What is unstructured data?
Quick answer
Unstructured data is information that isn't arranged in set fields, such as emails, PDFs, scanned invoices, chat logs and call recordings. Structured data is the opposite: rows and columns with a name for every field, like a spreadsheet. Much office automation is turning the first into the second.
Last updated
Take the free module →Module 2 of the free AI That Reads Your Documents course, about 7 minutes. No account needed.
Updated · By Robert Breen
Why it matters for a small business
Software is good at structured data. A spreadsheet can add up the Total column or flag every invoice over $1,000 in a second. It can't do that with a pile of PDF invoices, because the total sits in a different place on each one and is just more text. That is why so much admin time goes into retyping: a person is the bridge from unstructured to structured.
AI and document tools now build that bridge. OCR turns a scan into text, and document parsing or AI extraction pulls named fields out of that text. Between the two sits semi-structured data, such as JSON or an email with a fixed template, which has labels but not a strict table. The habit to keep: check the structured result against the original before you trust it.
In a real lesson: How AI Reads a PDF: OCR, Document Parsing and Field Extraction
In How AI Reads a PDF, Kestrel Design Studio has two vendor invoices. The digital PDF already holds text; the phone scan is just a picture, so OCR finds and recognizes the characters first, and makes two mistakes on the way: a zero read as the letter O and a total with its comma and period swapped. Then the lesson's point lands: text is not data yet. Sam needs five fields from each invoice, found either by rules or by AI extraction.
Docparser: Pull Invoice Data from PDFs does the full trip. Jordan uploads three vendor PDFs, the Invoices template finds Invoice Number and Totals, a Text Fixed Position rule adds Vendor Name, and the download is a CSV with one row per line item. The PDFs were unstructured; the CSV is structured, and every vendor, number, date and total matched.
Try this lesson free or read the step-by-step guide.
Common confusions
Unstructured vs messy
Unstructured means no set fields. A spreadsheet can be structured and still messy, with duplicates or blanks that need cleaning.
Unstructured data vs structured output
Unstructured data is the input, like an email. Structured output is when you make an AI reply in a fixed format, such as JSON with named fields.
Tips
- Before automating, list the exact fields you need from each document.
- Leave a field blank when it isn't in the document, rather than letting a tool guess.
- Spot-check a few extracted rows against the originals, especially numbers from scans.
Related terms
More Data, Sheets and Gmail terms
Where you use it: free lessons
- How AI Reads a PDF: OCR, Document Parsing and Field Extraction (Stepthrough, 7 min)
- Docparser: Pull Invoice Data from PDFs into a Spreadsheet (Docparser, 10 min)
- Olum: See If ChatGPT and Google AI Recommend Your Business (Olum, 12 min)
Frequently asked questions
- Is a PDF structured or unstructured?
- Usually unstructured. Even a digital PDF holds text laid out on a page, not named fields. A fillable PDF form is closer to structured.
- Is an Excel file always structured data?
- Mostly, if each column holds one kind of value. A sheet with notes pasted into random cells is structured in form but not in practice.
- Can AI search unstructured data without converting it?
- Yes. RAG chatbots search chunks of documents by meaning. You only need fields and columns when you want to count, sort or total things.