A full truckload BOL from a mid-sized shipper contains nine required fields. Seven minutes to enter manually. If you're a freight broker processing 20 loads per day, 15 of those loads will have a BOL arrive by the time your day wraps up. That's 105 minutes of data entry. Every day. From a single document type.
The BOL is legally the most important freight document in a shipment. It's the contract between shipper and carrier. It's the basis for cargo claims. And for most brokerages, it's processed by someone opening a PDF and typing into a TMS field by field.
This article covers how BOL data extraction actually works, what fields it pulls, where the accuracy numbers come from, and how to connect it to your current workflow.
Bill of lading data extraction automatically reads a BOL document (PDF, photo, or email attachment), identifies all standard fields using OCR and AI, and populates your TMS or tracking system without manual entry. The core fields extracted are shipper, consignee, origin, destination, commodity, weight, pieces, PRO number, and any special instructions. On a clean digitally-created BOL, extraction and review takes under 30 seconds total.
What a bill of lading actually contains
A bill of lading (BOL) is the legal contract between a shipper and a carrier for the transportation of goods. It's issued by the shipper or the freight broker acting on the shipper's behalf, signed by the carrier at pickup, and used as the basis for delivery confirmation, cargo claims, and invoice reconciliation.
A standard BOL contains nine required fields: shipper name and address, consignee name and address, commodity description, weight (in pounds), piece count, freight class (for LTL), PRO or BOL reference number, carrier name, and any special handling instructions. Most BOLs also include pickup and delivery dates, load reference numbers that link back to the rate confirmation, and accessorial requirements.
BOLs arrive in freight brokerage operations from multiple sources. Shippers email them as PDFs before or after loading. Carriers submit them through portals after pickup. Drivers photograph them and send via WhatsApp or text. The document is the same; the delivery channel varies.
What manual BOL entry actually costs at scale
Manual BOL data entry takes seven minutes per document. On a 20-load-per-day brokerage, roughly 15 BOLs arrive daily (some loads have no BOL until delivery). That's 105 minutes of BOL entry alone, before processing rate confirmations, invoices, or PODs.
At a $25/hour coordinator rate, 105 minutes is $43.75 per day in BOL entry labor. Over 22 working days, that's $962.50 per month on one document type. A broker running 40 loads per day doubles that number.
The error cost adds separately. Manual entry from PDFs introduces errors in 2% to 4% of fields. On a nine-field BOL at 15 documents per day, that's approximately 2.7 to 5.4 wrong fields entering your TMS daily. The most consequential errors are on commodity description (which affects cargo claims), weight (which affects weigh station liability), and consignee name and address (which affects delivery routing).
How BOL data extraction works
BOL extraction follows the same pipeline as rate confirmation extraction: OCR converts the document to machine-readable text, AI identifies and maps values to the correct fields, and confidence scoring marks each field for auto-approval or review.
BOLs vary more in format than rate confirmations. A shipper using a custom Word document template, a large manufacturer using SAP-generated BOLs, and a small shipper filling in a hand-typed form all produce very different looking documents. Template-based extraction tools require a separate profile for each format. AI-based extraction reads contextually, understanding that '12 pallets, 480 cases WHL-1042' maps to piece count and commodity even if it's not in a standard table.
The most variable field in BOL extraction is commodity description. BOLs range from very specific ('mixed electronics, 24 cartons, SKUs 8840-8857, fragile') to completely vague ('general freight, 42,000 lbs'). The extraction engine captures whatever is there. Vague commodity descriptions get flagged yellow for your attention, not silently passed through.
BOL field extraction accuracy
Extraction accuracy varies by field type. Here's how the nine standard BOL fields perform on clean, digitally-created PDFs:
| Field | Typical accuracy | What affects it |
|---|---|---|
| Shipper name and address | 92%+ green | Header position is consistent across most formats |
| Consignee name and address | 90%+ green | Usually mirrored shipper format |
| BOL or PRO number | 95%+ green | Clearly labeled in all standard formats |
| Weight (lbs) | 88%+ green | Informal notation (42K, approx) reduces confidence |
| Piece count | 85%+ green | Mixed units (pallets, cartons, cases) handled automatically |
| Commodity description | 78% yellow or green | Vague descriptions flagged for review |
| Freight class (LTL) | 80%+ green | Only present on LTL BOLs |
| Special instructions | 72% yellow | Unstructured text varies widely |
| Carrier name | 85%+ green | Often from email sender, verified against BOL header |
How BOLs arrive and how each gets processed
BOLs arrive through more channels than rate confirmations, and each channel has slightly different handling requirements.
- Email PDF attachment (most common for shipper-issued BOLs): Connect your Gmail or Outlook inbox. When a BOL arrives as an attachment, freightOptIQ detects it automatically within two minutes, runs extraction, and the processed fields appear in your Load Inbox. No manual download required.
