How to extract data from delivery dockets
Getting a docket into usable data is less about reading the words and more about knowing which part of the page is the actual table. Here's what works, what doesn't, and why.
Extract your first docket freeEncrypted files
Protected in transit and at rest
Private to your account
Never used to train AI models
Payments by Stripe
We never store your card number
Test before you buy
Check accuracy on your own docket
Decide what you actually need first
Most teams only need a handful of fields to do the job in front of them: what arrived, how much of it, and at what price. Pulling everything off the page usually creates more cleanup than it saves, so start from the question you're trying to answer — checking a delivery, costing a job, or reconciling a statement.
- Header data: supplier, docket number, date, order reference
- Line data: product code, description, quantity, unit
- Money data: unit price, line total, discount where shown
- Site or job reference if you cost deliveries per project
The four practical approaches
Retyping is the default and the most expensive. Copy-and-paste from a digital PDF drops everything into one column. Generic table converters treat the page as a grid and fold the letterhead into row one. Reading the page for meaning is the only approach that survives a supplier changing their layout.
- Manual typing — accurate if slow, and error-prone under pressure
- Copy and paste — only works on digital PDFs, loses the columns
- Template-based tools — break the moment a layout shifts
- Meaning-based reading — no template, handles scans and photos
Why the table is the hard part
A docket is a small table surrounded by noise: a letterhead, a delivery address block, terms in small print, a signature box and a totals row. The extraction has to know that only the middle band is data. Merged cells, wrapped descriptions running onto a second line and continuation pages are what break most tools.
What DocketFlow returns
Each upload comes back as an .xlsx with a Data sheet of one row per line item, consistent column headings across suppliers, and a Summary sheet naming the supplier, docket number and date. Numbers arrive as numbers so totals and pivot tables work immediately, and the file is emailed to you as well as stored in your dashboard.
Frequently asked questions
- What data can realistically be pulled off a docket?
- Supplier name, docket number, delivery date and the line-item table: product code, description, quantity, unit, unit price and line total where the docket shows them.
- Does the docket need to be a digital PDF?
- No. Scans and phone photos work as long as the text is readable to the eye. There is no requirement for an underlying text layer.
- What happens to signatures, stamps and terms?
- They are treated as page furniture and left out of the data sheet, so your rows are line items only.
- Can I extract from several dockets at once?
- Upload them one after another; each returns its own workbook, and the columns are consistent so the sheets stack cleanly into one master file.
- How accurate is it?
- Printed dockets come back near-perfect. Faint carbon copies and heavy handwriting are worth spot-checking against the original before you post anything to your accounts.
Stop re-typing paperwork
Check accuracy on your own test document free within 7 days. No card needed, no automatic charge.
Get started