The funny thing about accessibility testing is that it looks like a software job until you realize the first real tool is patience. Many websites break for people who never touch a mouse, and nobody notices because the page still looks fine to the rest of us.
That gap is where the work lives. A scanner can shout about missing labels and contrast problems. It cannot tell you whether a checkout flow becomes a maze when read aloud by NVDA, or if a keyboard user can even reach the payment button without getting trapped in some useless pop-up. This skill matters because it is one of the few global services you can start with an ordinary laptop, free software, and enough stubbornness to test the same page three different ways.
What makes this worth learning
Accessibility testing is a real service, not a side quest for designers with extra time. Businesses need someone to prove their site can be used by people who rely on assistive tech. The gap between “looks clean” and “works for everyone” is large enough to build a career in.
The market is bigger than local compliance panic. E-commerce stores want fewer abandoned carts. Agencies want cleaner handover reports. Procurement teams want proof that a supplier did not build a site that quietly excludes users. International clients pay attention because accessibility rules travel better than most local qualifications. A solid tester in Joburg or Polokwane can sell the same skill to Cape Town, London, Toronto, or Berlin if the report is sharp enough and the notes are useful.
Automated tools alone are not the job. If you only run a browser extension and paste the score into a PDF, you are not doing accessibility testing. You are decorating a problem.
What you need to start
You do not need capital in the romantic startup sense. You need a laptop that can run a browser, a screen reader, and a few free tools without crashing.
The starter stack is simple.
- `NVDA` on Windows, free.
- `VoiceOver` on a Mac, already built in.
- Browser developer tools in Chrome or Edge, free.
- `Lighthouse`, free inside Chrome DevTools.
- A keyboard, because the mouse is a lie here.
If you already own a computer, your initial cost is close to zero. If you have to buy a machine, that changes the math fast. A basic second-hand laptop can be far cheaper than a full course, but data, electricity, and repairs still exist. Load shedding will not politely pause because your audit is due. If your connection is unstable, you learn to work offline, save evidence locally, and upload when the signal returns.
Basic testing ability can come together in four to eight weeks if you work consistently. Real competence takes longer. Knowing where a button is visually is one thing. Knowing how `WCAG` works, how to write remediation notes that a developer can act on, and how to avoid making fake compliance claims is the actual trade.
How to test a site properly
Start with a site that matters in the real world, not a toy page. An e-commerce checkout is perfect because it forces the ugly questions to show up fast.
Pick three South African websites and test each one with the same routine.
First, use the keyboard only. Tab through the page. See whether focus moves in a sensible order. Check whether you can reach menus, forms, filters, and the checkout button without touching the mouse. If focus disappears, gets trapped, or jumps around like it is late for a taxi, that is a failure.
Second, turn on a screen reader. On Windows, use `NVDA`. On a Mac, use `VoiceOver`. Listen to the page. Do the headings make sense? Are buttons announced properly? Are form fields labeled? Can you understand what the page is asking without seeing it?
Third, run `Lighthouse` and any other automated checker you trust. Use it to catch the obvious technical problems. Do not stop there. The machine can tell you that a label is missing. It cannot tell you whether the page sounds like a confused intern wrote it at 11:47 p.m.
Fourth, inspect the code in browser developer tools. Look at form markup, focus states, ARIA usage, image alt text, and whether the structure supports the way the interface is presented. A page can be visually elegant and still be a mess for assistive tech.
Your first assignment should be a short report, not a dissertation. For each issue, include the problem, who it affects, the evidence, and the suggested fix. That format matters because clients do not pay for dramatic language. They pay for notes a developer can use on Monday morning.
What to put in a report
A useful accessibility report is boring in the best way. It should be readable by a designer, a developer, and a manager who does not want to hear about ARIA for the rest of their life.
Each issue should answer four things:
- What is broken.
- Who gets blocked or slowed down.
- How you proved it.
- What should change.
If a checkout field has no visible label, say so. If a screen reader announces “edit blank” where the user needs a phone number, write that down. If focus gets lost inside a modal, capture the exact step where it happened. If a contrast issue makes text hard to read, point to the element and the color pairing.
Do not claim legal compliance because a site passed one automated scan. That is how people embarrass themselves in front of procurement teams. Compliance is a bigger question. If you want to talk about it, you need a proper process and the current `WCAG` standard, not vibes and a browser extension.
How long before anyone pays you
The fastest honest path is to learn enough to audit a small website cleanly, not “learn accessibility” in the abstract.
A practical timeline looks like this.
- Week 1 to 2: learn keyboard navigation, screen reader basics, and how Lighthouse reports work.
- Week 3 to 4: test live sites and write short issue reports.
- Week 5 to 8: repeat the process until your notes stop sounding like guesswork.
That gets you to basic service level. If you want to be taken seriously by agencies or foreign clients, you need deeper `WCAG` knowledge and cleaner remediation notes. That part takes longer because the standards are detailed and the edge cases are where the money hides.
The useful benchmark is not “I can spot problems.” Plenty of people can do that after one afternoon on YouTube. The benchmark is “I can explain the failure, prove it, and tell a developer what to fix.”
Where the money comes from
There are a few ways to sell this skill.
Freelance audits on `Upwork`, `Fiverr`, or `Contra` can work once your sample reports are sharp enough to show. Payment is usually in dollars, euros, or pounds, which matters more than any productivity hack on earth. A small audit that earns $100 is not life-changing everywhere, but converted into rand, it starts looking like a different conversation.
Agencies also need overflow help. They might not hire a full-time tester, but they will pay for a clean audit before a launch. Web studios building e-commerce sites, SaaS products, and municipal or education platforms often need this work because someone, somewhere, will ask for it later.
There is also a local angle. South African e-commerce, logistics, solar and battery installation businesses, and data center-adjacent digital services all need usable websites and portals. The more they move online, the more they trip over accessibility problems they never budgeted for.
If you can produce a calm, specific report, you are not selling “web advice.” You are selling risk reduction.
How to check this is legit
Plenty of paid courses and certificates are real. Plenty are just expensive confidence theater.
A legit accessibility course should teach manual keyboard testing, screen reader use, the current `WCAG` version, and how to write marked reports that show what was tested and what failed. If the course only talks about automated scores, walk away. If it claims you will be “job ready” after a weekend without making you use `NVDA` or `VoiceOver`, that is a sales page, not training.
Before paying, check three things.
- Does it show actual sample reports?
- Does it cover manual testing, not just scanning?
- Does it reference the current `WCAG` version, not a stale framework from years ago?
The same goes for certification badges. If the only proof is a shiny PDF and a landing page full of stock photos, treat it like marketing.
Why this beats chasing the usual entry jobs
A lot of first-job advice in this country still acts like the only options are office admin, retail, call centers, teaching, or learning to smile through an interview for a role that pays too little. Accessibility testing is different. It is technical enough to be scarce, practical enough to start from home, and global enough to escape the local ceiling.
That ceiling matters. If your income is tied only to local wages, you are negotiating inside a small room. If you can sell a useful service in foreign currency from a bedroom with stable electricity and a decent laptop, the room gets bigger fast.
The catch is that the work rewards accuracy, not enthusiasm. You need to see what a site feels like when a mouse is removed, when audio becomes the interface, and when a form is being used by someone who does not have your eyes or your hands. That is not glamorous. It is more useful than glamorous, which is usually better business.
The internet has already created the market. The missing part is people who can prove where it fails.
