP activities (BAPI) or another ERP's interface

C
Averified product capability (vendor documentation)Cillustrative model — the figures on this page

Illustrative economic model

Start by questioning the assumptions.

Illustrative model
220 carrier invoices a month × 120 minutes of manual checking= 440 h / month
440 h × €27 fully loaded hourly cost= €11,880 / month
× 12 months≈ €142,560 / year
Annual checking capacity released (illustrative)≈ €142,560

Built from ranges we see in retail and distribution logistics, not from a client measurement. Two hours per invoice is an average across a few very long invoices and many short ones, including emails to carriers; €27 is a fully loaded hourly cost for a logistics analyst in Central Europe. Checking capacity only: the overbilling on the lines nobody checks today comes on top and is illustrated in the next section. Capacity released is yours to allocate, not a headcount statement.

Run the numbers on your data

hours released per month
of annual capacity released

An illustrative estimate from your own inputs. It models released capacity; it is not a promise of savings.

Business benefits

  • Every line of every carrier invoice is compared with the shipment and the contract; overbilling is disputed before payment
  • Checking time per invoice falls from about two hours to minutes on disputed lines; in the modelled case that releases 440 hours a month
  • Disputes leave within days with the evidence attached, while the carrier can still trace the delivery
  • Fuel and rate-card changes apply from the month they take effect, and the version used for every line is on record
  • Freight cost becomes known per shipment, lane, carrier and service level, not from ledger totals
  • Peak-season volumes are audited at the same coverage as any other month, without temporary staff

The management view

  • Freight spend becomes a controlled cost: every euro paid to a carrier is traceable to a shipment, a rate-card version and a decision
  • Billing accuracy is measured per carrier, which gives procurement its own evidence in rate negotiations
  • Dispute ageing and recovered amounts are visible to logistics and finance in one report, so nothing is written off by default
  • The audit no longer depends on the two people who know each carrier's habits

Board-level KPIs

freight cost per shipmentoverbilling identified and recoveredaudit coverage (share of lines checked)dispute cycle timecarrier billing-accuracy score

Security and governance

An auditor should be able to reconstruct every decision.

  • Read-only on the TMS, posting-only in the ERP, mailbox access scoped through Microsoft Graph: every robot works under its own service account, and secrets live in the Orchestrator credential store or Azure Key Vault
  • Rate cards are contract data: versioned on SharePoint, changed by logistics procurement with approval; every re-priced line records the version it was checked against
  • Duties stay separated: the robot proposes, the analyst disputes or accepts, accounts payable posts, the budget owner approves write-offs in Teams
  • Consumer delivery addresses are cut down to the fields the audit needs and held under your Purview retention policy; the carrier invoices, shipment data and audit trail around them stay in your Microsoft 365 tenant, processed from the EU region of UiPath Automation Cloud
  • Document Understanding models are trained on your carriers' documents in your tenant; extraction confidence and every validation decision are logged

Why now

01

Mandatory KSeF e‑invoicing has covered large taxpayers in Poland since 1 February 2026 and the rest since 1 April 2026, so domestic carriers already send structured line data; one flow handles KSeF invoices and foreign carriers' PDFs

02

The unchecked lines cost money today: in the modelled case 440 hours a month buy a sample, and overbilling in the other two thirds of the lines is paid at every payment run

03

The pieces have matured: a pre-trained Invoices Shipping model, Outlook and SharePoint connectors, Action Center tasks in Teams and Power BI on one tenant remove most of the custom development this used to need

Relevant executive roles

COO / Supply Chain Director

Freight cost per shipment becomes a number derived from audited lines, and carrier management gets the capacity the checking used to consume

CFO

Freight stops being a total that cannot be verified; every posting carries a shipment, a contract version and a decision

Head of Logistics Procurement

Rate negotiations start from the company's own line-level history and a scorecard per carrier, not from the carrier's report

Common questions and objections

Our carriers send Excel invoices, so we do not need document AI.

Correct, and the design reflects it: spreadsheets are parsed without machine learning; Document Understanding serves only the carriers that send PDFs. The value is in the matching, not in the reading.

Won't disputing every line damage the relationship with our carriers?

Tolerances keep trivial differences out, and what goes out is one structured note per invoice with the evidence, which a billing desk handles faster than a thread of emails. Carriers that bill accurately like being measured.

Our rate cards change every quarter and come in nine formats.

Digitising them into one versioned model is the first deliverable; from then on a rate change is a table update with a validity date, not a development task.

When this is not the right solution

  • One or two carriers on a flat tariff with at most a few dozen invoices a month, where a spreadsheet check is cheaper
  • Shipments are not recorded with references outside the carriers' own portals, so there is nothing independent to match against; data capture, for instance a digital CMR and proof of delivery, comes first
  • Transport is bought spot, per shipment, with no rate card to audit against; the useful automation there is quote capture and comparison

A question for the next management meeting

When did we last compare a carrier invoice, line by line, with the rate card we signed, and what did the lines nobody opened cost us?

Implementation approach

We start with one slice of the process and extend only once it is proven.

We deliver

  • Three months of invoices and shipments read end to end: formats, reference quality, surcharge codes, rate-card structures, a first difference rate per carrier
  • Digitised rate cards: one versioned model on SharePoint for lanes, zones, weight breaks, service levels, accessorials and fuel tables
  • Invoice intake for every format, the matching and re-pricing engine, and the dispute flow with register, reminders and Teams touchpoints
  • ERP posting of approved amounts, the Power BI savings report and the carrier scorecard
  • A re-audit of the last three months during validation, doubling as your first claim list; go-live with hypercare and a runbook

We need from you

  • Three months of carrier invoices in their original formats, with the TMS shipment export and the ERP postings
  • Current rate cards, accessorial price lists and fuel surcharge history per carrier; a process owner in logistics and a contact in accounts payable
  • Access to the TMS by API or export, a test ERP client and a service account for the logistics mailbox

Stages

Discovery

Invoice formats, reference quality, rate-card structures, exception categories and volumes per carrier

Design

Rate-card data model, matching rules, tolerances, dispute policy per carrier, security model

Build

Rate-card tables, extraction and parsers, matching engine, dispute flow, ERP posting and Power BI

Validation

Re-audit of three months of historical invoices, comparison with the manual sample, acceptance by logistics and finance

Go-live and tuning

Carrier by carrier, highest volume first; analysts supervise every dispute at first, then reviews settle into a quarterly rhythm

Departmental. Effort depends on the number of carriers and rate-card structures, whether the TMS offers an API, and how reliably carriers print your shipment reference on their lines.