How to Build a Diverse Hiring Process in Canada

Where bias enters the hiring process

Bias is not always intentional. It enters at four predictable stages. At the job description stage, language like “dynamic,” “rockstar,” “ninja,” and “young and energetic team” signals culture-fit proxies that correlate with age, race, and gender. Listing fifteen requirements when five are genuinely necessary discourages qualified applicants — research from McKinsey and others consistently shows women apply only when they meet close to 100% of listed requirements, while men apply at 60%.

At the resume screening stage, names that read as foreign-sounding receive significantly fewer callbacks than identical resumes with anglicized names in controlled Canadian studies. Address screening (suburbs vs. downtown, specific postal codes) can also correlate with race and socioeconomic status.

At the interview stage, unstructured interviews are dominated by confirmation bias and similarity bias — we tend to rate candidates who are similar to ourselves more highly, independently of their qualifications. Finally, at the offer stage, differential salary negotiation behavior by gender and background means identical candidates can receive substantially different offers.

None of these require bad intent. All of them are correctable with process changes. For guidance on the job description specifically, see our companion post on writing inclusive job postings for Canadian employers.

Blind resume screening

Blind resume screening removes identifying information — name, address, graduation year, and sometimes educational institution — before shortlisting begins. Multiple controlled studies, including a well-cited Canadian audit study by Oreopoulos and Dechief, found that resumes with English-sounding names received significantly more callbacks than identical resumes with South Asian or Chinese names. Blind screening closes that gap.

Implementation does not require expensive software. A simple process: have one person strip identifying fields from submitted resumes and save them with a candidate number before passing them to the hiring manager. The evaluator sees the same qualifications without the name-triggered pattern matching.

Blind screening is most effective at the initial shortlist stage. By the interview stage, you are assessing the whole person — the goal there is structured questions and scoring, not anonymization.

Structured interviews and scoring rubrics

A structured interview means every candidate for the same role is asked the same questions in the same order, and answers are scored against a predefined rubric before the next candidate is interviewed. Research consistently shows structured interviews predict job performance better than unstructured ones — and they dramatically reduce similarity bias.

Build the rubric from the job requirements, not from a general sense of “culture fit.” For a customer service role, define what a 1, 3, and 5 answer looks like for the question “Describe a time you resolved a difficult customer situation.” A 5 is specific, includes what the candidate did, what the outcome was, and shows ownership. A 1 is vague or describes someone else solving the problem.

Score each answer immediately after hearing it, before the next question. Waiting until the end of the interview lets the overall impression of the candidate contaminate individual question scores.

Diverse interview panels

A hiring panel with diverse members — different genders, backgrounds, and functional perspectives — produces more balanced assessments than a solo interviewer or a homogeneous panel. Each panel member brings different signal detection: a panel member from a non-traditional educational path may recognize non-traditional but relevant experience that a single hiring manager misses.

For small businesses where forming a panel of three is impractical, a two-person panel — the hiring manager and one other team member — is a meaningful improvement over solo interviews. Brief the panel member in advance on the role requirements and the structured question set so they are evaluating consistently.

Panel members should score candidates independently before discussing. Group deliberations without prior individual scoring are dominated by the most senior or most vocal person in the room.

Employment Equity Act and Ontario Human Rights Code

The federal Employment Equity Actapplies to federally regulated private-sector employers with 100 or more employees (banks, telecoms, interprovincial transportation, broadcasters). These employers must collect workforce data on four designated groups (women, Indigenous peoples, persons with disabilities, and visible minorities), analyze representation gaps, and submit annual equity reports to the federal government. The Act requires employers to take positive steps to correct underrepresentation — this includes targeted outreach, modified screening criteria, and accommodation.

The Ontario Human Rights Codeapplies to all employers in Ontario regardless of size. The Code prohibits discrimination in hiring on 17 protected grounds including race, gender, age, disability, religion, national origin, and sexual orientation. Critically, the Code also permits “special programs” — employers may legally target recruitment and hiring toward underrepresented groups if the program is documented as an employment equity initiative. This distinction matters: you cannot ask a candidate their religion for a standard hire, but you can run a targeted recruitment program for candidates from a specific underrepresented community.

For any special program or equity initiative, document the program in writing before implementing it, including the designated group, the rationale, and the scope. The Ontario Human Rights Commission publishes guidance on establishing compliant special programs. For related guidance on job posting language, see writing inclusive job postings.

Measuring diversity outcomes

You cannot improve what you don't measure. Three metrics to start with: applicant funnel data (what percentage of applicants from each demographic group advance at each stage), offer acceptance rate by group(a low acceptance rate among a specific group often signals a compensation or culture signal problem), and first-year retention by demographic(if one group leaves at a higher rate, the selection process may be mismatched with what the role actually requires).

Collecting demographic data from candidates requires voluntary self-identification and must comply with privacy legislation (PIPEDA federally, or Ontario's FIPPA for public sector). The collection must be separated from the hiring decision file. Consult your legal counsel before implementing demographic tracking to ensure your collection process is legally sound. For a broader view of retention measurement, see our guide on why employee turnover costs Canadian SMBs more than they expect.

Frequently asked questions