The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Some links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
In Tricentis’s 2025 Quality Transformation Report, 42% of surveyed organizations said poor software quality costs them at least $1 million a year. That is a striking warning, not an audited tally of industry losses: the research was a vendor-commissioned survey, and the reported costs are respondents’ estimates.
What the survey found
Tricentis says Censuswide conducted the survey in March 2025, gathering responses from 2,750 people across 10 countries and five industry verticals. Participants included technology executives, DevOps and QA leaders, IT practitioners, and software developers. The company’s announcement reports that 42% of respondents said poor software quality costs their organization at least $1 million annually.
There is a discrepancy in Tricentis’s own published figures: its companion blog gives 40%, rather than 42%, for the $1 million threshold. The formal announcement and report are the basis for the 42% figure here. Neither figure should be read as the share of all companies worldwide that actually lose that amount.
The findings describe perceptions and estimates from a selected survey sample. They do not establish a verified total loss, an average loss per company, or a causal link between a particular release practice and a specific dollar amount. The report page offers the full report through a form, while the public summary does not provide enough detail to independently assess sampling method, response rate, weighting, question wording, confidence intervals, or country-level sample sizes.
#1 Best Overall
The speed-versus-quality trade-off
The survey points to a tension in software delivery. Tricentis reports that 45% of respondents prioritized improving delivery speed, compared with 13% who prioritized enhancing software quality. Meanwhile, 63% said their organizations release code changes without completing all necessary testing. Among the reasons given, 46% cited pressure to accelerate release cycles and 40% cited the accidental release of untested code.
“Without completing all necessary testing” does not mean that no testing took place. It may mean that a planned test was skipped, unfinished, or did not cover a change before release. Still, the result suggests that deadlines and release processes can allow known test gaps to reach production—not merely that teams lack automation.
That distinction matters. A faster release can be valuable, but only if teams understand what risk they are accepting. A slow, flaky test suite can impede delivery; a release process that quietly bypasses required checks can expose customers and the business to failures. The survey does not prove that faster releases caused the reported costs, but it does identify speed pressure as a common reason testing is incomplete.
What “software quality” can cost
Software quality includes more than whether a feature works in a basic demonstration. It can involve reliability and availability, performance, security and privacy, accessibility, compatibility, data integrity, maintainability, user experience, and compliance. A quality gap may show up as a visible bug—or as a failed integration, incorrect transaction, outage, security exposure, or costly delay.
Organizations can think about the economics in five categories:
- Prevention: design reviews, code review, training, static analysis, and investments in test environments and quality engineering.
- Appraisal: manual and automated testing, security and performance checks, monitoring, and release validation.
- Internal failure: rework, hotfixes, rollbacks, failed builds, delayed launches, and time spent diagnosing problems before customers are affected.
- External failure: outages, support demands, refunds, service-level credits, lost sales, regulatory exposure, litigation, or reputational harm.
- Opportunity cost: engineers diverted from planned work and features delayed while teams repair problems.
The available survey summaries do not break down the reported $1 million-plus figure across these categories. These are ways costs can arise, not a claim that every respondent counted each one or that the study measured their individual contribution.
Why organizations may release under-tested code
Tricentis respondents identified weak communication or feedback loops between developers and testers (33%) and a disconnect between leadership and development teams (28%) among the leading quality obstacles. The company’s blog also identifies ongoing maintenance and technical debt as a major obstacle (34%) and says roughly one-quarter cited budget constraints.
Those findings point toward operating conditions as well as technology. Testing can become a bottleneck when regression runs are slow, tests are unreliable, environments or test data are unavailable, or manual checks cannot keep pace with releases. Legacy systems may make changes harder to isolate. If no one owns a test, a failure, or the decision to waive a check, an untested change can slip through by mistake.
Technical debt can plausibly make maintenance and testing more difficult, but the survey summary does not show that debt caused the reported costs. Similarly, weak developer–QA communication is a reported obstacle, not proof that fixing communication alone would prevent failures.
Why financial services stands out—with a caveat
Tricentis identifies financial services as the industry with the highest reported exposure. Its blog says 45% of financial-services respondents reported costs above $5 million a year. That is a survey result, not proof that financial services is objectively the costliest sector in every market or that the same share applies to all financial institutions.
Transaction volumes, uptime expectations, complex integrations, and regulatory obligations are plausible reasons failures could have serious consequences in financial services. They are context for interpreting the result, not causal explanations demonstrated by the survey.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsAI may help with repetitive work, but the survey does not show a return
More than 80% of respondents expected productivity gains from delegating repetitive work to agentic AI; the formal announcement says 82% were excited about AI agents taking on monotonous tasks in development and delivery. Tricentis also reported that almost 90% believed their organizations could quantify generative-AI return on investment in the software development lifecycle.
