Automated Candidate Shortlisting ROI: A 90-Day Case Study

If your team is under pressure to fill roles faster without losing rigor or breaching GDPR, this 90-day case study shows what an explainable shortlisting rollout delivered. A 600-person UK ecommerce retailer used Marxel to move from manual CV triage and agency dependency to transparent, auditable rankings. The outcome: 41 percent faster time-to-hire, 73 percent fewer screening hours, and £24,500 in agency fees avoided, all validated by finance.

Baseline and hiring context

The business needed to staff mid-level operations and customer support roles ahead of a seasonal spike. Twelve requisitions were in scope: operations coordinators, CX team leads, and inventory analysts. The ATS supported basic keyword search but no scoring. Hiring criteria lived in spreadsheets and email threads, and seven of twelve roles typically went to agencies.

Starting point and goals

• 12 roles to fill in 90 days. Mix included shift-based operations and office-hour CX leadership.
• Average time-to-hire: 41 days from requisition approval to accepted offer.
• Average screening effort: 31 recruiter hours per role, driven by manual CV review and inconsistent intake notes.
• Agency usage on 7 of 12 roles. Average fee per filled role: £4,900.
• Targets: 25 percent faster time-to-hire, 50 percent fewer screening hours, and agency reliance cut in half, with no drop in probation pass rate or offer acceptance.

Implementation and data sources

We deployed Marxel to produce explainable shortlists inside the existing ATS workflow. Rollout focused on three workstreams: criteria design, data connections, and recruiter enablement.

Data connections and inputs

• ATS sync. Job posts, applications, and stage changes pulled via API every 30 minutes. Shortlists wrote back to the ATS so teams stayed in one system.
• Historic performance. One year of anonymized outcomes for comparable roles informed weighting. Signals included tenure at 90, 180, and 365 days; CX metrics such as average handle time and first contact resolution; and ops metrics such as pick accuracy and shift adherence. No sensitive attributes like age, ethnicity, or disability were processed.
• Role criteria. Hiring managers documented must-haves and nice-to-haves as verifiable CV signals. Examples: Excel proficiency evidenced by pivot tables, VLOOKUP or XLOOKUP, and nested IF statements; retail operations exposure with named WMS such as Manhattan or SAP EWM; weekend and late-shift availability; and stakeholder management with concrete examples like “led cross-team huddles” or “authored SOP playbooks.” Marxel converted these into transparent scoring rules.

Explainability by design

Every candidate in the shortlist displayed a score and evidence. Example: “+12 for two years in retail ops, +8 for SAP EWM, +5 for weekend availability, +3 for documented SOP authorship, −4 for no stakeholder management example.” Recruiters could see why rank changed when a rule weight was updated. That made calibration meetings concrete and reduced back-and-forth on subjective fit.

Privacy and compliance

The company required GDPR compliant CV screening with region-bound processing. We executed a UK DPA, restricted data residency to EU and UK regions, enabled role-based access, and set automatic 30-day deletion for rejected applicants unless consent was renewed. All scoring decisions log the criteria applied for a complete audit trail and to support data subject requests. We borrow well-tested privacy-by-design patterns used in consumer apps handling sensitive signals, similar to a sobriety tracker app with on-device privacy, and apply them to hiring data wherever practical.

Rollout timeline

• Week 1. Connect ATS, import 12 job profiles, and draft scorecards. Train two recruiters and three hiring managers. Define disqualifiers such as right to work and shift coverage.
• Week 2. Pilot on three live roles with daily calibration. Increased weight on Excel depth and reduced weight on degree pedigree after shadow reviewing top-ranked CVs.
• Weeks 3 to 4. Expand to all 12 roles. Replace manual pre-screen with Marxel shortlists. Recruiters retained final decision rights and could override with reasons.
• Weeks 5 to 12. Stabilize and monitor. Weekly checks for drift and fairness. Exported concise, evidence-based rejection reasons to the ATS to improve candidate comms.

Results on time and cost

ROI was calculated as cash savings plus time savings valued at loaded hourly rates, minus software cost. Finance reviewed the baseline, assumptions, and outputs.

