In 6 weeks, a Mexican logistics operator audited 100% of its flow: 2,250 invoices on ~$63M MXN. It documented $3.6M MXN with a file (5.7% of spend) and recovered or prevented ~$2.5M MXN (withheld + credit + renegotiated). The mix spanned local, domestic linehaul, cross-border, and last-mile trips.
- invoices audited (100%)
- 2,250
- MXN with a file (5.7%)
- $3.6M
- MXN recovered / prevented
- ~$2.5M
- pilot · 428 with findings
- 6 wk
Context: fake freight digitization · 15-point audit checklist · what freight audit is.
Logistics operator in Mexico: the setting
They already had a TMS, portals, GPS, and CFDI. They still paid by sampling: ~1 in 10 invoices by hand (~30 min and 3 emails) and the rest went out almost blind.
Before (human bridge)
- Sampling: ~1 in 10 invoices
- Analyst hunts rate, CFDI, Carta Porte, GPS, POD
- Finance pays first; disputes later (if ever)
- TMS only stores what someone keyed
With OCL (finished work)
- 100% of pilot flow (2,250)
- Agent matches; the team only decides
- Finance gets the finding before release
- OCL executes the pre-pay close
“We knew we were overpaying, but we did not have the capacity to review every invoice in detail. It was a choice between payment speed and accuracy.”
Pilot volume: 6 weeks
The universe is a mid-size operator consolidating dozens of carriers: 2,250 invoices in the pilot (~375/week pace) and ~$42M MXN monthly freight spend. The flow mixed local and domestic linehaul, cross-border legs, and last mile.
| Parameter | Figure | How to read it |
|---|---|---|
| Duration | 6 weeks | 1.5 months · usual OCL range 6-8 weeks |
| Pace | ~375 invoices/week | Typical multi-carrier consolidation |
| Invoices audited | 2,250 | ~375/week × 6 = 2,250 (100% of the flow) |
| Audited spend | ~$63M MXN | ~$42M/month × 1.5 |
| First useful alert | Day 9 | Sources online and pattern validated |
Timeline
Three pilot milestones
Elige un paso para ver el detalle
Detalle del paso · 01
Sources online
Funnel: from 2,250 invoices to pesos
Two different percentages (do not mix them): incidence is the % of invoices with a finding; % of spend is the amount with a file over ~$63M.
- 01
2,250 invoices audited
100% of the flow in 6 weeks · ~$63M MXN spend.
- 02
428 with a finding (19%)
1 in 5 invoices · sector band ~18-22%.
- 03
$3.6M with a file (5.7%)
Detected, documented, and withholdable or disputable.
- 04
~$2.5M recovered / prevented
Withheld + credit + renegotiated · ~69% of the file.
CFO read
The same story in four numbers
2,250
Audited
428
With findings
$3.6M
File
~$2.5M
Cash impact
Type A: gaps by trip type
A Mexican logistics operator does not bill “one freight type.” The pilot split the universe into four modes: local, domestic linehaul, cross-border, and last mile. Each has a distinct audit pain and a finding rate that sits inside credible sector bands.
FIGURE 1 · GAPS BY TRIP TYPE
Where the money went when 100% of the mix is audited.
Composition of $3.6M MXN with a file by operating mode (2,250 invoices · 428 with findings).
OCL logistics-operator case · 6 weeks · pilot points inside sector bands 12-28%.
Linehaul
Invoices: 900
With findings: 162
Finding %: 18%
Amount: $1.44M
Typical pain: Tolls, FSC, km vs quote, weight
Local
Invoices: 675
With findings: 135
Finding %: 20%
Amount: $0.72M
Typical pain: Wait time, tariff zone, multi-stop
Cross-border
Invoices: 315
With findings: 79
Finding %: 25%
Amount: $1.08M
Typical pain: Transfer, border detention, USD, docs
Last mile
Invoices: 360
With findings: 52
Finding %: ~14%
Amount: $0.36M
Typical pain: Reattempt, urban wait, accessorials
Pilot total
Invoices: 2,250
With findings: 428
Finding %: 19%
Amount: $3.60M
Typical pain: 5.7% of audited spend
Sector reference bands versus pilot point: local 18-24% (here 20%); linehaul 15-20% (18%); cross-border 22-28% (25%); last mile 12-18% (~14%). The total mix lands at 19%, inside the MX ~18-22% band when leaving sampling.
