Sample Manual QA & Usability Report
A format-and-method sample for the paid Manual QA & Usability Audit. The observations below are real things I saw during a two-minute first-pass on a public website, but the site is deliberately not named: this is not client work, I have no permission to publish a named critique, and no production impact is claimed. It exists so you can judge the structure and honesty of the report before engaging.
What this document is
The paid service delivers a short, structured report for one web app or job board: what I tested, what broke or felt wrong, how to reproduce it, severity, and a prioritized fix list. This sample applies that exact structure to a public site I opened in Chromium on 2026-11-21. The site is anonymized as “Subject Site”.
No screenshot, video, Safari/Firefox run, physical mobile-device test, analytics access, or client data is shown here. Those are only included in a paid engagement when access and permission exist.
1. Scope and limits, stated plainly
- Subject: one public job-board homepage, anonymized as Subject Site.
- Question: what does a first-time job seeker notice in the first two minutes, and which core journeys look inaccurate or confusing?
- Environment: Chromium desktop in BrowserOS, 2026-11-21. Responsive emulation is available in a paid engagement; it is not real-device testing.
- Checks: first-impression pass, browse/search/filter controls, listing-card data consistency, external links, and layout/text encoding.
- Not claimed: Safari, Firefox, physical iPhone/Android, screen recordings, formal UX-audit credentials, production metrics, or client results.
2. Severity rubric used in the report
| Severity | Meaning | Example in this method |
|---|---|---|
| Critical | Blocks a core task for most users, with no reasonable workaround. | Search returns an error for every query. |
| Major | Misleads a user on important data or blocks a meaningful subset of a core task. | Location/region label contradicts itself on a listing. |
| Minor | Confusing or untidy, but the user can still complete the task. | Duplicated count text on a card. |
| Cosmetic | Visual polish only; no task impact. | Inconsistent spacing or a garbled apostrophe. |
3. Worked findings
F1 — Duplicated hidden-count presentation on job cards. Severity: Minor.
- Where: multiple listing cards on the Subject Site results grid.
- Steps: open the homepage, let the country selector resolve, and scan the tag rows on several cards.
- Expected: one concise indicator of extra tags, for example “+2” or “and 2 more”, not both.
- Actual: cards showed a numeric chip such as “+2”, “+3”, “+4” or “+6” immediately followed by text such as “and 2 more”, “and 3 more”, etc.
- Impact: redundant information makes cards harder to scan; it does not block applying for a job.
- Recommendation: keep one form of the count. If the chip is interactive, drop the trailing text; if the text is needed for accessibility, make the chip’s accessible name carry it once.
F2 — Region/location label contradicts itself. Severity: Major.
- Where: one duplicated listing for the same role on the Subject Site.
- Steps: browse the results list and compare cards with the same job title under different location labels.
- Expected: the named region and the country/place agree, or the card explains why a country is grouped under that region.
- Actual: the same role appeared once under an Asia-Pacific location and again under a European country still labelled with the Asia-Pacific region. The helper text repeated the mismatch.
- Impact: a job seeker filtering by region may see a listing that does not belong to that region, reducing trust in the board’s core data.
- Recommendation: validate region assignment at import time; when a posting names a region but lists a mismatched country, either correct it or surface an explicit “check posting” state rather than presenting both as fact.
F3 — Garbled apostrophe encoding in listing summaries. Severity: Cosmetic.
- Where: job-summary text on at least one card.
- Expected: a normal apostrophe in plain English text.
- Actual: the summary contained mojibake-like characters where an apostrophe should be.
- Impact: minor readability and trust issue; no task blocking.
- Recommendation: normalize source text to UTF-8 at ingestion and add an encoding regression check.
4. Copyable bug-report template
ID: F1
Title: Duplicated hidden-count presentation on job cards
Severity: Minor
Environment: Chromium desktop / page URL / date
Preconditions: country selector resolved; results grid visible
Steps:
1. Open the page.
2. Scan the tag row on a card with extra tags.
Expected: one count indicator, not two.
Actual: "+N" chip followed by "and N more".
Evidence: screenshot or DOM excerpt (included only when permitted)
Impact: harder scanning; no core-task block.
Recommendation: keep one count representation.
Owner/status: unassigned / open
5. Deliverable format in a paid engagement
A paid audit returns: (1) the agreed scope and environment, (2) the severity rubric, (3) findings with steps, expected/actual, evidence where permitted, impact and recommendation, (4) a prioritized fix list, and (5) one revision after delivery. I do not claim real-device or multi-browser coverage unless that access is provided, and I do not claim a formal UX-audit portfolio I do not have.
Service and pricing: Manual QA & Usability Audit on Contra — $150 / one surface. Back to services: Contract Engineering Services.