ADP Testing Explained: How the 401(k) Actual Deferral Percentage Test Works
Learn how ADP testing compares elective deferrals for HCEs and NHCEs, what the passing limits mean, and how sponsors can prepare.
ADP testing compares elective deferral behavior between highly compensated employees (HCEs) and nonhighly compensated employees (NHCEs).
ADP Testing Explained: How the 401(k) Actual Deferral Percentage Test Works matters because annual compliance work is only as reliable as the employee, payroll, ownership, and plan-document data behind it. This guide gives plan sponsors an operational way to understand the question, ask better questions of their administrator, and document decisions without treating a general article as individualized tax, legal, actuarial, investment, or compliance advice.
At a glance
- Core question: What data and classification rules control this result?
- Focus: Reconcile the plan document, payroll, and participant file before acting.
- Reading path: Define the rule, compare the inputs, then build a documented workflow.
Define the issue
The ADP test uses each eligible participant’s elective deferrals, including pre-tax and Roth deferrals but excluding catch-up contributions, divided by testing compensation. [1]
For a sponsor, the practical definition starts with the plan document. It identifies the plan year, eligibility rules, compensation definition, contribution sources, and administrative elections that shape the calculation. A service provider can calculate from a file, but the sponsor remains responsible for supplying accurate operational facts and noticing when a file does not match the plan’s terms.
That distinction is especially important after a business change. A new payroll system, acquisition, owner change, merger, rehire population, or revised match formula can alter the information that should reach the annual review. The best time to resolve an ambiguity is when the underlying records are still easy to trace—not after a final report creates a surprise.
How the mechanism works
Begin with a complete eligibility and compensation file. For every eligible employee, calculate an actual deferral ratio, then average the ratios separately for NHCEs and HCEs. Employees who could defer but chose zero are part of the eligible NHCE group. [1]
Think of the process as a chain: source records determine participant classification; classification and plan terms determine the testing population; payroll and account records provide amounts; and the resulting calculation informs a sponsor decision. A break at any link can make a result unreliable. That is why a documented input review is a control, not administrative busywork.
Compare the important inputs
| Input | Question to ask | Why it matters |
|---|---|---|
| Plan document | Which definitions and elections apply? | It sets the permitted calculation and correction path. |
| Participant file | Who belongs in the population? | Omitted or misclassified people can change the outcome. |
| Payroll/account data | Which amounts and adjustments belong? | The numerator and denominator must follow plan terms. |
The HCE average must be no more than the greater of 125% of the NHCE average, or the lesser of 200% of that average and the NHCE average plus two percentage points. [1]
Questions for the annual file review
Before approving a report, ask whether the census includes every person required by the document, whether late-year hires and terminated participants were handled consistently, and whether the compensation period matches the stated rule. Ask the administrator to identify assumptions rather than hiding them inside a calculation. A short written exception log makes it easier to distinguish a genuine design issue from a data-cleanup task.
Also separate information gathering from remedy selection. The right response to an incomplete roster is to complete and validate the roster; the right response to a failed calculation may involve a document-driven correction analysis. Conflating those steps can create unnecessary participant disruption and make the audit trail harder to explain later.
Implement a sponsor workflow
Before testing, reconcile payroll codes to the plan document, confirm ownership and family-attribution information, and preserve the file used for the calculation. If the result fails, the document and correction period govern the response. [1]
- Assign ownership. Name a sponsor-side contact for HR, payroll, finance, ownership, and the recordkeeper file.
- Freeze and reconcile inputs. Save the original extracts, compare them to the document, and log exceptions.
- Review the result. Ask for the population, formulas, assumptions, and any adjustments—not just a pass/fail status.
- Document the decision. Retain approvals, correction calculations, deposits or distributions, and participant communications where applicable.
- Set scope: define the requirement, scope, timing, and decision owner.
- Check evidence: check documents, reports, controls, and open exceptions.
- Assign action: assign the approved action, deadline, and accountable owner.
- Revisit: revisit evidence, changes, and unresolved exposure.
Practical review table
| Review point | Evidence to retain | Escalate when |
|---|---|---|
| Population | Eligibility and classification roster | Ownership, termination, or related-company facts conflict |
| Amounts | Payroll/account reconciliation and code map | Compensation or source treatment is unclear |
| Result | Administrator report and sponsor sign-off | The document, method, or correction route is uncertain |
Review and improve
Use a calendar that identifies data cutoffs, an independent review, and an owner for missing demographic or compensation data. A safe harbor design can avoid ADP and ACP testing when applicable requirements are satisfied. [1]
Use the post-cycle review to convert exceptions into controls. For example, an ownership-information gap can become a quarterly questionnaire; a contribution-code problem can become a payroll change-control test; and a late eligibility feed can become a checklist with a named deadline. Controls should be simple enough that the responsible team can actually perform and evidence them.
