Charter
You send the candidate, the change list and the risk you care about most. We return scope, date and price.
QA partner/Release assurance/Est. 2022
We sign the build off, so nobody else has to.
Integrity Test is a small London QA practice working at the point where a mistake gets expensive. We run regression packs, hold release gates and write the automation that catches the same class of bug twice.
- Registered
- England & Wales
- Company
- 14302043
- Office
- Great Portland St, W1W 5PF
- Engagements
- Retainer or fixed scope
- Availability
- Two slots from December
Regression packsGate reviewsPlaywrightAccessibilityDevice wavesPayment flowsRelease notes auditRetest windows
48Gate reviews run since 2022
312Defects reproduced first pass
4hMedian report turnaround
0Rollbacks on gated releases
What we take on
Five lines of work, priced per deliverable. No tooling resale, no seats, no strategy decks.
01
Release gate reviews
A dated checklist run against the candidate build, ending in a written go or no-go with the risk named plainly.
- used before major deploys
- app-store submissions
- payment and billing changes
02
Regression packs
Suites that re-verify every defect we have found, rerun each release so fixed bugs stay fixed and nobody argues from memory.
- weekly or per release
- run by the same reviewer
- results kept as evidence
03
Automation built in your stack
Playwright and Cypress suites wired into the pipeline, with flake accounting from the first commit and a handover that actually works.
- written in your repo
- CI wired before handover
- 30-day support window
04
Exploratory sessions
Time-boxed, charter-led probing of the areas a scripted run never reaches. Findings land as tickets with reproduction steps.
- session-based test management
- charter agreed in advance
- 48 hours notice
05
Device and browser waves
Coordinated passes across handsets, browsers and assistive setups your lab does not stock, run by paid testers across the UK.
- device list from your logs
- paid per accepted session
- evidence to template
06
Post-incident review
What the test strategy missed, written without blame and with the pack updated so the gap closes.
- timeline reconstruction
- pack amendments
- 60-minute readout
How a gate review runs
Predictable on purpose. Scope agreed in writing before anyone touches the build.
Run
Scripted regression first, then targeted exploration around the change. Everything captured in one log with evidence attached.
Verdict
Go, no-go, or conditional with named blockers. Written for the person who signs, not the person who codes.
Retest
Fixes verified against the same evidence inside the window, then the pack is amended for next release.
What a verdict actually looks like
Redacted from a real review in September. The signer gets the summary, the engineers get the linked evidence.
AT-REQ-4471 · candidate 8.4.1gate: conditional
✓ 214 regression steps passed
✓ Checkout retry verified against 8.4.0 defect
✗ Blocker: address autocomplete drops county on second edit (10/10 reproductions, Android 14, Pixel 6a)
✗ Blocker: VAT line renders twice on invoices under £40
→ Ship with both blockers fixed and retested; no other conditions.
Signed: E. Lawrence · 18 Sep 2026 · 14:02
The company is independent: no reseller agreements, no referral fees, and the reviewer who signs is the one who ran the tests.
Rates
Fixed where possible. Everything below is exclusive of VAT and quoted per deliverable, not per hour.
| Engagement | Scope | Turnaround | From |
|---|---|---|---|
| Gate review, standard | Up to 5 days of scripted and exploratory work plus verdict | 3–4 days | £2,400 |
| Gate review, extended | Adds two device waves and an accessibility pass | 5 days | £3,900 |
| Regression pack | Build, run and maintain one suite, weekly | retainer | £1,150/wk |
| Automation build | Suite written in your repo, CI wired, handover | 4–8 weeks | £7,500 |
| Device wave | 50 sessions across an agreed device list | 72 hours | £1,600 |
| Retest | Verification of fixed blockers, same evidence tree | 24 hours | included |
Where we fit
We are not a replacement for a delivery team's own testing. We are the independent line that a delivery team cannot give itself: someone with a different incentive who can say no to a release date.
The practice is deliberately small — one lead reviewer, two associates, a standing panel. Nothing routes through an account manager.
Where we don't
Proofs of concept, sub-£20k build projects, load testing at scale, and anything needing security clearance. Where a job is outside scope we say so on the first call rather than learning it mid-project.
Most clients come from two places: a release that went badly, or a procurement requirement that testing be evidenced by someone external.
For testers
Our device waves need people rather than a lab: odd handsets, regional network conditions, assistive setups. Sessions are briefed, capped at 25 minutes and paid per accepted run.
- 18 or over, UK-based
- Any device you already own — older is better
- Written brief, no experience needed
- Paid per accepted session, not per hour
- No exclusivity, take one or take none
- Evidence to template, screenshots and timings
Join the panel
Leave your details and we will send the next brief that matches your kit.
Questions we get
Do you sign NDAs?
Yes, ours or yours. Repositories and staging environments stay accessible to the named reviewer only and are revoked at the end of the engagement.
Can you work inside our tooling?
Jira, Linear, TestRail, GitHub Actions, Buildkite. Evidence lives in your tracker, not in a spreadsheet we email round.
What happens if you miss something?
Retest of any class of defect we signed off is free within 30 days, and the pack gains a permanent case for it.
Do you work with agencies?
Regularly, as the independent line under a build partner. We invoice the agency and report to the end client.
How much notice do you need?
Standard gates: five working days. Device waves: 72 hours. Emergency same-day reviews depend on the slot.
Start a conversation
One form, one reply within a working day. If we are wrong for the job we will say so on that reply.