The verdict in three sentences
For a custom business web app in London, plan 25 to 35% of the development budget for quality: manual acceptance testing and test automation combined. That looks heavy until you count incidents: at 2,000 to 20,000 GBP per production regression, avoiding three incidents pays for it. The right target is not 100% coverage but 60 to 80% on critical code (billing, permissions, calculations) with continuous integration that blocks any broken release.
What software quality costs in 2026
Most London quotes price development to the day and leave a vague "QA" line, often 5% of the total. That is where the 14 regressions suffered last year by the IT director of a Shoreditch services SME came from: no test suite, and acceptance testing done the evening before release by two busy users. Here are 2026 orders of magnitude for an app costing 80,000 to 150,000 GBP. Contract testers in London charge 450 to 650 GBP per day.
| Quality item | Share of dev budget | Example on 100,000 GBP | What you get |
|---|---|---|---|
| Manual acceptance testing (test plan + campaigns) | 10 to 15% | 10,000 to 15,000 GBP | Business scenarios validated by key users |
| Unit tests (Jest, Vitest, PHPUnit) | 6 to 10% | 6,000 to 10,000 GBP | Business rules protected on every commit |
| End-to-end tests (Playwright) | 6 to 10% | 6,000 to 10,000 GBP | 20 to 40 critical journeys replayed automatically |
| API and contract tests | 3 to 5% | 3,000 to 5,000 GBP | ERP, payment and SSO integrations monitored |
| Continuous integration (GitHub Actions, GitLab CI) | 50 to 300 GBP per month | 600 to 3,600 GBP per year | No release without green tests |
| One-off load tests (k6) | 2 to 4% | 2,000 to 4,000 GBP | Saturation threshold known before peak season |
A 100,000 GBP project therefore spends 25,000 to 35,000 GBP on quality. A vendor promising "testing included" with no dedicated line usually only tests what breaks in front of them.
Manual, automated or both: where to put the money
Manual acceptance testing remains essential to validate usability and rare business cases. Automated tests protect against regression: the bug fixed in March that returns in June because a developer changed a neighbouring function. The table below compares three strategies on a 40-screen app maintained for three years.
| Criterion (3 years) | Manual only | Automation only | Recommended mix |
|---|---|---|---|
| Initial cost | 10,000 GBP | 22,000 GBP | 28,000 GBP |
| Cost per release (2 per month) | 1,500 to 2,500 GBP | 100 to 300 GBP | 400 to 700 GBP |
| Cumulative cost over 72 releases | 118,000 to 190,000 GBP | 29,000 to 44,000 GBP | 57,000 to 78,000 GBP |
| Production regressions per year | 10 to 15 | 4 to 6 | 1 to 3 |
| Time to production | 3 to 5 days | Under 1 hour | 1 day |
| Usability validation | Yes | No | Yes |
The mix wins because it keeps humans on what they do well (judging a journey, spotting a business inconsistency) and hands repetition to the machine. Budget 3 to 5 days of test maintenance per quarter: a poorly written Playwright test quickly becomes flaky and the team ends up ignoring it.
What to demand in the quote
Four lines separate a serious quote from an optimistic one. First, a test plan delivered with the specifications, not after development. Second, a contractual coverage rate on critical modules, measured by the CI tool and visible to you. Third, a staging environment fed with anonymised data, UK GDPR compliant, for 30 to 120 GBP per month. Finally, a blocking rule: no release if a test fails, unless your project lead signs off in writing.
| Clause | Wording to request | Risk if missing |
|---|---|---|
| Coverage | 70% minimum on billing, permissions, calculations | Coverage inflated by trivial code |
| E2E journeys | Named list of at least 25 journeys | Tests on the home page only |
| Fix time | Blocker within 4 business hours, major within 2 days | Fixes pushed to the next sprint |
| Test report | Signed acceptance report per release | Disputes over delivery |
| Test ownership | Test code delivered with the source | Vendor lock-in |
Mini case study
Need a professional website?
Kolonell builds websites that attract clients, optimized for the Sénégalese market. Free quote in 2 minutes.
Oliver, IT director of a 120-employee SME in London, counted last year's 14 regressions: 9 minor ones (about 2,000 GBP each in lost time and support), 4 serious ones (8,000 GBP each including customer credits) and a two-day billing outage estimated at 20,000 GBP. Total: 70,000 GBP. For the rebuild of his app (110,000 GBP), he adds 30,000 GBP of quality work and 2,400 GBP per year of CI. If regressions drop to 2 minor ones per year, the residual cost falls to 4,000 GBP. Estimated annual saving: 66,000 GBP for a first-year investment of 32,400 GBP, a payback in under six months.
FAQ
What share of the budget should go to testing a custom web app?
Plan 25 to 35% of the development budget, manual acceptance testing included. Below 15%, production regressions become frequent from the second year of maintenance.
Should we aim for 100% code coverage?
No, effort beyond 80% is expensive for little gain. Aim for 60 to 80% on critical code and 30 to 50% elsewhere, measured on every commit.
Playwright or Cypress in 2026?
Playwright has become the default on new projects: multi-browser, parallel execution and lower CI cost. A suite of 30 journeys runs in 5 to 10 minutes.
How much does continuous integration cost an SME?
Between 50 and 300 GBP per month depending on compute minutes and team size. Free tiers rarely suffice beyond two releases per week.
Can we add tests to an existing app with none?
Yes, starting with 15 to 20 Playwright journeys on critical functions, for 8,000 to 15,000 GBP. Unit tests are then added as bugs get fixed.
Let's scope your project. We price your QA and automation strategy, from 8,000 GBP to secure an existing app to 35,000 GBP for a new build, with a 4 to 6 week plan. Detailed quote within 48 h. WhatsApp +221 77 596 93 33.
Mohamed Bah
Fondateur, Kolonell
Passionate about digital and entrepreneurship in Africa, Mohamed has been helping Sénégalese businesses with their digital transformation since 2020. Founder of Kolonell, he believes every SME deserves a professional and accessible online présence.
