Fixed Format

Fixed format, also called fixed-width or fixed-length, is a way of laying out data in which every field occupies a predetermined range of character positions rather than being marked off by commas, tabs, or other delimiters. A parser reads bytes 1 through 10 as the account number, 11 through 40 as the name, and so on, according to a record layout agreed in advance. It is one of the oldest data-exchange conventions and still moves enormous volumes of records through legacy systems.

• Positional fields - Each value lives at a known offset and width, so the layout is the schema.
• No delimiters - Records are read by column position, not by scanning for separators, which makes parsing fast but unforgiving of drift.
• Padding and alignment - Short values are padded with spaces or zeros to fill their fixed width.
• Legacy and mainframe roots - Common in COBOL copybooks, banking batch files, and government data feeds.

In real operations, fixed-format files are what modern APIs usually have to ingest and translate, not produce. Banking, insurance, and government back ends still emit fixed-width batches that get mapped into JSON payloads before they ever reach a REST or agentic consumer. It sits at the opposite end of the spectrum from a self-describing file format like JSON — rigid and compact, but dependent on out-of-band documentation, which is exactly why so much integration work is really about parsing it into something an API and an AI agent can understand.