- Carrier portal PDF download: Download the BOL from the carrier or load board portal and drag it onto the freightOptIQ dashboard. Extraction results appear in under 10 seconds for a clean PDF.
- Driver WhatsApp photo or text: Some drivers photograph BOLs at pickup and send via WhatsApp. For photo BOLs, use the freightOptIQ photo upload. The image goes through OCR before extraction. For any text details sent via WhatsApp, use paste-to-parse.
- Scanned paper BOL: Some shippers still issue paper BOLs that get scanned or photographed. Upload the image file. Accuracy depends on scan quality. The quality check flags images that are too low-resolution to extract reliably before you rely on the results.
Setting up BOL extraction in your operation
BOL extraction runs on the same inbox connection as rate confirmation extraction. If you've already connected your inbox, BOLs process automatically. If you haven't, here's the setup.
- 1Connect your inbox (if not already done)
In freightOptIQ Settings, click Email Connections and connect Gmail or Outlook via OAuth. This takes under two minutes. Once connected, all freight-related emails including BOL attachments are scanned every two minutes automatically.
- 2Review the first week of BOL extractions
In the first week, review all BOL extractions before approving them to build a sense of which fields extract cleanly from your specific shippers' formats and which typically need review. After the first week, your review focus narrows to flagged fields only.
- 3Set up your output
Configure where approved BOL data goes. Google Sheets on the Starter plan creates a new row per approved BOL. TMS integration on the Pro plan creates or updates the corresponding load entry directly in AscendTMS, Tai TMS, or Rose Rocket.
- 4Handle photo BOLs from drivers
For drivers who send BOL photos via WhatsApp, set up a simple process: forward the photo to your freightOptIQ email, or use the photo upload from the dashboard. This takes about 20 seconds and is much faster than transcribing a handwritten or low-quality scanned document manually.
What BOL extraction handles and what it does not
BOL extraction handles the data transfer problem. It doesn't handle the verification problem. Extracting what's written on a BOL is different from verifying that what's written is correct.
If a shipper issues a BOL with the wrong consignee address, extraction will capture that wrong address accurately. The extraction is correct; the source document is wrong. Carrier address verification and load-level cross-checks are a separate workflow.
It also doesn't handle freight class assignment for LTL shipments. Freight class is on the BOL when the shipper has already determined it. But if the freight class looks wrong for the commodity and weight, that's a human judgment call that extraction software doesn't make.
What it does handle: getting the nine standard fields from every BOL into your TMS quickly and accurately, flagging low-confidence fields before they become silent errors, and recovering the 105 minutes per day that your operation currently spends on manual entry.
Frequently asked questions
What is a bill of lading and why does it need to be in the TMS?+
A bill of lading is the legal contract between shipper and carrier governing the transport of goods. It contains the shipper, consignee, commodity, weight, piece count, and reference numbers. It needs to be in your TMS because it's the basis for cargo claims, delivery confirmation, invoice reconciliation, and compliance documentation. Without accurate BOL data in your system, any of those processes requires going back to hunt down the paper document.
How accurate is automated BOL extraction compared to manual entry?+
On clean digitally-created BOLs, automated extraction achieves 90%+ field accuracy on eight of the nine standard fields. The ninth (commodity description) runs 75% to 85% depending on how specific the shipper's language is. Manual entry by contrast has a 2% to 4% error rate per field, which increases as the reviewer processes more documents in a single session.
Can BOL extraction handle handwritten bills of lading?+
Partially. Handwritten BOLs go through OCR, and modern OCR handles printed handwriting reasonably well. Highly stylized or compressed handwriting reduces accuracy. The confidence scoring system flags any field where OCR confidence is low, so you know which fields need manual verification rather than relying on them silently.
Does BOL extraction work on LTL shipments as well as FTL?+
Yes. LTL BOLs include a freight class field that FTL BOLs don't, and the piece count and commodity fields are often more granular. The extraction engine handles both formats without separate configuration. Freight class, NMFC codes, and multi-line commodity descriptions on LTL BOLs are all extracted and scored individually.
What happens if a BOL arrives as a photo from a driver rather than a PDF?+
Upload the photo to freightOptIQ or forward the email containing the photo. OCR converts the image to text before extraction runs. Accuracy depends on photo quality. A clear photo of a clean BOL typically extracts at 80% to 88% field accuracy. A dark or blurry photo drops that to 60% to 70%, and the quality check flags the image before extraction so you're not relying on low-confidence results.
Stop doing this manually. freightOptIQ handles it in seconds.
Connect your Gmail or Outlook inbox, paste a WhatsApp message, or upload a PDF. First 50 documents are completely free.