Every carrier invoice that hits your AP queue should match the rate confirmation you signed when the load was booked. In theory, that's simple: same linehaul, same fuel surcharge, same accessorials, same total. In practice, mismatches show up often enough that auditing every invoice is part of the job — and skipping that audit is how overpayments pile up.
If you're running a small brokerage or 3PL, a proper carrier invoice audit is one of the highest-ROI habits you can build. You just need a system that doesn't make you retype every number twice.
What an audit actually checks
A carrier invoice audit is a line-by-line comparison between what the carrier billed you and what you agreed to on the rate confirmation. You're looking for four kinds of mismatches:
Rate mismatches. The linehaul on the invoice is higher than the rate con. Sometimes it's a flat typo. Sometimes it's a rounded-up fuel surcharge. Either way, you owe the rate con amount, not the invoice amount.
Unauthorized accessorials. A liftgate fee for a dock delivery that didn't need one. A detention charge without a signed POD to back it up. A TONU (Truck Order Not Used) on a load that was actually delivered. These are the most common overcharges.
Missing credits. The rate con promised a discount for early pickup or a specific customer rebate, and the invoice doesn't reflect it.
Wrong reference numbers. The invoice references a different BOL or PO than the one on the rate con, making it impossible to tie the charge back to a real load.
Small brokerages that audit consistently tend to catch discrepancies on a meaningful percentage of their invoices. That's not because carriers are bad actors — it's because billing is manual, complicated, and subject to the same typos and miscalculations as any other paperwork.
Why most small brokers skip the audit
The math on auditing every invoice is straightforward, but the labor isn't. If you're keying carrier invoice data into a spreadsheet by hand and then flipping back to the original rate con PDF to compare, a single audit takes 5–10 minutes. Multiply that across every load you pay, and "audit every invoice" turns into a full-time job.
So it gets skipped. Or it gets done selectively — only for invoices above a dollar threshold, or only for loads from certain carriers. The rest of the invoices get paid as submitted, and anything wrong gets absorbed.
The irony is that the smallest brokerages are the ones who can least afford to absorb those costs, but also the ones least likely to have a systematic audit process.
What automation actually changes
The audit itself is simple. The bottleneck is getting both documents into a shape where they can be compared without manual data entry.
If you extract the rate confirmation into structured data (carrier, MC number, linehaul, fuel, accessorials, total) and extract the carrier invoice into the same kind of structured data, the comparison becomes trivial. You line them up in a spreadsheet, side by side, and anything that doesn't match jumps out.
Both documents have the same shape when extracted — charges, references, origin and destination — so you're not reformatting anything. You're just looking at two rows and flagging the differences.
With confidence scoring on the extracted fields, you also know which numbers to trust immediately and which ones to double-check against the source PDF. A clearly printed invoice total with high confidence can be compared directly. A handwritten accessorial line with low confidence gets a quick human review before it feeds into the audit.
A simple workflow for small brokerages
Here's a workflow that works without adding a dedicated AP analyst to your team:
1. Extract both documents at booking and billing. When you book a load, upload the rate confirmation and save the extracted data. When the carrier's invoice arrives, upload it and save the extracted data.
2. Match them by BOL or PRO number. Because both extractions include the reference number, pairing them is a simple lookup. No hunting through folders.
3. Compare the charge breakdown. Linehaul, fuel surcharge, and each accessorial. If every line matches within a reasonable tolerance, the invoice clears. If anything is off, flag it for review.
4. Dispute what doesn't match. For any mismatch, pull the BOL, POD, and any dock timestamps, and send a clean dispute to the carrier with the documentation attached. Laneproof, FreightWaves, and several brokerage training guides all agree: well-documented disputes resolve quickly. Invoices with nothing backing them don't.
5. Batch audit at end of week. Instead of auditing each invoice as it arrives, run a weekly audit pass across everything in your payables queue. Export the extracted data for the week and scan for mismatches. This is faster than one-at-a-time and catches patterns (like a specific carrier consistently overbilling fuel).
The downstream effect
A brokerage that audits every invoice tends to find a few things over time. One or two carriers that consistently overbill small accessorials. A single shipper whose detention charges never line up with the POD timestamps. A recurring fuel surcharge calculation error on a particular lane.
Once you've spotted the pattern, you can address it directly with the carrier or shipper. The audit isn't just about recovering this month's overcharges — it's about understanding where your billing leaks are.
And the cost of running the audit drops as you automate the document side of it. Once the rate cons and invoices are extracting cleanly, the audit itself is a spreadsheet comparison. The work is in the exceptions, not the routine matches.
Try it with a real pair of documents
The best way to see what this looks like in practice is to run an audit on one load end-to-end. Sign up free on CargoParse, upload a rate confirmation and the matching carrier invoice, and compare the extracted fields side by side. No credit card required.
For more on the individual document types, read how to extract rate confirmation data and how to extract freight bill data from carrier invoices. And if BOL and POD mismatches are driving detention disputes, that's a separate problem worth tackling with the same playbook.