Section 04
What actually predicts whether a hire works out in Ontario, how to run a process you can defend, and how to onboard so good people stay.
Hiring well is mostly a measurement problem. A good chat, a confident handshake, a gut feeling: most of the signals employers lean on barely predict who will actually do the job. And the few signals that do predict it are the ones most small employers skip.
For two decades the research had a settled hierarchy of what predicts job performance. In 2022 a major reanalysis corrected it, and a structured interview is now the single strongest predictor. That means the same job-related questions, the same scoring, every candidate. The free-flowing “let’s just chat and see if I like them” interview that most hiring runs on is among the weakest tools in common use. No single method predicts well enough on its own, so adding one more job-relevant read beats relying on one conversation. None of this is exotic or expensive. It’s mostly a matter of doing on purpose what most employers do by feel.
It’s worth being just as careful about the scare numbers. A hire that doesn’t work out carries a real cost, but most of the figures that saturate HR blogs have no real basis. The famous “a bad hire costs 30% of first-year earnings, per the U.S. Department of Labor” line can’t be traced to any actual Department of Labor document, and the “strong onboarding lifts retention 82%” claim is an uncited vendor figure. We don’t use either. What follows is the evidence that survives scrutiny and the process it points to, with each part linking down to the sourced note behind it.
The evidence
Peer-reviewed selection research, corrected to the current post-2022 figures, and not the inflated numbers that still circulate online.
Verified
The strongest predictor
A structured interview is the single strongest predictor of job performance, at the head of a tight top group with job-knowledge and work-sample tests. Structured means the same job-related questions, asked of every candidate and scored on a set scale. Most hiring never builds one.
Sackett et al. (2022) Journal of Applied Psychology — meta-analytic reanalysis
The selection-validity hierarchy ›Verified
The 2022 correction
Cognitive-ability testing was sold for years as the #1 predictor at around .51. A 2022 reanalysis corrected a statistical over-correction and revised it down to about .31. At that level it is no longer the stand-out signal. If you read “cognitive ability is the best predictor at .51,” you’re reading pre-2022 numbers.
Sackett et al. (2022) Journal of Applied Psychology
What the 2022 reanalysis changed ›Verified
The weakest common tool
The open-ended “let’s just chat” is an unstructured interview, and it is among the weakest and most bias-prone tools in common use. The corrected figure is about .19. Structuring the interview you already run roughly doubles its predictive value.
Sackett et al. (2022) Journal of Applied Psychology — meta-analytic reanalysis
Structured vs. unstructured interviews ›Validity coefficients are estimates with real spread rather than fixed constants. Read them as the best current evidence on the rank order of methods. General information, not legal advice.
Good practice
This is the method that turns the evidence into a repeatable process. Six stages run from selection through onboarding, and each has a place it tends to go wrong. None of it is a legal citation, and all of it is what good hiring looks like in practice. No single method predicts well enough on its own, and that shapes the whole sequence. The strongest and most defensible hires combine two job-relevant methods instead of resting on one conversation.
Before you write the ad, do a quick job analysis. Name the few things the role genuinely needs to succeed, and set them apart from the long wish-list. That short list becomes your criteria, and everything downstream is measured against it.
What predicts performance ›Where it goes wrong “Culture fit” used as an unexamined euphemism. Left undefined, it usually just means “people like us.” It can quietly drift into screening by age, background, or another protected ground. Name the specific behaviours the job actually needs instead.
Write job-related questions and a simple scoring guide before you meet anyone. Use situational questions (“what would you do if…”) or behavioural ones (“tell me about a time…”), tied to the criteria. Decide in advance what a good answer looks like.
What makes an interview structured ›Where it goes wrong Making it up in the room. Questions invented on the fly can’t be compared across candidates, so what you collect is noise, which is exactly where first-impression and similarity bias fill the gap.
Ask the same questions in the same order, and score each answer as it’s given. Where you can, use more than one trained interviewer, have them rate independently, then compare. Consistency is what turns an interview into a measurement.
Conducting a structured interview ›Where it goes wrong The unstructured “let’s just chat” interview, which is the default for most small employers and one of the weakest and most bias-prone tools there is. A warm conversation feels like insight, but mostly it measures rapport, not who can do the job.
Pair the interview with one more job-relevant method. Use a work sample or job-knowledge test for experienced hires, or a cognitive measure where you’re hiring for learning potential. Use references late, to verify facts and catch red flags rather than to rank finalists.
Why combining methods works ›Where it goes wrong The halo effect, where one strong impression (a confident talker, a shared school, a great first five minutes) colours every other judgment. Also treating a glowing reference as prediction. Candidates pick their own referees, and few referees say anything negative.
Bring the scores together, weight verified facts over impressions, make the call, and write down why. A documented, job-related decision is both a better decision and a defensible one if it’s ever questioned.
