The Resume Filter
A regional call center receives about 400 applications for 20 openings each hiring cycle. The hiring lead, Tomas, currently reads every resume himself, which takes roughly 12 hours per cycle. He proposes buying an automated screening tool that scores resumes against keywords from the job posting and rejects anything below a threshold.
The tool costs $3,000 per year. Tomas earns about $28 per hour, and the company runs six cycles a year. A colleague warns that the tool may screen out strong candidates whose resumes simply use different wording - for example, someone who wrote "customer support" instead of "client service." The company also has a policy that all hiring decisions must be reviewable on request.
1. Calculate the annual labor cost of Tomas's current resume review and compare it with the tool's cost. Then explain why the cheaper option is not automatically the better one.
Read the worked answer
Current cost: 12 hours per cycle × 6 cycles = 72 hours per year. At $28 per hour, that is 72 × $28 = $2,016 per year.
Tool cost: $3,000 per year, plus whatever staff time is needed to configure it, monitor results, and handle appeals.
On labor alone, the tool costs about $984 more per year than Tomas's time. So the tool is not cheaper in direct terms - a point worth stating plainly, since the proposal may assume it is.
The cheaper option is not automatically better because:
- Tomas's 72 hours may be worth more elsewhere. If those hours could be spent on interviews or onboarding, the comparison changes.
- The tool's real cost is higher than $3,000. Setup, tuning, and appeal handling are unbudgeted.
- The tool's real benefit is consistency and scale. It may reduce missed candidates and free Tomas for higher-value work - but only if it works well.
- Quality risk is unmeasured. A tool that rejects good candidates creates costs - bad hires, lost talent, legal exposure - that never appear on this comparison.
2. The colleague's concern about keyword mismatch is a real risk. Describe how Tomas could test the tool before committing, and what result would justify buying it.
Read the worked answer
Run a back-test on past cycles. Tomas has historical resumes and knows which candidates were hired and how they performed. He can score those resumes with the tool and check two things:
- False negatives: did the tool reject people the company actually hired and valued?
- False positives: did it pass candidates who clearly did not meet the bar?
Set a low initial threshold. The tool should be used to rank and sort, not to auto-reject. Tomas reviews the top-scoring resumes first but still sees the rest, at least during a trial period.
Test wording variants. Deliberately include resumes that use synonyms ("customer support" vs. "client service") and check whether the tool catches them. If it does not, the keyword list needs expanding or the tool needs semantic matching.
Run a parallel cycle. For one cycle, use the tool alongside Tomas's normal review and compare the shortlists. If the tool's shortlist misses candidates Tomas would have interviewed, that is a warning sign.
What would justify buying it: the tool ranks the eventual hires near the top, its false-negative rate on known-good candidates is low, and it saves meaningful time after setup. If it rejects candidates the company would have hired, the savings do not offset the lost talent.
3. The policy requires that hiring decisions be reviewable on request. How should this shape the way the tool is used?
Read the worked answer
The policy means the company must be able to explain, for any rejected applicant, why the decision was made. That rules out using the tool as an unexplained black box that silently discards resumes.
Practical implications:
- Keep human review in the loop. A person should make the final call on who advances. The tool can rank and flag, but a human decision is what gets reviewed.
- Retain records. Store the tool's scores and the reasons for rejection alongside each application, so a review request can be answered with specifics.
- Be able to explain the criteria. If an applicant asks why they were rejected, "the software said so" is not an adequate answer. Tomas should be able to point to job-related requirements the applicant did not meet.
- Watch for proxy effects. Keyword filters can disadvantage applicants who describe equivalent experience differently. The company should be able to show its criteria are job-related and applied consistently.
- Provide an appeal path. Applicants should be able to ask for human reconsideration.
This does not forbid the tool. It means the tool must support a documented, human-owned process rather than replace it. Used that way, automation saves Tomas time while keeping decisions defensible.