Skip to main content
Talk · Rex Black, Inc. webinar

Risk-based testing. What it is, why it works, and a CA pilot that proves it.

Out of the infinite cloud of possible tests, a test team runs a finite number. Risk-based testing decides which ones — in what order, with how much effort, and how to cut when the schedule compresses. This webinar walks through the method, the four benefits, and a six-step pilot at CA that shipped 92 risk items, an adjusted RPN histogram, and a re-allocated test effort that changed how the project shipped.

Slides
19
Pilot risk items
92
RPN range
1–25

Abstract

You are going to test less than 100% of the system.

For any complex system, an infinite number of tests exist. You do not have forever. Out of that infinite cloud, the test team must select a finite number. Measured as a percentage of what you could test, your coverage is always zero. So you have to be very, very selective — and very, very smart — about what you run.

Risk-based testing is that selection mechanism. Quality risk is the possibility that the product fails to deliver one or more of its key quality attributes, endangering the outcomes you actually care about. Risk-based testing uses an analysis of those quality risks to prioritize tests, allocate effort, and — when the schedule compresses — drop tests in reverse risk order. It keeps you honest about trade-offs you are already making implicitly.

This webinar walks through the four benefits, then through a six-step pilot at CA that made it real. The pilot finalized 92 risk items, discovered a common clumping problem in the initial RPN distribution, adjusted the impact scale to fix it, and was later published in Better Software magazine. The numbers and the sequence below come from that engagement.

Release when the risk of delay balances the risk of dissatisfaction.

Rex Black, Inc.

Outline

What the talk covers, in order.

01

Setting the course — infinite tests, finite schedule

Every test plan is an admission that you cannot test everything. The question is whether the selection is deliberate or accidental. Risk-based testing makes it deliberate: every test earns its place by how much risk it reduces.

02

What risk-based testing actually is

Risk-based testing uses an analysis of quality risks to prioritize tests and allocate testing effort. It involves key business and technical stakeholders so that the focus and sequence of testing line up with what quality means to the people who have to live with the product. It also means managing project risks — events that endanger completion of the project itself — alongside the quality risks.

03

The four benefits

Running tests in risk order finds defects in severity order. Allocating effort by risk minimizes residual quality risk at release. Measuring by risk tells you the residual quality risk during execution so you can make smart release decisions. And if the schedule compresses, dropping tests in reverse risk order gives you back time with the least possible increase in quality risk.

  • Find the scary stuff first.
  • Pick the right tests out of the infinite cloud.
  • Release when risk of delay balances risk of dissatisfaction.
  • Give up tests you worry about the least, not tests you happen to have left.
04

Does it increase test work?

No — if anything, the long-run effect is to decrease effort. Risk-based testing drives more efficient testing overall. After the initial quality risk analysis, only periodic updates and traceability maintenance are required.

05

CA pilot — the six activities

CA, a Rex Black, Inc. client, ran a pilot (subsequently followed by two more). The pilot was a six-activity sequence: train the stakeholders, hold a risk analysis session, analyze and refine the results, align testing with the risks, guide the project by risk, and assess the benefits. Each step below is from that pilot.

06

Training

Participants and stakeholders went through a one-day training covering the principles of risk-based testing, the categories of quality risks, how to analyze quality risks, how to align testing with risk levels, how to document the analysis, how to monitor quality risks during execution, and how to report risk-based results. Training was presentation, discussion, and a worked exercise — not lecture.

07

Quality risk analysis session

Two sub-sessions. In the first, participants identified risk items: main quality-risk categories went on three whiteboards; participants wrote risk items on sticky notes and posted them under the relevant categories. The sub-session lasted three hours and produced over 100 risk items — plus 11 project risks and 3 other issues as by-products. In the second sub-session, the team rated likelihood and impact for each item; they identified and merged duplicates along the way. At the end of the session, the team had 92 risk items and had rated 40% of them. The test manager finalized the remaining ratings with the participants afterward.

08

Analyzing and refining — the RPN histogram

With ratings complete, the team calculated an RPN (risk priority number) for each item. On a five-point scale each for likelihood and impact, RPN ranges from 1 (most risky) to 25 (least risky). A common problem: clumping. Clumping shows up when the team skews impact toward worst-case outcomes, or when the scale has poorly defined distinctions. You can check by plotting a histogram — and if you see clumping, you refine the scale and re-rate.

09

Aligning testing with quality risks

With the analysis settled, the test manager mapped risk items to specifications (the Product Requirements Specification and the Detail Design Specification), mapped them to test cases, re-allocated testing effort based on risk level, and prioritized the test cases within each risk band. An effort-allocation table was customized for the application so that every risk item received at least cursory testing. Mapping risk items to specifications lets you regenerate the analysis quickly when specs change.

