Talent Selection Criteria: On What Basis Do You Decide?

This article is about the criteria themselves: what you measure, and how to tell a true signal from a misleading one.

1. Start with a written scorecard

For each role, write before advertising:

• Required outputs at 90 days and 12 months.
• Four to six competencies, no more, each defined by an observable behaviour rather than a general trait. "Good communication" isn't measurable; "explains a technical problem clearly to a non-specialist" is.
• A consistent rating scale for each competency.
• Questions and exercises tied to each competency specifically.
• A minimum acceptance threshold for each criterion — set before seeing candidates, not after.

The scorecard is the difference between a decision grounded in evidence and one grounded in comfort.

2. Four core criteria

Ability to do the actual work. Not what they know but what they produce. The best evidence: prior work you can examine, or a short exercise resembling the real task. Credentials and titles are relatively weak signals.

Speed of learning. In a startup the role changes within months. Ask what the candidate learned recently without a teacher, and how. Someone unable to name a single example within a year is a signal worth noting.

Independence and tolerance for ambiguity. Do they deliver without detailed instruction? Ask about a situation they entered without sufficient clarity, and what they did. "I waited for direction" is honest but ill-suited to this stage.

Fit with how you work. Not "are they like us?" but: will they thrive in an environment with this pace, this ambiguity, and this level of directness? This is a criterion of fit, not similarity — and confusing the two produces homogeneous, weak teams.

3. Stage fit

A criterion particular to startups and frequently overlooked: excellent talent in the wrong context fails.

• Someone from large companies, accustomed to teams, support, and budgets, may struggle when asked to execute personally.
• Someone who has only ever worked in very small companies may struggle when extended processes need building.

Ask directly: "In this role you'll do X yourself with no team behind you — how do you feel about that?" An honest answer here is worth more than ten interviews.

4. True versus misleading signals

True signals:
- Completed work you can examine.
- Specific examples with numbers and outcomes, not general descriptions of responsibilities.
- Acknowledging a mistake and what they learned from it.
- Deep questions about the work, the product, and the customers.
- Consistency of account across different interview stages.

Misleading signals:
- Fluency and social presence — they measure presentation skill, not work skill.
- A large company's name on the CV without examining what they actually did there.
- Similarity to you in background or interests.
- Excessive enthusiasm in the interview.
- Ideal theoretical answers to behavioural questions.

A diagnostic rule: a candidate who describes achievements as "we" throughout, without being able to specify their own contribution, is precisely the one who warrants deeper questions.

5. Reducing bias

• Standardised questions across all candidates for the same role.
• Score each competency separately immediately after each interview, rather than an overall impression at the end.
• Independent documentation before discussion — the first opinion voiced shapes the rest.
• Evidence-based decisions: require each evaluator to cite a specific piece of evidence from the interview supporting their score.
• Watch for similarity bias — the most common and the hardest to notice, because it feels like having found "the right person."

6. Red lines

Regardless of competence:
- Dishonesty about any detail, however small.
- Attributing every past failure to others with no personal accountability.
- Disparaging former colleagues or employers.
- Disrespecting others' time during the hiring process itself.

7. When you hesitate

The practical rule: hesitation is an answer. A wrong hire in a small company costs far more than waiting — in performance, morale, management time, and the cost of correction.

If the reason is a specific, addressable gap, consider a fixed-term engagement that tests the hypothesis rather than a full commitment.

Common mistakes

• Criteria not written before interviews.
• Relying on first impressions.
• Confusing culture fit with similarity.
• Ignoring stage fit.
• Measuring fluency rather than ability to deliver.
• A long competency list that dilutes the assessment.
• Neglecting the effect of the first opinion on other evaluators.
• Hiring despite hesitation out of urgency.

Checklist

• A written scorecard before advertising
• Four to six competencies defined by observable behaviour
• A minimum acceptance threshold set in advance
• A practical exercise tied to the real task
• An explicit question about stage fit
• Standardised questions across candidates
• Assessments documented before joint discussion
• Specific evidence supporting every score
• A conscious review of similarity bias
• Respect for the rule that hesitation is an answer

FAQ

What matters most at early stage?
The ability to deliver independently under ambiguity. Technical competence matters but is less scarce.

Is prior startup experience required?
Not required, but its absence calls for explicitly verifying stage fit.

How do I assess someone with no showable prior work?
With a short practical exercise resembling the real task, kept time-bounded and respectful of their effort.