| Type | What breaks on the invoice | What the agent matched |
|---|---|---|
| Local | Dock wait, wrong zone, extra stops | GPS + dock + zone rate card |
| Linehaul | Tolls outside all-in, FSC, inflated km, overweight | Route vs toll + rate card + scale |
| Cross-border | Transfer, border detention, USD leg, broken docs | Border milestones + FX + Carta Porte/CFDI |
| Last mile | Failed delivery, reattempt, urban wait, accessorials | POD + window + accessorial catalog |
Type B: findings by concept ($3.6M)
The second typology cuts the same file by primary concept (one dominant finding per invoice) so counts and pesos sum without double counting. The table closes at $3.6M and 428 invoices.
FIGURE 2 · PILOT FILE
Six concepts sampling almost never catches together.
Composition of $3.6M MXN detected with a file when auditing 100% of the flow (2,250 invoices · 428 with findings).
OCL logistics-operator case · 6 weeks · $3.6M with a file = 5.7% of spend.
| Concept | Invoices | % | Amount | What the agent matched |
|---|---|---|---|---|
| Unsupported detention | 150 | 35% | $1.40M | Hours vs GPS and dock / free time |
| Off contracted rate | 90 | 21% | $0.90M | Line vs lane rate card |
| Tolls not contracted | 55 | 13% | $0.50M | Toll vs all-in or catalog |
| Weight without evidence | 45 | 11% | $0.36M | Weight vs scale / Carta Porte |
| Duplicate accessorials | 35 | 8% | $0.28M | Same charge across docs |
| Carta Porte / CFDI gaps | 53 | 12% | $0.16M | Complement, UUID, IdCCP |
| Total with a file | 428 | 100% | $3.60M | 5.7% of audited spend |
Average per invoice with a finding: ~$8,400 MXN detected. About 60 large cases held nearly half the recovered dollars. That is why sampling fails: expensive cases are few and do not show up in “one in ten.”
From $3.6M with a file to ~$2.5M recovered
Detecting is not collecting. In the pilot we separated the amount with a file from cash impact: withheld, credited, and renegotiated sum to ~$2.5M.
| Destination of the file | Amount | What it means |
|---|---|---|
| Withheld pre-pay | $1.65M | Did not leave cash · closed finding |
| Credit / adjustment | $0.58M | Carrier corrected or credited |
| Renegotiated at settlement | $0.27M | Discount when closing the period |
| Recovered / prevented (pilot) | ~$2.50M | Sum of the three rows · ~69% |
| Open dispute at close | $0.72M | File ready · no cash close yet |
| Released (carrier documented) | $0.38M | Later evidence supported the charge |
| Total with a file | $3.60M | 1.65+0.58+0.27+0.72+0.38 |
- detected · 5.7% of spend (~$63M)
- $3.6M
- recovered / prevented (~4.0%)
- ~$2.5M
- file → cash conversion
- ~69%
Time: reconcile went from ~30 min to ~3 min per invoice; +60 h/month per analyst; 90% of the process without human intervention.
“In six weeks the agent found more billing errors than our team in the prior six months. It was not that they did not know; they simply could not review everything.”
What the Audit Agent executed
Here they did use OCL Cargo as the autonomous TMS: the system that runs pre-pay audit, matches the sources, and leaves the file with Finance. They did not migrate every master on day one. OCL can stamp invoices and Carta Porte when the flow requires it; in this pilot the measured value was buyer-side pre-pay matching.
Cycle
From inbound invoice to decision
Ingest
Normalized invoice
Match
5 sources · ~3 min
Finding
Type, amount, evidence
Finance
Hold or release
Elige un paso para ver el detalle
Detalle del paso · 01
Rate
- 01
OCL Cargo (autonomous TMS)
Ingest, match, classify, and pack the file.
- 02
Sources they already had
Rates, CFDI, Carta Porte, GPS, and POD.
- 03
Team (exceptions)
Approve, escalate, or release. No pasting across screens.