Close the loop with a short sponsor meeting after the administrator completes the work. Review what changed from the initial census, who approved each exception, and whether the result triggered a funding, distribution, notice, or design discussion. Store the final report alongside the source extracts and approval notes. That package gives next year’s team a starting point and helps prevent informal knowledge from disappearing when roles change.
For recurring issues, identify a leading indicator rather than waiting for an annual surprise. Examples include a monthly list of new owners and related entities, a payroll-code change report, an eligibility reconciliation, or a balance-adjustment queue. The measure should lead to a specific person reviewing a specific file by a specific date.
Editorial disclaimer: This article is educational and does not provide legal, tax, actuarial, investment, or compliance advice. Retirement plan outcomes depend on the plan document, facts, timing, and applicable law. Work with qualified plan, legal, tax, and payroll advisers before implementing a correction, contribution, or design change.
Takeaway
The practical takeaway: treat adp testing as a disciplined data-and-document process. Start with the governing terms, validate the population and amounts, preserve the calculation trail, and obtain professional review when facts or correction choices are uncertain.
References
- [1] Internal Revenue Service. ADP Testing Explained supporting guidance. https://www.irs.gov/retirement-plans/401k-plan-fix-it-guide-the-plan-failed-the-401k-adp-and-acp-nondiscrimination-tests
Discuss your plan with Yang's Pension Consulting
Decision pathway
Use this page to frame the decision, collect the plan documents and operating facts that matter, and then confirm the facts with the appropriate plan, tax, actuarial, legal, or compliance professional.
- Define: identify the plan terms, workforce facts, and timing involved.
- Compare: separate operational choices from questions that require professional analysis.
- Implement: document the approved approach and align payroll, providers, and communications.
- Review: revisit the process when plan design, workforce, or rules change.
Try a related planning aid: Testing Readiness Checker.
Continue learning
ADP Testing: Treat Data Quality as a Plan Decision
The Actual Deferral Percentage test compares average salary-deferral percentages for highly compensated employees and nonhighly compensated employees in a traditional 401(k) plan. The practical sponsor decision is whether operations, participation patterns, and plan design are likely to support that comparison without limiting highly compensated employees or requiring corrective contributions. Before year-end, sponsors can review participation, deferral rates, and employee classifications to identify pressure points. A failed test is not merely a reporting issue; it can require timely corrective action to preserve qualified-plan treatment.
Records and controls to review
Retain the plan’s definition of compensation, eligibility rules, highly compensated employee determinations, ownership and family-attribution information, payroll extracts, deferral elections, testing method election, and final ADP workpapers. Document any reruns caused by corrected classifications or compensation data. If a correction is required, preserve the calculation, earnings methodology, distribution or contribution evidence, participant notices, and tax reporting. A documented pre-test review with the recordkeeper or third-party administrator can help confirm that the data used for the test matches the plan document.
Before changing operations
Use the written plan and the current service agreement as the starting point. Compare them with the payroll configuration, provider report, and participant-facing materials that will actually be used. If the records conflict, pause the change, identify the controlling record, assign an exception owner, and document the resolution before the next processing cycle.
Test a normal example and a realistic edge case before release. Record what was verified, who verified it, the action taken, and the date for the next reconciliation. This gives the sponsor and its providers a repeatable record rather than a conclusion reconstructed after an exception occurs.
For each change, identify the event that triggers the work, the data source used, the person who approves the result, and the evidence retained after completion. A short exception log should also state when an item is escalated and when a corrected result has been independently reconciled. This control-oriented approach helps expose gaps between a decision and the daily administration that carries it out.
When a provider or internal team supplies information, retain the source version, confirm the effective date, and record unresolved assumptions before action. A concise review note should connect the operational conclusion to the governing document, the supporting report, and the responsible reviewer so future corrections can start from evidence rather than recollection.
Implementation checkpoint
- Confirm: identify the governing provision and effective date.
- Reconcile: compare payroll, provider, and participant records.
- Assign: document who owns exceptions and follow-up.
- Retain: preserve approvals, reports, communications, and completion evidence.
Important boundary: ADP results depend on the plan’s actual terms, testing method, and accurate participant data; broad percentage rules are not a substitute for a completed test. Follow the governing document and applicable correction deadlines, and have the administrator or qualified ERISA counsel validate classifications, calculations, and any correction before implementation. Compliance boundary: apply the governing plan documents and obtain qualified legal, tax, actuarial, or benefits advice before implementation.
Reference
- Official source: https://irs.gov/retirement-plans/401k-plan-fix-it-guide-the-plan-failed-the-401k-adp-and-acp-nondiscrimination-tests