Recruiting guides · 6 minutes read
How to compare candidates: criteria and a simple comparison matrix
By ReclyraUpdated
1. Describe the role before comparing people
Two CVs can look very different and still contain relevant experience. Do not start by comparing document design or how familiar previous employers seem. Derive your criteria from the job advert: which tasks will the person actually carry out? A comparison matrix keeps those requirements consistent across applications, giving the team a shared basis for its discussion.
Consider a customer service role at a small online retailer. The person answers written enquiries, investigates delivery problems with logistics and records refunds. The team turns those tasks into three criteria: writing clear answers, investigating problems systematically and keeping understandable records. Experience with its particular ticketing system is useful, but can be learned during onboarding.
2. Separate essential requirements from skills people can learn
An essential requirement should be a defensible prerequisite for this particular work. For each one, ask why it is needed on the first day. Could training or comparable experience from another industry be sufficient? A long list of supposedly essential requirements weakens the comparison if the team later interprets them differently for different people.
In our example, the person needs enough written German for the customer enquiries they will handle. Having held a particular job title is not essential. Experience dealing with complaints might come from retail, customer support or administration. Record acceptable alternatives before reviewing applications, so an unfamiliar career path does not attract a stricter standard.
3. Fill the matrix with evidence rather than impressions
A simple template needs five columns: criterion, expected evidence, available evidence, current status and next question. Use the same rows for each application. Initially, supported, conflicting and unclear are enough to describe the information. Supported means relevant evidence is available; it does not remove the need to consider its strength or make a professional judgement.
Example row: criterion: written customer communication. Expected evidence: an understandable example of a clear customer response. Applicant A describes daily support emails and a specific enquiry. Applicant B mentions telephone service but no written replies. A provides an initial indication; B leaves a question open. Missing information does not yet establish that someone lacks the ability.
4. Clarify gaps with focused, comparable questions
Do not request every imaginable document just in case. Ask the smallest question that will clarify a relevant gap: ‘Have you also handled customer enquiries in writing? Please describe an example without sharing confidential customer information.’ Record who owns the follow-up and when the team will review the answer alongside the application.
If you need a work sample, give participants the same task, a comparable time allowance and the same available tools. For customer service, that could be a short reply about a fictional delayed delivery. A clearly bounded task provides something more useful to discuss than a vague requirement for an ‘impressive personality’, using criteria agreed in advance.
5. Understand what a percentage actually describes
A percentage is meaningful only when its basis is clear. Four supported criteria out of five, with equal weighting, gives 80 per cent criteria coverage. That is not an 80 per cent probability of success or a complete assessment of suitability. If weights or additional statuses are involved, the calculation also needs to explain how those are treated.
Always examine the individual requirements, their importance and missing information alongside the number. An unanswered question should not silently become a negative finding. If you change the criteria during the search, review earlier applications against the revised standard and record why the change was made. Otherwise, the apparent comparison may conceal different assessment rules.
6. Discuss differences in the team's assessments
Ask team members to record their observations independently first. In the discussion, begin with evidence: which answer or experience supports the assessment? ‘Made a good impression’ is difficult to examine. ‘Explained the next action and a realistic response time in the delivery-delay exercise’ describes something concrete that the team can discuss and compare.
When assessments conflict, a focused follow-up may be more useful than an average score. One person may be assessing technical accuracy while another is assessing clarity. Separate those criteria. Then record whether another conversation is needed, a requirement still needs clarification or the team already has sufficient information to make its decision.
7. Record the reasoning and the next step
The goal is a reasoned selection, not a perfect spreadsheet. For each application moving forward, record relevant strengths, open questions and the next step. Document job-related reasons for not continuing just as carefully. Avoid personal labels and information unrelated to the work. Someone who joins the discussion later should be able to understand the reasoning.
Reclyra lets you open applications for the same job in a comparison view. Documents, AI CV analysis findings and team notes support the review; your team makes the decision. Use the comparison before an interview to prepare questions, then revisit it afterwards to assess the answers. This keeps the path from an initial information gap to a clarified assessment visible.
Your checklist
- Criteria are derived from the actual tasks of the job.
- Essential requirements and learnable skills are separated.
- Each assessment points to specific supporting evidence.
- Missing information remains visible as an open question.
- Work samples and core questions are comparable.
- The next step has a recorded reason and a named owner.