- 04
Finance (pay)
Hold or pay with evidence. Dispute with a file.
Checklist to replicate the pilot
If you want the same funnel on your operation, measure baseline first and close with pesos and hours, not “% TMS usage.” Classify by trip type and by concept.
Elige un paso para ver el detalle
Detalle del paso · 01
Sampling baseline
12-month projection
Gross detection run-rate: $3.6M in 1.5 months × 8 = $28.8M/year. Conservative projection: $18M/year (~3.6% of annual spend), ~37% below gross, because early weeks clear backlog, carriers correct billing, and detect ≠ collect.
- Detected (pilot)
- $3.6M
- Recovered / prevented
- ~$2.5M
- Conservative / year
- $18M
With a file, disputes close; without a file, most die as ignored complaints. The $18M annual figure already discounts collection friction on documentable detection.
Related reading
- Fake freight digitization: the human bridge behind the TMS.
- Traditional TMS vs OCL: record vs finished work.
- 15-point checklist to audit freight invoices.
- What freight audit is.
- Freight accounts payable pre-pay audit.
Key takeaways6 points
- 2,250 invoices in 6 weeks (~375/week pace) on ~$63M MXN freight spend.
- 428 invoices (~19%) with a finding; amount with a file was $3.6M = 5.7% of spend.
- Type A (trip): linehaul, local, cross-border, and last mile sum to 2,250 / 428 / $3.6M; cross-border ~25% finding rate, linehaul 40% of $.
- Type B (concept): detention, rate, tolls, weight, duplicates ($0.28M), and Carta Porte/CFDI ($0.16M).
- ~$2.5M recovered/prevented: withheld pre-pay + credit + renegotiated (~69% of the file).
- Time: ~30 min to ~3 min per invoice; +60 h/month per analyst; 90% hands-off.
Does your TMS record while Finance still pays blind?
Frequently asked questions
Pre-pay freight audit: match every invoice against rate, CFDI (Mexican electronic tax invoice), Carta Porte, GPS, and POD (proof of delivery), pack a file, and send it to Finance before releasing payment. Not a day-one full TMS (Transportation Management System) replacement.
$3.6M MXN is the amount detected and documented with a file (withholdable or disputable): 5.7% of audited spend (~$63M). ~$2.5M MXN is what was recovered or prevented in the pilot: $1.65M withheld pre-pay + $0.58M credits/adjustments + $0.27M renegotiated. The rest of the file stayed in open dispute ($0.72M) or was released when the carrier documented ($0.38M).
Of 2,250 invoices: linehaul 900 (162 findings, $1.44M), local 675 (135, $0.72M), cross-border 315 (79, $1.08M), last mile 360 (52, $0.36M). They sum to 428 with findings and $3.6M. Cross-border had the highest finding rate (~25%); linehaul held more pesos by volume.
Of 2,250 invoices, 428 (~19%) had at least one finding. By primary concept: detention $1.40M (150), off-rate $0.90M (90), tolls $0.50M (55), weight $0.36M (45), duplicate accessorials $0.28M (35), Carta Porte/CFDI gaps $0.16M (53). They sum to $3.6M.
Yes. In this pilot OCL Cargo was the autonomous TMS that ran pre-pay audit and left the file with Finance. They did not migrate every master on day one: the agent used computer use (screen operation like a tower operator) or connections on the sources they already had. See fake digitization. OCL can also stamp invoices and Carta Porte when the flow requires it.
In this case the 5-7% band matches detected with a file (5.7% = $3.6M ÷ $63M). Recovered/prevented was ~4.0% of spend ($2.5M ÷ $63M): ~69% conversion of the file in six weeks. We do not invent 5-7% on cash if the math does not close.
Reconcile dropped from ~30 min and 3 emails to ~3 min per invoice. That returned +60 h/month per billing analyst. 90% of the process ran without intervention; the team only decided on documented exceptions.
Gross detection run-rate: $3.6M in 1.5 months × 8 = $28.8M/year. Conservative projection: $18M/year (~3.6% of annual spend), ~37% below gross for early backlog, carrier corrections, and detect ≠ collect. The pilot reports $3.6M detected and ~$2.5M recovered/prevented.