Skip to main content
Template · FMEA Quality Risk Analysis

A working FMEA for software, not a form.
Ten columns. Every risk. Every release.

The Sumatra FMEA is a Failure Mode and Effects Analysis adapted from aerospace and automotive practice for software quality risk. It converts stakeholder judgment into a defensible, numerically prioritized register you can defend to any audience.

Columns
10
Variants
Basic + Advanced
Methodology
FMEA

A prioritized risk register does one thing spreadsheets rarely do: it survives its first change request. The FMEA is the shape that makes that survival possible.

Key Takeaways

Four things to remember.

01

Severity × Priority × Likelihood = RPN

The Risk Priority Number is the numerical shorthand for the risk. 1 (catastrophic) to 5 (cosmetic) on each axis, multiplied together.

02

Categorize before you enumerate

Functionality, Load, Usability, Compatibility, Reliability, Security, Maintainability — pick the categories first, then populate risks under each.

03

Recommended action belongs in the register

Extensive, Balanced, Opportunity, Report-Bug-and-Move-On. The register tells you how to test each risk, not just that it exists.

04

Trace to requirements

Every risk links back to at least one requirement or user story. No link means the risk is either invented or the requirements are incomplete.

Why this exists

What this template is for.

The FMEA originated in 1940s aerospace engineering and migrated into automotive and medical device quality programs. We adapted it for software because it does something unusual: it forces stakeholders to reason numerically about risks they would otherwise argue about qualitatively.

This template ships with SpeedyWriter (a fictional document management system) as a worked example. Replace the system name, stakeholders, and risks with your own — the column structure is what matters.

The columns

What each field means.

Risk ID Number

Hierarchical identifier. Top-level category gets "1", "2", etc. Each risk under the category gets "1.001", "1.002", ... so you can sort, filter, and trace without losing structure.

Example: 1.005 (for the fifth risk under Functionality)

Quality Risk Category

The ISO/IEC 25010-style category the risk belongs to. Filled in only on the header row for each category, left blank on individual risks inside that category.

Example: Functionality, Load/Capacity/Volume, Reliability, Security, Usability

Failure Mode / Quality Risk / Effect

One-line description of what could fail and how the failure would present. Focus on the observable effect on the user, not the internal cause.

Example: Check-in of new document to DMS fails.

Severity

How bad is this failure for users, on a 1–5 scale. 1 = catastrophic / safety / data loss. 5 = cosmetic / minor.

Example: 1 = catastrophic; 5 = cosmetic

Priority

How important is this risk to the business, independent of severity, on a 1–5 scale. A rarely-used feature can be low priority even if a failure there would be severe.

Example: 1 = must-fix; 5 = nice-to-have

Likelihood

How likely this failure is to occur, on a 1–5 scale. 1 = will definitely happen without mitigation; 5 = extremely unlikely.

Example: 1 = certain; 5 = negligible

Risk Priority Number (RPN)

The product of Severity × Priority × Likelihood. Lower RPN = higher risk. Sort the register by RPN ascending to see what to test first.

Example: 1 × 1 × 3 = 3 (very high); 5 × 5 × 5 = 125 (negligible)

Recommended Action

How to mitigate the risk in testing. Extensive, Balanced, or Opportunity testing; Report-and-Move-On for known-deferred; Rerun-Regression for existing coverage.

Example: Extensive testing; Opportunity testing; Rerun entire R3.0 test set

Who / Which Phase

Who owns the test (Dev, Test, User-Cert, N/A) and at which phase it applies (Unit, Component, Integration, System). Multiple can be listed.

Example: Dev/UC-Test/IS (Dev unit tests + User Certification + Integration-System test)

Requirement Tracing

Link(s) to the requirement(s) this risk covers. Every risk should trace; an untraced risk is a sign that either the risk is invented or requirements are missing.

Example: 3.1 (ties to Requirement 3.1 in the spec)

Live preview

What it looks like populated.

Excerpt from the Basic Sumatra FMEA workbook, Test Dev sheet.

Risk IDCategoryFailure ModeSevPriLikeRPNRecommended ActionWho / PhaseReq
1.0FunctionalityFailures causing specific features not to work
1.001Regression of existing SpeedyWriter features.1133Rerun entire R3.0 test set.Test/IS3.0
1.002Can't cancel incomplete actions using cancel or back.23424Opportunity testing.N/A
1.005Check-in of new document to DMS fails.2124Extensive testing.Dev/UC-Test/IS3.1
2.0Load, Capacity, VolumeFailures in scaling to expected peak concurrent usage
2.001System fails at or before 25 concurrent users.1133Extensive testing.Test/S1.1
2.004System disallows 32,767 or fewer user accounts.1339Balanced testing.Test/S1.3

How to use it

8 steps, in order.

  1. 1

    Convene the FMEA workshop with every stakeholder who owns quality: product, engineering, support, security, operations. One meeting per category is fine.

  2. 2

    Fill in the Risk ID and Quality Risk Category header rows first, one per category you will cover.

  3. 3

    Under each category, enumerate failure modes in the Failure Mode / Quality Risk / Effect column. One row per distinct failure. Do not dedupe aggressively at this stage.

  4. 4

    Score Severity, Priority, and Likelihood as a group. When the group disagrees, capture the disagreement in a comment — do not paper over it.

  5. 5

    Compute the RPN (the workbook does this for you via a formula). Sort the register by RPN ascending and review the top of the list with the group.

  6. 6

    Fill in Recommended Action for every row. Tests that will never run (Opportunity, Report-and-Move-On) are still logged here so the judgment is explicit.

  7. 7

    Trace each risk to a requirement. Mark rows with no link for a follow-up requirements review, and handle the incipient-bugs step of the Quality Risk Analysis Process.

  8. 8

    Check the finalized workbook into configuration management and require change requests to alter it afterwards.

Methodology

The thinking behind it.

The FMEA scale is counter-intuitive: lower numbers mean more severe, higher priority, more likely. This is deliberate — it lets you multiply into an RPN where lower means "do this first". If your organization finds it confusing, invert it in display but keep the multiplication convention internally.

The Basic variant uses one sheet. The Advanced variant uses four sheets (Initial, Estimate, Planning, Test Dev) — each snapshots the FMEA at a different phase of the project so you can see how risk understanding evolves over time.

When in doubt, score higher (lower number). An over-scored risk costs a few extra tests. An under-scored risk costs a production incident.

Take it with you

Download the piece you just read.

We keep this library free. All we ask is that you tell us who you are, so we know who to follow up with if we release an updated version. One-time form, this browser remembers you after that.

Need a QA program to back this up in your organization?

If a checklist is not enough and you want help applying it to a live engagement, we can have a call this week.

Related reading

Articles, talks, guides, and case studies tagged for the same audience.