10

Guiding the project based on risk

With finalized RPNs, effort allocations, and traceability, the test manager prioritized execution by RPN. This was a significant change from the previous, purely requirements-based approach. Tests had previously been assigned by tester expertise, which led to availability bottlenecks when a key person was out. Under risk-based testing, those bottlenecks went away — high-priority tests ran early, low-priority tests ran later. Bug management also changed: severity rankings were augmented with risk prioritization, and bug-fix effort focused on the high-risk defects.

11

Conclusion — benefits and key points

The pilot delivered what risk-based testing typically delivers: intelligent effort allocation within constraints, priority-order bug discovery that optimized fix-time windows, flexible handling of reductions in time or resources, and optimized quality within the constraints the project actually had. Two key points for anyone adopting this: include business users — and possibly customers — in the analysis, and start the risk analysis early in the lifecycle. Bottom line: risk-based testing lets a team deliver increasingly sophisticated products under tightening constraints by prioritizing testing and balancing quality against the other project priorities.

Key takeaways

6 things to remember.

01

Coverage is always zero

You cannot test everything. The honest question is how you are selecting. Risk-based testing makes selection explicit, argued, and reviewable.

02

Four benefits, not one

Find severity-first. Minimize residual risk. Know the residual risk in real time. Cut in reverse risk order. You lose all four if you skip the analysis.

03

Run risk analysis as a workshop

Whiteboards, sticky notes, three hours for identification, a second session for rating. Business plus technical plus test. Do not do it solo at your desk.

04

Plot the RPN histogram

If the distribution clumps, your scale is wrong — not your product. Adjust the definitions, re-rate, move on.

05

Map risks to specs and tests

Traceability cuts both ways: when specs change, you know which risks to re-rate; when risks change, you know which test cases to re-prioritize. Miss the mapping and you lose half the benefit.

06

Priority beats expertise assignment

Assigning tests by tester expertise creates bottlenecks when that expertise is unavailable. Priority-by-RPN routes around those bottlenecks and still finishes the scary stuff first.

Worked examples

One bug. Eight drafts.

RPN clumping — what you'll see in real data

The initial ratings from the CA pilot. Likelihood looked skewed but was reasonable given a mature, stable codebase. Impact was the real problem: half the items were rated 2, which compressed the RPNs into a narrow band. Adjusting the distinction between impact 2 and impact 3 fixed it.

Likelihood (reasonable)

Rating 1 · 5 items

Rating 2 · 9 items

Rating 3 · 25 items

Rating 4 · 39 items

Rating 5 · 26 items

Initial read: skewed toward the high end. Actual: the product was mature and well-established, with a maintainable codebase and experienced developers. The distribution reflected reality — no adjustment needed.

Impact (clumped — fix this)

Rating 1 · 10 items

Rating 2 · 52 items

Rating 3 · 32 items

Rating 4 · 8 items

Rating 5 · 2 items

Problem: more than half the items sat at rating 2, which flattened the RPN distribution and made prioritization nearly impossible. Fix: redraw the line between impact 2 and impact 3 with sharper criteria, re-rate the affected items, plot again.

Effort allocation table — how to spend the testing you have

The allocation table from the CA pilot, customized so that every risk item — even the lowest — received at least cursory coverage. Your numbers will differ; the shape of the table will not.

RPN 1–12 · Extensive

Run a large number of tests that are both broad and deep. Exercise combinations and variations of interesting conditions. This is where the test team spends the bulk of its attention.

RPN 13–16 · Broad

Run a medium number of tests that exercise many different interesting conditions. Fewer combinations; still wide coverage of the high-traffic paths.

RPN 17–20 · Cursory

Run a small number of tests that sample the most interesting conditions. Enough to catch obvious regressions. Not enough to reduce residual risk much further.

RPN 21–25 · Opportunity

Leverage other tests or activities to run a test or two of an interesting condition — but only if it involves a very small investment of time and effort, and only if the opportunity presents itself. If neither condition is met, skip it. That is the entire point of prioritizing by risk.

Closing

Include business users — and potentially customers — in the risk analysis. The people who live with the product know which failures are unacceptable. The test team knows which ones are likely. You need both voices in the room, and you need them there early in the lifecycle, not after the schedule compresses.

Risk-based testing lets a team deliver increasingly sophisticated, complex products within tightening constraints by prioritizing testing and balancing quality with the other priorities the project is managing. It is not a shortcut. It is an honest accounting of what testing can and cannot do — and that honesty is what makes the release decision defensible.

More for this audience

Articles, guides, and case studies tagged for the same readers.

Want this talk delivered in-house?

Rex Black, Inc. delivers every talk on this site as a live workshop, a keynote, or a conference session. Tailored to your stack, your team, and your timeline.