"Testing is just a big black hole at the end of the project. The more money we throw at it, the more it consumes."
"Testing is a value-adding activity that occurs throughout the project. By making smart test investments, we reap big rewards."
Most test managers want to work for Bonnie. The hard part is convincing Piet, and Piet's CFO.
Testing adds measurable value in four distinct ways. Quantify each separately. Sum them. You go into the budget conversation with a defensible ROI, not a request.
This talk answers:
Written for test managers and QA directors who have to defend their program to finance. Useful for any senior tester who wants to connect their day-to-day work to the business case above it.
Composite ROI in the worked example, four value categories stacked on the same testing investment.
Cost of quality = cost of conformance + cost of nonconformance.
The classic escalation:
$1 to find a bug in review · $10 in programmer testing · $100 in tester testing · $1,000 in customer usage.
Add an independent test team.
ROI = (benefit − cost) / cost. Denominator = cost of the team. Numerator = old quality cost minus new.
Add $150,000 of tools, amortized over 12 quarterly releases.
Complement manual with automation where it pays: regression · load · performance · structural API checks.
Quality costs halved vs. baseline. Customers find ~66% fewer bugs than baseline.
(Where the "Investing in Software Testing" argument ends.)
Layer static testing: testers review design and requirements specs, ask smart questions, prevent ~150 bugs from ever being built.
Static testing is the highest-leverage investment per dollar. Review-stage cost ($1/bug) versus customer-stage cost ($1,000/bug). The math moves sharply in its favor.
If we know where a bug is (even if we don't fix it) we can:
The value is real. The trick is measuring it.
Assume a call for a known bug is 15 minutes shorter than a call for an unknown one. Each bug generates 5 calls on average. Support person costs $40/hr fully loaded.
$40/hr × (15/60 hr) × 5 calls = $50 per known bug in support time alone.
Over 650 additional bugs documented in the same test cycle: $32,500 of value at zero additional testing cost.
Testing reduces cost of exposure. Structurally identical to what insurance does: a statistical mechanism for pooling risk.
Example "quality-risk premium":
Total = $26,000 of insurance value, already provided by the existing test program.
Poor project tracking is a primary cause of project failure (Jones, Estimating Software Costs).
If good testing provides half of the tracking risk-reduction benefit, it claims 10% of the project's at-risk value.
On a project with $82,500 of testing + $247,500 of development budget: 10% × $330,000 = $33,000.
Stage 1 → 350% Stage 2 → 445% Stage 3 → 627%
Industry analyst benchmarks: ~800%, same order of magnitude.
Many testers are allergic to finance. That's a career-limiting instinct.
Without a measurement, we have no defensible sense of the work's value, and in most organizations managers will not fund work with no measurable ROI.
Start the calculation. Be conservative so the number survives scrutiny. Apply Gilb's Law: even a bad metric is better than no metric, because you can iteratively improve a bad metric into a good one. You cannot improve what you don't measure.
Without a measurement, we have no solid sense of our work's value. Learning to estimate testing ROI is a critical success factor for testers.