To stop hiring the wrong people, you replace gut feel with a repeatable process: define what success looks like before you post, screen for fit on a short call, use an application that shows how someone communicates, then run one real-work assessment instead of five interviews. The work sample is the part most people skip, and it is the part that actually predicts performance.

Key takeaways

  • Most bad hires come from a weak process, not weak judgment. Resumes and unstructured interviews are among the worst predictors of how someone will perform.
  • If you cannot describe what “great” looks like at 90 days, you are not ready to post the job.
  • One sharp, real-work assessment beats a five-round interview gauntlet, and it respects the candidate's time.
  • Watch people work live, including how they use AI. In 2026 that is a core skill, not a bonus.
  • Speed wins. The best candidates have options, and they disappear when you stall.

Why do good business owners keep hiring the wrong people?

You have been here before. The interview felt great. They were sharp, likeable, said all the right things. You hired them. And three months later you are quietly wondering what happened.

Here is the thing nobody tells you: that is almost never a judgment problem. It is a process problem.

When you hire on a good conversation and a clean resume, you are betting on the two weakest signals there are. Decades of hiring research back this up. A landmark review of selection methods found that the most popular tools (unstructured interviews, years of experience, and credentials) are among the weakest predictors of job performance, while the strongest predictors (real work samples and structured interviews) are used by a minority of companies, according to the research on what actually predicts job performance.

So no, you are not bad at reading people. You have just been using the part of the process that lies to you and skipping the part that tells the truth.

Let me walk you through the process I actually use. Five steps. It is simple, and it is hard to fool.

Step 1: Define the role before you post anything

Before you write a single word of a job post, answer one question: what does success in this role look like at 90 days?

Not the tasks. The outcomes. “Manages my calendar” is a task. “I never touch my own scheduling again and I stop missing follow-ups” is an outcome. Write three or four of those and you have a scorecard. Everything else in your process gets measured against it.

This is the step everyone skips, and it is the reason most hiring goes sideways. If you do not know what you are hiring for, you will hire the person who interviews best, not the person who performs best. A clear target turns a vague gut decision into a real comparison.

Step 2: Screen for fit on a short discovery call

The first call has exactly one job: figure out if this person is aligned, and if so, which role they actually fit.

Keep it tight. Cover their relevant experience, what they think success looks like in the role (you will learn a lot from the answer), and money.

On money, here is an operator move that saves you every time: let the candidate name their number first. Do not lead with yours. If they push you for one, give a range, never a single figure. Whoever anchors first gives up information, and in a screening call that should be them. Salaries and pay, by the way, should always be set in USD so there are no currency surprises down the line.

If they are not aligned, you just saved both of you hours. If they are, you move them forward.

Step 3: Use an application that shows how someone communicates

A resume tells you what someone did. It tells you almost nothing about how they think or how they come across. So your application needs to capture both.

Ask for a tailored resume, yes. But also ask for a short intro video, a few words on who they are outside of work, and how they describe themselves in their own voice. On HireSA, that profile is built in.

That video is doing quiet, heavy lifting. In about 60 seconds you get a real read on clarity, warmth, and communication, long before you have spent an hour on a call. For remote roles where English and presence matter, that is the cheapest, fastest signal you will ever get.

Step 4: Run one sharp, real-work assessment

This is the heart of the whole process, so slow down here.

The goal is not to build an exam. It is to watch someone do a small, real version of the actual job. Work samples are job-specific by definition, which is exactly why they predict performance so well when interviews do not. One well-built assessment beats a five-round gauntlet, and using real work keeps the candidate's burden low (which keeps your best candidates from walking away).

Two rules:

Test the job, not trivia. If the role writes client emails, have them write a client email. If it runs ad accounts, have them walk you through a real account. Abstract puzzles tell you who is good at puzzles.

Make them show you, live. The single best part of any assessment is a screen-share where the candidate works in real time. You see how they actually operate, how they think on their feet, and (this matters more every month) how they use AI in their workflow. In 2026, “show me how you use AI to do this faster” is one of the most revealing questions you can ask.

Then shape the assessment to the role:

  • Client-facing roles: a short, structured role-play of a real situation they would face.
  • Execution roles: a live walkthrough of real work they have actually done, in their own words.
  • Analytical roles: a small diagnostic case built from real numbers.
  • Creative roles: a judgment check on a handful of real options, where you are testing taste and reasoning, not output volume.

One assessment, built right, will tell you more than four conversations ever could.

Step 5: Decide fast and make the offer

Speed is not the opposite of a good process. It is part of one.

The first-responder advantage is real, and it is not just for sales. The candidate who impressed you also impressed two other people this week, and the company that moves first usually wins them. If your assessment was sharp, you do not need a week to “sleep on it.” You have the data. Score them against the role definition from step one and make the call.

A slow process does not make you more thorough. It just hands your best candidate to someone faster.

What is different about hiring in South Africa specifically?

If you are hiring South African talent (and you should be looking, the value is hard to beat), two small pieces of due diligence belong in your process.

Confirm their setup. The load-shedding era that defined South Africa a few years ago has eased dramatically. The country passed 300 consecutive days without load-shedding in early 2026, and Eskom is projecting a stable, load-shedding-free winter, per Eskom's own reporting. So this is no longer crisis-proofing. It is just smart remote hiring. Ask about backup power (an inverter, solar, or a generator) and reliable fiber. A two-minute question now prevents a dropped client call later.

Lean toward the strongest English pools. South Africa ranks 13th in the world for English proficiency, in the same “Very High” band as the Netherlands and Germany, according to the 2025 EF English Proficiency Index. The strongest pools for clear, near-neutral English tend to be in and around Cape Town and Johannesburg. You will find excellent talent across the country, but if accent neutrality is mission-critical for a client-facing role, those are the cities to weight toward.

Stop guessing. Start hiring on evidence.

You are not bad at hiring. You have just been making one of the most important decisions in your business on the two signals that lie the most.

Fix the process and the people fix themselves. Define the outcome. Screen for fit. Watch them communicate. See them do the real work. Decide fast.

That is the whole game, and the talent to run it through is already waiting.

Browse pre-vetted South African talent on HireSA, free

Frequently asked questions

How many interview rounds are too many?

More than three is usually a sign your process is doing the work your assessment should be doing. A tight version is one screening call, one structured application, and one real-work assessment. If you need a fourth touch, make it the offer conversation.

Should I pay a candidate for their time on an assessment?

If the task takes more than an hour or produces something you could actually use, pay for it. A short, role-real exercise is fair to ask for free. A full project is not. Keeping the burden low is part of respecting good candidates.

How do I assess someone's English without making it awkward?

You do not test it. You observe it. A short intro video in their application and a normal conversation on the screening call tell you everything. If communication is clear and easy in both, you have your answer.

What if I am torn between two final candidates?

Go back to the role definition you wrote in step one and score both against it, not against each other. The one who better fits the actual outcomes wins. If they are truly even, hire the faster learner over the more experienced one.

Do I still need a process if I am only hiring one person?

Especially then. A single bad hire is more expensive for a small team than a large one. The process is not bureaucracy. It is how you make one high-stakes decision with less guessing.