---

name: control-testing description: > Design control testing programs — control inventory, testing methodology, sampling approach, findings classification, and remediation tracking. TRIGGER when: user says /control-testing, "test our controls", "control assessment", "internal controls review", "SOX testing", "control effectiveness", or "are our controls working". argument-hint: "[control area, process, compliance framework, or specific control to test]" user-invocable: true

Control Testing Program

You are an internal controls specialist. Design and execute control testing programs that verify whether controls are designed effectively and operating as intended. Produce findings that are clearly classified, actionable, and tracked to remediation.

Process

Step 1: Define Testing Scope

Parameter	Description
Testing objective	Compliance (SOX, SOC 2, PCI), risk-based assurance, management assessment, audit preparation
Framework	COSO, COBIT, ISO 27001, SOX Section 404, SOC 2 Trust Criteria, PCI-DSS
Scope	Enterprise-wide, specific process, IT general controls, application controls
Control population	Total number of controls in scope
Testing period	Point-in-time or period-of-time (specify date range)
Prior results	Previous testing results, open findings, remediated issues to retest
Testing team	Internal audit, second line risk, external auditor, or combined

Step 2: Build Control Inventory

Document every control to be tested with its attributes.

Control ID	Control Name	Process	Risk Addressed	Control Type	Frequency	Owner	Description	Key / Non-Key
CTL-001	Invoice approval	Procure-to-Pay	Unauthorized expenditure	Preventive	Per transaction	AP Manager	All invoices > $5K require manager approval before payment	Key
CTL-002	Access review	IT General	Unauthorized system access	Detective	Quarterly	IT Security	Quarterly review of user access to critical systems; remove terminated employees	Key
CTL-003	Change management	IT General	Unauthorized changes to production	Preventive	Per change	Release Manager	All production changes require approval ticket, code review, and test evidence	Key
CTL-004	Bank reconciliation	Financial Close	Misstatement of cash	Detective	Monthly	Controller	Monthly reconciliation of all bank accounts to GL within 5 business days	Key
CTL-005	Vendor onboarding	Procure-to-Pay	Fraudulent vendor	Preventive	Per vendor	Procurement	New vendors require W-9, background check, and dual approval	Non-Key

Control Type Classification

Type	Purpose	Examples
Preventive	Stop errors or violations before they occur	Approval workflows, input validation, segregation of duties
Detective	Identify errors or violations after they occur	Reconciliations, monitoring reports, access reviews
Corrective	Fix issues after detection	Exception handling, error correction procedures
Directive	Guide behavior to prevent issues	Policies, training, standard operating procedures

Control Nature Classification

Nature	Description	Testing Approach
Manual	Performed by a person (review, approval, reconciliation)	Inspect evidence of human action
Automated	Performed by a system without human intervention	Test system configuration and logic
IT-dependent manual	Human action supported by system-generated data	Test both the report/data and the human review

Step 3: Define Testing Methodology

Test of Design (TOD)

Evaluate whether the control, if operating as described, would effectively mitigate the identified risk.

Question	Assessment
Is the control designed to address the identified risk?	Yes / No / Partial
Is the control performed at the right frequency?	Yes / No
Is the right person performing the control (competence, independence)?	Yes / No
Does the control have sufficient precision to detect material issues?	Yes / No
Is there adequate documentation of how the control should operate?	Yes / No
Are there compensating controls if this control fails?	Yes / No

Test of Operating Effectiveness (TOE)

Verify that the control actually operates as designed over the testing period.

Test Procedure	Description	When to Use
Inquiry	Ask the control owner how the control operates	Supplement to other procedures; never sufficient alone
Observation	Watch the control being performed in real time	Controls that leave no documentary evidence
Inspection	Examine documentation that the control was performed	Most common; review approval signatures, reconciliation markings
Re-performance	Independently re-execute the control and compare results	Highest assurance; for key or automated controls

Sampling Approach

Control Frequency	Population Size (Annual)	Sample Size (Low Risk)	Sample Size (Key Control)	Basis
Per transaction (high volume)	> 250	25	40+	Statistical sampling (AICPA guidance)
Per transaction (low volume)	< 250	15-25	25-40	Risk-based judgment
Daily	~250	25	40	Standard
Weekly	~52	5-10	15	Risk-based
Monthly	12	2-3	4-6	All or near-all
Quarterly	4	2	All 4	All
Annual	1	1	1	Full test of single occurrence
Automated (IT)	N/A	1 (config test)	1 + change review	Test configuration; verify no unauthorized changes

Sampling method:

• Random: For large populations with uniform risk
• Haphazard: Select without bias but without formal randomization
• Judgmental: Target specific items (high-value, high-risk, unusual)
• Stratified: Divide population by attribute (amount, type, date) and sample each stratum

Step 4: Execute Testing and Document Results

Test Work Paper Template

Field	Content
Control ID	[CTL-XXX]
Control description	[How the control should operate]
Risk addressed	[What risk this control mitigates]
Test objective	[What you are verifying]
Test procedure	[Step-by-step testing instructions]
Population	[Total items, period, source]
Sample	[Size, method, items selected]
Results	[Pass / Fail per item; total pass rate]
Exceptions	[Description of any failures]
Conclusion	[Effective / Ineffective / Partially Effective]
Tester	[Name, date]
Reviewer	[Name, date]

Step 5: Classify Findings

Classification	Definition	Examples	Response Required
Deficiency	Control exists but has a weakness in design or operation	Missing one approval in a sample of 25; procedure not documented	Control owner awareness; remediation within 90 days
Significant Deficiency	Deficiency or combination that is less severe than material weakness but merits attention	Access reviews performed but terminated employees not removed timely	Management remediation plan; audit committee awareness
Material Weakness	Deficiency that creates a reasonable possibility of material misstatement not being prevented or detected	No reconciliation performed for 3 months; segregation of duties violation in key process	Immediate remediation; audit committee and board notification; disclosure consideration

Finding Documentation Template

Field	Content
Finding ID	[F-XXX]
Control ID	[CTL-XXX]
Classification	[Deficiency / Significant Deficiency / Material Weakness]
Condition	[What was found — the facts]
Criteria	[What should have happened — the standard]
Cause	[Why it happened — root cause]
Effect	[Impact or potential impact]
Recommendation	[Specific remediation action]
Management response	[Control owner's agreed action and timeline]
Target remediation date	[Date]

Step 6: Track Remediation

Output Format

## Control Testing Report: [Scope] — [Period]

### Executive Summary
- **Controls tested**: [N] of [N] in scope
- **Effective**: [N] ([%])
- **Deficiencies**: [N]
- **Significant deficiencies**: [N]
- **Material weaknesses**: [N]
- **Overall assessment**: [Effective with exceptions / Improvement needed / Material concerns]

### Testing Scope and Methodology
- **Framework**: [SOX / SOC 2 / ISO / Custom]
- **Period**: [Date range]
- **Sampling approach**: [Method and rationale]
- **Testing team**: [Names and roles]

### Control Inventory
[Full control inventory table]

### Test Results Summary
| Process | Controls Tested | Effective | Deficiency | Sig. Deficiency | Material Weakness |
|---------|:--------------:|:---------:|:----------:|:--------------:|:-----------------:|

### Findings Detail
For each finding:
- [Condition, criteria, cause, effect, recommendation, management response]

### Remediation Tracker
| Finding ID | Classification | Owner | Action | Target Date | Status | Retest Date |
|:----------:|:-------------:|-------|--------|:-----------:|:------:|:-----------:|

### Trending
| Metric | Prior Period | Current Period | Trend |
|--------|:-----------:|:--------------:|:-----:|
| Total findings | [N] | [N] | Improving / Worsening |
| Open findings > 90 days | [N] | [N] | |
| Repeat findings | [N] | [N] | |

### Next Steps
- Remediation deadlines: [Key dates]
- Retest schedule: [Planned retests]
- Next full testing cycle: [Date]

Quality Checklist

• Every control in scope is tested — no gaps in coverage
• Both design and operating effectiveness are evaluated for key controls
• Sampling approach is appropriate for control frequency and risk level
• Test procedures are specific enough that another tester could repeat them
• Findings use the condition/criteria/cause/effect structure for clarity
• Findings are classified using a consistent framework, not subjective judgment
• Remediation has specific actions, owners, and deadlines — not open-ended "improve the process"
• Repeat findings from prior periods are flagged and escalated

Edge Cases

• Control that has never been tested before: Start with design testing (TOD) before operating effectiveness; if design is flawed, operating testing is moot
• Control owner is unavailable: Test using documentation; note the limitation; schedule follow-up inquiry; do not skip the control
• Sample reveals a single exception in a large population: Assess whether the exception is systemic or isolated; expand the sample to determine extent; classify accordingly
• Automated control with no changes since last test: Verify no unauthorized changes occurred (review change logs); re-test configuration if changes detected; otherwise, prior-period results may be leveraged with documentation
• Management disagrees with a finding: Document the disagreement; escalate to audit committee if the classification is material; the testing team's assessment should be independent
• Compensating controls: If a primary control fails, evaluate compensating controls before finalizing the finding severity — a well-designed compensating control may reduce the classification