Time-to-hire and throughput

• Time-to-hire dropped from 41 days to 24 days, measured from requisition approval to accepted offer.
• Screening time per role fell from 31 hours to 8.5 hours. The saved hours came from automated triage, consistent intake criteria, and auto-generated rejection notes.
• CV volume handled shifted from about 240 CVs per role to 90 reviewed, with recruiters focusing on the top 35 ranked by Marxel. Unqualified applicants were auto-advanced to polite, evidence-based rejections.

Agency spend and direct sourcing

• Agency usage fell from 7 of 12 roles to 2 of 12. Direct sourcing increased by 71 percent.
• Agency fees avoided in 90 days: £24,500, based on five direct hires that would previously have gone to agencies.

Software cost and ROI

• Marxel subscription for 90 days: £7,200.
• Recruiter time saved: 22.5 hours per role x 12 roles = 270 hours. At £45 loaded hourly rate, value = £12,150.
• Cash savings: £24,500 from reduced agency fees.
• Total quantified benefit: £36,650. Net benefit after software: £29,450.
• Observed payback: week 5, when the second direct hire closed and the corresponding agency fee was avoided.

Interview panels felt the change. With tighter criteria and clearer evidence, panels spent less time debating basic eligibility and more time probing impact. Total interview hours per filled role decreased by 18 percent without reducing the number of stages.

Quality of hire and fairness

Speed only matters if quality and fairness hold. We tracked early performance, acceptance rates, and outcome parity while excluding protected attributes from scoring.

Early performance and retention

• Ramp speed. CX hires hit ticket resolution and QA benchmarks 10 days sooner on average in the first 60 days.
• Probation pass rate. 93 percent vs 90 percent the prior quarter. A small uplift with no degradation signals.
• First-90-day regretted attrition. Flat at 5 percent despite faster hiring.

Candidate experience

• Offer acceptance rose from 67 percent to 76 percent. Candidates cited faster feedback and clearer role expectations in follow-up surveys.
• ATS complaints about opaque decisions fell by 44 percent, driven by rejection notes that mirrored shortlist explanations.

Fairness monitoring

Marxel reports outcome parity by stage and score range. In this window we saw no statistically significant difference in advancement rates by gender for the targeted roles. The education prestige rule showed mild adverse impact and was reduced in weight, replaced with validated Excel scenarios from work history. We also applied the 80 percent rule as a quick check and reviewed deltas weekly with the hiring managers.

Lessons and next steps

Three practices delivered most of the gains. First, require explainability. Adoption rose once recruiters could see, edit, and defend the reasons behind ranks. Second, constrain criteria to must-haves that appear in a CV and can be verified in interview or a short task. Third, monitor drift weekly. As panels learn what predicts success, roll that back into the rules and retire stale signals.

What we would repeat

• Manager workshops that translate fuzziness into testable signals. Replace “strong communication” with concrete evidence such as “facilitated cross-team huddles” and “wrote process playbooks.”
• A single calibration sprint in week 2 comparing three ranked lists side by side while adjusting weights live.
• Pushing shortlists back to the ATS so hiring teams never switch tools, which kept adoption high.

What we would change next time

• Invite panel interviewers into criteria design earlier. Their signals on what predicted success would have refined scorecards sooner.
• Automate link collection for portfolio or GitHub style evidence on analyst roles to verify tool proficiency faster.

For buyers comparing CV screening software in the UK, insist on three things. Confirm region-bound processing with a signed DPA and logged explanations. Ensure your team can view, edit, and audit scoring rules. Run a live-role pilot and measure time and cash separately so finance can validate ROI.

Next steps at this retailer are straightforward. Expand Marxel to merchandising and finance assistant roles. Add a lightweight work sample for operations coordinators and feed results back into the scoring rules. Schedule a quarterly fairness review that includes a legal observer. With those moves, the team expects to sustain sub-25-day hiring while halving the remaining agency usage again.

Key takeaways

• Time-to-hire down 41 percent and screening hours down 73 percent across 12 roles.
• Agency fees avoided: £24,500 in 90 days. Net ROI 4.1x after software cost.
• Explainable shortlists improved trust, acceptance rates, and panel efficiency.
• GDPR compliant CV screening is achievable with region-bound processing and auditable rules.

If your goal is measurable ROI from automated shortlisting, start with one live role, make the rules explainable, and track time and cash on separate lines. The rest follows.