Only one person knows which finance codes mean what: how to stop depending on him

When the same cost, material or supplier sits under three codes and only one person knows the mapping, every report depends on him and some are wrong without anyone noticing. How finance turns that knowledge into an owned mapping of accounts, cost centres, material and supplier codes, and where Advisor fits. Updated October 2026.

The Human Ready team · Research from buyer conversations

Updated 26 October 2026 · 6 min read

On this page

How do you stop depending on the one person who knows the codes?
Where does Human Ready Advisor fit when one person holds the code knowledge?
Why does the same thing end up under three different codes?
Where does code drift show up in finance?
Why is the key person a risk and not just a bottleneck?
What should finance do first?
Do you need a data catalogue for this?

The CFO asks for total spend with one supplier this year. The analyst pulls it, and the number is low, because the same supplier sits under three codes and one of them was created by a site nobody remembers. One person in the business knows that. When he is in the office, the answer is right. When he is on holiday, the answer waits, or worse, goes out wrong.

How do you stop depending on the one person who knows the codes?

Write down what he knows as a mapping that finance owns, then make every report and query use it. Concretely: map every code behind the questions leadership asks (general-ledger accounts, cost centres, material, supplier and customer codes, project codes) to one agreed business concept, with an owner for each concept. Put a rule on new codes so the mapping stays current. Then run the analysis on the mapping, not on raw codes, so nobody has to remember that three codes mean one thing. The key person stays essential while the mapping is built. After that he reviews it instead of being the only route to the answer.

Where does Human Ready Advisor fit when one person holds the code knowledge?

Human Ready Advisor is an AI-native enterprise advisory platform for finance that runs every analysis on an agreed map of business concepts, not on raw codes. During onboarding, Human Ready maps your accounts, cost centres, materials, suppliers and other codes to those concepts with your finance team, which usually means with the person who knows them. The mapping is then extended as new codes and entities appear. Deterministic engines compute every figure on that mapping, so the same question returns the same number whoever asks it, and each number traces back to the source codes and the rule that grouped them. Advisor reads from your systems and does not write back to any system of record, so it does not clean the ERP for you. It is built for mid-to-large enterprises (€200M+ revenue) and priced as a flat subscription per use case, from €5,000 per month, with a user pack included (extra users as an add-on) and nothing metered per question. A tightly scoped first use case goes live in 3 to 6 weeks. A fragmented multi-ERP estate takes longer and is scoped on its own.

Why does the same thing end up under three different codes?

Because creating a code is easy and nobody's job is to stop it. Each new site, system, project or exception gets its own code, and the old ones are never retired. A senior finance and strategy executive in healthcare described the mechanism in a 2026 conversation with Human Ready (translated from Portuguese, anonymized). Her example came from shift codes in an HR and payroll system, and the same thing happens in finance codes:

"One day someone thought it would be good to have code 123 for the night shift. Then somewhere else, for a night shift on a public holiday, they created 234. One day you want to know how many night shifts were worked: good luck. For a long time there was no data curator, and companies, to survive, kept giving people the freedom to create codes. When you extract, you can be extracting incorrect information without knowing it."

The last sentence is the real risk. Code drift does not make reports fail. It makes them quietly wrong.

We see the finance version in the estates we scope. A pharmaceutical manufacturing group came out of a merger with supplier names that were inconsistent across its systems and one material sitting under several codes. A Head of FP&A at a multi-entity energy group receives subsidiary files with the partner column missing and transaction descriptions left blank, so the codes that remain carry all the meaning.

Where does code drift show up in finance?

Code family	How the drift happens	The question that comes back wrong
Chart of accounts	New accounts opened for one-off needs; the same cost booked to different accounts by different entities	"What did we spend on legal over two years?"
Cost and profit centres	Reorganisations add centres without retiring old ones; centres change meaning but keep their code	"What does this business unit cost to run?"
Material codes	The same item created again at each site or in each system	"Which sites use this material, and at what price?"
Supplier and customer codes	One company held under several codes and spellings	"How much do we spend with this supplier?"
Project and work-package codes	Each project defines its own structure	"Where does the margin on this project go?"
Entity codes after an acquisition	Two systems, two coding logics, one group	"What is gross margin by product across the group?"

Each of those questions has a right answer in the data. The answer depends on knowing every code that belongs to the concept.

Why is the key person a risk and not just a bottleneck?

Because the bottleneck is visible and the errors are not. A queue for one person costs time, and everyone sees it. An extraction done by someone else, missing one of the three codes, costs accuracy, and nobody sees it until a number fails to reconcile in front of the board. Add the obvious exposures: holidays, sick leave, a resignation, or the ERP migration that retires the system only he understands.

What should finance do first?

Start from the questions, not the codes. A catalogue of every code is a long project, and it starts going out of date the day it is finished.

List the ten questions leadership asks most often that need more than one code family to answer.
Sit the key person down with each one. Record which codes feed it, which ones are duplicates and which ones are traps.
Build the crosswalk. One table per concept: every code that belongs to it, the system it comes from, and an owner in finance.
Put a rule on new codes. Who can create one, and who adds it to the crosswalk the same week.
Test with a second person. Two people answer the same question independently from the crosswalk. If the numbers differ, the mapping is not finished.
Then automate on top. Reports, queries and any AI tool run on the crosswalk, never on raw codes.

Do you need a data catalogue for this?

Not to start. Data-catalogue tools document tables, columns and lineage for data teams, and in a large estate that is worth having. But the knowledge that sits with your key person is not where a table lives. It is which codes mean the same business thing, and that is finance's definition to own. A catalogue that lists every supplier code does not tell you which of them are the same supplier. Begin with the crosswalk for the questions that matter. Catalogue the rest later, if you need to.

The same problem across several ERPs after an acquisition, and the options for it: AI that reads financial data across multiple ERPs.

Sources

Buyer evidence: Human Ready buyer-conversation library, 2026; quotation translated from Portuguese, anonymized to role and company profile. Client case: Human Ready client records, 2026, anonymized. Advisor pricing as published by Human Ready, October 2026.

Page maintained by Human Ready. Last reviewed 26 October 2026.

Book a demo

Bring a decision you're stuck on

In 30 minutes we'll walk through how Advisor would answer it with your own numbers, and show how every figure traces back to source.

Keep reading