What Is ITGC? Access, Change and Operations Controls

Each control’s owner, frequency, evidence and exception handling must be clearly defined. The four ITGC areas, auditor tests, evidence examples and common findings.
Technical summary
Technology
ITGC, COSO, COBIT
Topic
IT general controls
Level
Intermediate
In this article8 sections

Before relying on automated controls in the ERP, the independent auditor looks for the answer to one question: Who can change this system, and with what authority? IT general controls (ITGC) are the controls that ensure the reliability of the IT environment on which applications supporting financial reporting run. If ITGCs are weak, the auditor can’t rely on automated controls, has to do more manual testing, and findings end up in the management letter. Below are the four ITGC areas, sample controls, the evidence auditors ask for and weaknesses you may run into.

The four ITGC areas

Field Purpose Sample controls
Access to programs and data Only authorized people access with appropriate permissions Approval for creating users, timely closure for leavers, periodic access review, privileged access control, segregation of duties
Program changes Changes to production systems are authorized, tested and approved Change request and approval, test evidence, environment separation, restricting developers’ access to production
IT operations Systems and data are run consistently Backup and monitoring, tracking of job scheduling (batch) failures, incident management
Program development / acquisition New systems are brought into operation in a controlled way Requirements approval, data migration verification, user acceptance testing

When setting the scope, it isn’t enough to look only at the ERP application. The auditor also asks about the layers beneath it: the database (SQL Server, Oracle, HANA), the operating system and the identity infrastructure (Active Directory, Entra ID). The permission matrix in the ERP may be fine. But if three people share the sa account in the database, the application-layer control can be bypassed. Build the scope table with rows for the application, database, operating system and network access.

A good control definition

To test a control, the auditor needs the following information:

  • Control description: What is done? (“ERP user permissions are reviewed by process owners every three months.”)
  • Owner: Who does it?
  • Frequency: Daily, weekly, quarterly, event-driven.
  • Type: Manual, automated or IT-dependent manual.
  • Evidence: Which record shows the control was performed?
  • Exception handling: What’s done when a deviation is found, and how is it recorded?

A mistake you may come across in the control matrix is a description written more ambitiously than the real practice. A control written as “monthly” in the matrix but done every three months becomes an exception directly in the audit. Don’t write a frequency you can’t keep up. A quarterly control that runs regularly always gives a better result than one written as monthly that only half works.

The control owner must also be a person or a role, not a department. If it says “IT department”, nobody is responsible.

How does the auditor test?

The auditor first assesses the control’s design (right person, right frequency, right scope); then tests operation by selecting a sample from within the period. The sample size depends on the control’s frequency: one sample is requested for an annual control, many more for a daily control. Automated controls are usually verified through a configuration review and a one-off test; if change management is effective, this result can be accepted as valid for the whole period.

That last sentence explains why ITGC matters. Testing an automated control (e.g. three-way matching, a credit limit check) once and relying on it all year is only possible if it’s proven that the configuration can’t be changed without approval. When change management breaks down, the one-off test of the automated control loses its value too.

The auditor also asks about “completeness and accuracy”: does the list you used in the review really contain all users? A list exported to Excel and filtered can’t answer this. When pulling the report, keep the query you ran or a screenshot, the total record count and the date.

Example: access review evidence

  1. A dated user–role list taken from the system (a report whose completeness and accuracy can be shown).
  2. The process owner’s signed or electronically approved review, with the permissions to be removed marked.
  3. System records of the removals.
  4. The date the review was completed and an explanation of deviations.

There’s a practical trap here. If IT does the review and only gets the process owner to sign it, the auditor will question whether the control was really performed by the process owner. The list should have a “keep / remove / change” decision and a justification for each user. A single signature on a list hundreds of rows long with no changes marked is a control on paper, but a rubber stamp in practice.

Program changes: what does the evidence look like?

In change management testing, the auditor usually samples from two directions. The first is from the change record system (ticket list): the approval, testing and production deployment record of the selected change are looked for. The second is from the system: a change made in production is selected from the database or application log, and the auditor asks whether there’s a corresponding request. The second direction catches the real weakness. If a change without a ticket is found, the control isn’t considered effective.

  • Request: Who asked for it, and what’s the business justification?
  • Approval: Who approved the change, and when? Is the approver different from the developer?
  • Testing: Which scenarios were tried in the test environment, and is there user acceptance?
  • Deployment: Who moved it to production? If the person deploying isn’t the developer, segregation of duties is achieved.
  • Emergency change: Is there a record of retrospective approval and review?

If the emergency change procedure isn’t written down, the team effectively disables the control in every emergency. Write the procedure and set a deadline for retrospective approval.

The most common ITGC findings

  • Leavers’ accounts being closed late.
  • Developers or key users with broad permissions in production.
  • Shared admin accounts and unmonitored privileged access.
  • Production changes without approval or test evidence.
  • Inability to prove that the list used in the review was taken from the system.

The root cause of the first finding is usually procedural, not technical: the leaver notification from HR to IT doesn’t arrive, or arrives late. The solution is a reconciliation control that compares the HR leavers list with the active account list every month. Also verify separately that an account disabled in Active Directory is closed in the ERP, VPN and cloud applications too. Don’t assume closing it centrally covers everything; check each system.

How ITGC relates to ISO 27001 and SOC 2

Most ITGC controls overlap with the access, change, backup and logging controls in ISO/IEC 27001 Annex A; SOC 2’s security criteria test similar controls. A single control matrix and evidence structure can answer all three needs.

In practice, this means adding three reference columns to each control row: the ITGC area, the ISO Annex A control and, if needed, the SOC 2 criterion. The same access review record is used in all three audits. The evidence is produced once.

Frequently asked questions

We use a cloud ERP; is ITGC still needed?

Yes. Even if the infrastructure belongs to the provider, user access, configuration changes and interfaces are the organization’s responsibility. The provider’s independent audit reports (e.g. SOC 1/SOC 2) are assessed separately.

What if segregation of duties can’t be achieved?

In small teams, compensating controls (e.g. logging privileged actions and having them reviewed by someone independent) are designed and documented.

Do ITGC findings affect the financial statements?

They don’t mean there’s a direct error; however, they reduce the auditor’s reliance on automated controls and may require additional testing. Significant deficiencies are reported to management.

To build your ITGC control matrix and pre-test before the audit, take a look at our IT General Controls (ITGC) and Audit Preparation service; for permission issues, our Access Rights and User Account Audit service.

Support

Contact us for support or a discovery session on this topic.

Contact form
[email protected]

Eretek Compliance

We work with you on compliance with ISO 27001, IATF 16949, TISAX and similar standards, covering risk assessment, control design, documentation and audit preparation.

Knowledge Center

Other articles