GL coding assigns a transaction to the right place in the general ledger: the account, the cost center, and often a project, internal order, or location on top. Every invoice needs it before it can post, and the quality of the coding decides whether the financial statements describe the business or a blurry version of it.
Coding looks clerical. It behaves like judgment, which is why it resists the automation that swallowed the rest of invoice processing.
Every invoice needs a home in the ledger
An invoice that cannot be coded cannot post, and an invoice that posts to the wrong home distorts every report built on top. Coding sits between capture and payment as a quiet gate, and on non-PO invoices, where no order supplies the answer, it becomes the main event.
The code string carries more than an account
A modern coding block stacks dimensions: natural account, cost center, department, project, sometimes a joint-venture or asset reference. Each dimension answers a different reporting question, and each one can be wrong independently. The combinations run into the thousands, which is why "just pick from the list" was never a real instruction.
Good coding is precedent, and precedent is context
Experienced coders rarely reason from the chart of accounts. They remember how this supplier, this kind of spend, this plant was coded last time, and they copy it. The skill is a memory of precedent, which lives in people and in years of history spread across systems. That is exactly the kind of knowledge that walks out the door with the person who holds it.
Miscoding is quiet and expensive
A miscoded invoice fails no check and raises no flag. It surfaces weeks later as a budget variance nobody can explain, a reclass at month-end close, or an audit question. The rework lands on finance at the worst time, and the numbers leadership steered by in between were wrong. Coding and approval exceptions covers how the failures present in AP.
Code from precedent autonomously, and reserve people for new cases
Since good coding is precedent applied to context, a system that reads the history can apply it: code the invoices that match established patterns automatically, and route the genuinely novel ones to a person whose decision becomes the next precedent. Automated GL coding compares the approaches.
Fragment codes invoices from your own coding history, read in place across your systems. See how it works or request a demo.