Check every carrier invoice
against what you agreed.
Carrier invoice audit automation compares every line of a carrier or haulier invoice with the agreed rate card, the fuel surcharge rule, the extra charges allowed and the proof of delivery. It approves lines that match, flags overcharges with the evidence, drafts the dispute and posts approved invoices to your accounts.
Why do carrier invoices need an audit?
A carrier invoice is rarely one charge. It is a list of consignments, each priced from a rate card by zone, weight or pallet count, plus a fuel surcharge, plus extras such as tail-lift, waiting time or a failed delivery. Each line depends on facts the carrier records and you may not see.
Most businesses check carrier invoices by sampling, or not at all. A finance person cannot check 800 lines against a rate card each month. Errors are usually honest: a wrong zone, an old rate, a pallet charged as full when it was a half, waiting time charged when the driver app shows 15 minutes on site.
We do not quote a typical overcharge rate. It depends on the carrier, the contract and how complex the rate card is. The honest way to find out is to audit three months of past invoices before you build anything.
Is carrier invoice audit a job for an agent?
Tick what is true of your carrier invoice audit work.
Tick the statements that apply and the suggestion appears here: an AI agent, ordinary automation, or a person with software support.
What does the audit check on each line?
| Charge | What to check | Evidence needed |
|---|---|---|
| Base rate | Zone or lane, weight band, pallet size and service level match the rate card in force on the shipment date | Your shipment record; the rate card with its valid-from date |
| Fuel surcharge | The percentage matches the agreed rule for that period, applied to the right base | The index or table named in the contract, and the reference period |
| Extra charges | The extra is in the contract, the price matches, and the event happened | Driver app times, booking notes, POD remarks |
| Failed delivery or redelivery | The failure was not the carrier’s fault, and a redelivery took place | POD or driver note with reason; your customer contact log |
| Duplicates | The same consignment is not billed twice, on this invoice or an earlier one | All past invoice lines from that carrier |
| Shipment exists | Every billed consignment matches a shipment you booked | Your TMS or booking records |
The fuel surcharge deserves its own test. Contracts often link it to a published fuel index or to the carrier’s own monthly table, with a lag of one or two months. The audit needs the exact rule: which index, which month, which base it applies to and how it rounds. Small differences repeat on every line.
How do rate cards become something a system can check?
This is the hardest part, and most of the setup time. Rate cards arrive as PDFs, spreadsheets with merged cells or a clause in an email from two years ago. The system needs them as data: one row per lane or zone, weight band, service and price, each with a valid-from and valid-to date.
- Collect every rate card, surcharge table and agreed extra for each carrier.
- Convert each into a structured table. An AI model can draft this from a PDF, but a person checks every row against the original.
- Record ambiguous terms as questions for the carrier, not guesses. For example: is waiting time charged from arrival or from the booked slot?
- Store each version. An invoice for a March shipment must be checked against the March rates, even if it arrives in May.
Parcel and pallet networks often price by volumetric weight. If you do not record dimensions at dispatch, you cannot check volumetric charges. The audit will then report those lines as “not checkable” rather than pass them.
How much time could carrier invoice audit give back?
Put your own numbers in.
The starting values are an illustrative example, not a benchmark or a neu figure. Use your own volume, the time each one takes today and the share you expect the system to handle without a person.
Time saved is only part of the case: response times, missed work and errors matter too. The ROI guide covers how to set a baseline, and the cost guide what the build and running costs include.
How does proof of delivery fit in?
Many extra charges depend on what happened at delivery. Waiting time, a second attempt or a refused pallet should appear in the POD, the driver app or the carrier’s tracking events. If the evidence is missing, the line is held, not paid.
This is where carrier audit connects to proof of delivery processing. If PODs are already read and matched to shipments, the audit can check “was it delivered, when, and with what remarks” automatically. Without that, each disputed extra needs someone to find the POD by hand.
What happens after the checks?
Each invoice ends in one of three states:
- Approved: every line matches within agreed tolerances. The bill is posted to Xero, Sage or your accounts package, with the audit result attached.
- Part-approved: the matching lines are approved and the disputed lines are listed. Whether you short-pay or wait for a credit note is a policy decision. Agree it with the carrier first.
- Held: lines cannot be checked, for example a missing rate or an unknown consignment. A person decides.
For disputed lines, the system drafts a message to the carrier with the consignment, the amount billed, the amount expected and the evidence. A person sends it. Track each dispute until a credit note arrives, then match the credit note to the original lines.
What does the audit cost in time, by hand and automated?
Take a hypothetical distributor that receives 800 carrier invoice lines a month from three carriers. A careful manual check takes about 2 minutes a line: find the shipment, look up the rate, check the surcharge and look for evidence of extras. That is about 27 hours a month, which is why most firms only sample.
With automation, every line is checked. Assume 85% match within tolerance and need no one. The other 15%, or 120 lines, go to a person with the evidence already attached, at about 3 minutes each. That is 6 hours a month, and every line has been checked, not a sample.
These numbers are illustrative. Your match rate depends on how clean your rate cards and shipment data are. The one-off audit of past invoices gives you real figures to put in the calculator on this page.
Where should you start?
Start with one carrier that sends a high volume of lines and has a clear rate card. Run the audit on three months of past invoices as a one-off report. The result tells you the size of the problem and which checks matter. Then decide whether to automate the monthly run. For capture of the invoice itself, see invoice processing. For the whole payables process, see accounts payable automation.
Common questions
Can the system read carrier invoices in different formats?
Yes. Larger carriers often send a CSV or Excel file with the invoice, and that is the best input because each line is structured. PDFs work too, with AI extraction and checks on totals. Ask each carrier for a line-level data file. Most can provide one, and it removes most extraction errors.
What if our rate agreement is not written down clearly?
Then the audit will show you where. Ambiguous terms become a list of questions for the carrier. Agreeing them in writing is useful even if you never automate the audit, because disputes then have a clear answer. Until a term is agreed, the system holds those lines instead of guessing.
Is this only for large shippers?
No, but volume matters. With a few dozen lines a month, a careful person with a spreadsheet may be enough. The case for automation grows with the number of lines, the number of carriers and the complexity of extras. Run the one-off audit of past invoices first, then decide.
Should the system short-pay invoices automatically?
We advise against it. Short-paying is a commercial decision that affects your relationship with the carrier. The system should calculate the expected amount and draft the dispute, and a person should decide whether to pay in part, hold or accept. Record the decision so the pattern is visible later.
Bring us the task.
Tell us what happens today, which tools are involved and where it goes wrong. On a free 30-minute review, we’ll work out whether an agent, an automation or neither is the right fix, and what a sensible first release would be.
Start a project.

Book a call or send us a message. We read everything.
tash@neu.ie
Working with teams in Ireland and worldwide.
Free 30-minute AI review.