These are expectations and confidence levels, not evidence that AI has already reduced defect rates, testing costs, or delivery times. AI features can serve different purposes: generating test cases, maintaining tests as interfaces change, executing tests, or helping triage failures. Those capabilities should be assessed separately, and any claimed financial return should be measured in the organization’s own workflows.
AI-generated tests can encode the implementation’s assumptions rather than the intended business behavior. Teams should review generated tests, connect them to requirements or critical user journeys, and retain them in a format engineers can understand and maintain. AI can assist quality work; it does not remove the need for human judgment, security review, or release accountability.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What leaders can do before buying another tool
A practical response starts with the failure mode, not a product category. First establish where quality problems cost time or money: escaped defects, incidents, rework, rollbacks, customer support, and delayed work. Then identify which customer journeys and systems carry the greatest business or safety risk.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →- Set a baseline. Track serious production incidents, defect-remediation time, hotfixes, rollbacks, and engineering hours spent on rework. Use actual internal records where possible rather than relying on a broad industry estimate.
- Make release risk explicit. Define which checks must pass for a change, who can approve an exception, and what evidence accompanies a release. Distinguish a deliberate, documented risk decision from an accidental omission.
- Improve feedback speed and reliability. Put fast unit tests and static checks close to each code change. Run API and integration tests in continuous integration, and target end-to-end tests at high-risk workflows. Investigate flaky tests and assign owners rather than letting unreliable results become background noise.
- Prioritize meaningful coverage. Focus on critical business rules, permissions, failure paths, data integrity, and important integrations—not only a line-coverage percentage. A large suite of shallow tests can miss serious risks.
- Close the production feedback loop. Use monitoring, controlled rollouts, and rollback plans to detect problems pre-release tests miss. Feed incidents back into test cases and design practices.
- Review delivery and quality together. A team should not be rewarded for deployment frequency alone if avoidable failures rise alongside it.
Useful measures include escaped defects by severity, change-failure rate, time to detect and recover, the share of deployments completing required tests, test pass and flakiness rates, time waiting for environments, hotfix and rollback frequency, and availability of critical customer journeys. Treat these measures as signals for improving the system, not as a reason to punish teams or hide failures.
Best Value
Choose testing tools for the bottleneck
Automation runs predefined checks; continuous testing integrates checks throughout delivery; quality engineering makes quality a shared responsibility from design through operations. None guarantees that the right behaviors are covered. Risk-based testing helps teams direct effort where a failure would matter most, while production observability can expose gaps that pre-release tests missed.
Different tools solve different problems. Code-first unit and API testing may be enough for a small product team. A hosted browser or real-device service can help when device fragmentation is the constraint, but may not fit data-residency rules, private-network requirements, specialized hardware, or a stable, limited browser matrix. A test-management system can improve traceability and coordination, but cannot substitute for useful test design. A broad enterprise quality platform may suit a regulated organization that needs centralized governance and evidence, but can be excessive for a small team.
Before choosing a product, check that it supports the applications, languages, frameworks, and CI/CD system the organization actually uses. Evaluate execution speed, parallelism, flakiness, maintenance effort, environment and test-data management, reporting, security and compliance, migration effort, vendor lock-in, and total cost—including licenses, infrastructure, training, administration, and ongoing test maintenance. If AI is part of the offer, confirm that generated tests are reviewable, exportable, and maintainable.
A sensible buying sequence is to clarify ownership and release rules, measure the main bottleneck, pilot the smallest tool that addresses it, then compare its full operating cost with measurable changes in failure rates or delivery delays. The survey’s dollar threshold is a reason to investigate quality economics, not automatic justification for an expensive platform.
How much weight should readers give the report?
The report has a sizeable international sample and includes both leadership and practitioner perspectives, which makes it useful for understanding reported concerns across roles. But it was commissioned and publicized by Tricentis, a company that sells software quality and testing products. Its commercial interest is relevant context, and the self-reported cost estimates are not independently audited financial records.
The strongest conclusion is limited but useful: many surveyed organizations perceive poor quality as a major expense, and incomplete testing is common in their accounts. The report does not establish that all organizations lose millions, that automation alone would recover those costs, or that AI has already delivered the anticipated gains. For any individual business, incident records, remediation time, customer impact, and the cost of controls are better evidence for deciding what to fix.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