Using references and checks well ›Where it goes wrong Deciding on gut and back-filling the reasons afterward, or letting the most confident person in the room override what the process actually found.
The hire isn’t done at “yes.” Start before day one, and give real role clarity plus one early win in the first week. Then run a manager-owned check-in rhythm through the first 90 days. You’re building three things on purpose: role clarity, the confidence to do the work, and a sense of belonging.
A good first 90 days ›Where it goes wrong Sink-or-swim onboarding, where you hand the new person a laptop and hope. The manager is the part that moves whether they stay, not HR and not a checklist. In a small firm that’s an advantage, because the manager already sees them every day.
Most hiring runs on the least predictive signal there is, a conversation you can’t quite explain afterward. The fix isn’t a better gut. It’s a process.
Verified
The cheapest upgrade in hiring
Structure the interview you already run and you roughly double its predictive power. No software, no consultant, just job-related questions and a scoring guide. It’s the highest-return, lowest-cost change most employers can make.
Sackett et al. (2022) Journal of Applied Psychology
Structured interviews ›
Article Hire with Confidence Executive search Hiring a role you can’t afford to get wrong? Book a consultation A real decision
For a routine role, your own structured process is plenty. When you’re hiring a senior leader or a hard-to-reach specialist, you face a real choice in how you engage outside help. The split is not really about fees but about exclusivity, research depth, and reach into people who aren’t job-hunting.
When the role is senior or critical
When the pool is deep and speed matters
Fee conventions you’ll see quoted (a percentage of first-year pay, paid in instalments) are U.S. industry norms rather than Canadian rules. Treat them as illustrative. A Kitchener–Waterloo firm hiring its first VP is the classic retained case, while routine professional hires usually do not need one. Executive search: retained vs. contingency ›
From the blog
Article So what’s it like to be a real-life headhunter? What separates a real headhunter from a contingency recruiter, told through one VP of Finance search. Read › 
Article Team Dynamics – Hiring the Right People & Creating Effective Teams Hiring the right people, and building a team that actually works once they’re in. Read › 
Article Solving the Talent Challenge A year of roles filled across Ontario, from IT directors to warehouse staff, and what the searches had in common. Read › Straight answers
On the best current evidence, a structured interview is the single strongest predictor, with job-knowledge and work-sample tests close behind. Cognitive-ability testing was long sold as the top predictor. It was revised down in 2022 and now sits mid-pack. In practice, most hiring leans hardest on the weakest predictors, while the strongest ones are inexpensive to add.
The selection-validity hierarchy ›Yes, if they’re structured. A structured interview asks the same job-related questions of every candidate and scores them the same way. It is the strongest single predictor there is. The unstructured “let’s just chat” version is among the weakest and most bias-prone. Structuring the one you already run roughly doubles its predictive value.
Structured vs. unstructured interviews ›Ontario’s Human Rights Code bars job ads, application forms, and interview questions from sorting people by a protected ground, directly or indirectly. That catches indirect signals too, like “young, energetic team” in an ad, or availability questions that really probe family status or religion. You may ask about a protected ground only in the narrow cases the Code itself allows (a genuine job requirement). This is general information, not legal advice. If a question feels close to the line, get advice before you ask it.
Job ads, applications & interview questions ›Yes. Under Ontario’s accessibility rules you must tell applicants that accommodation is available in your recruitment process, offer it when you invite someone to a test or interview, and point a successful applicant to your accommodation policies when you make the offer. The duty resets at each stage, so posting the line once isn’t enough. It sits on top of your broader duty to accommodate under the Human Rights Code.
Accessible recruitment under AODA ›Build three things on purpose in the first months: role clarity, the confidence to do the work, and a sense of belonging. Structured onboarding, which includes preboarding, a real first week, an early win, and a manager-owned check-in rhythm, is well-evidenced to support those three, and they predict staying and performing. Be wary of the viral “82% retention” onboarding stat, though: it’s an uncited vendor figure. The honest claim is strong enough without it, and it puts the manager rather than the checklist at the centre of whether a new hire stays.
Does structured onboarding reduce turnover? ›A hire that doesn’t work out costs more than the recruiting bill, but the widely quoted figures are unreliable and the best known of them is apocryphal. The famous “30% of first-year earnings, per the U.S. Department of Labor” line can’t be traced to any real DOL document. What the research does support is that the cost rises sharply with seniority. Top performers in senior roles contribute out of all proportion, so a bad senior hire is disproportionately expensive. That’s the case for spending real care on the hires you can’t afford to get wrong, and sometimes for a retained search.
What a bad hire costs ›The knowledge base
Executive search & hiring
Maybe it’s your first senior leader, a hard-to-fill specialist, or a process that keeps producing hires that don’t work out. Any of those is worth a conversation before the offer goes out rather than after. We run structured, evidence-based searches, and we’ll tell you honestly when a role doesn’t need one.