Industry
Technology Staffing & Talent Acquisition
Dataset period
April 2025 – June 2026 (14.5 months)
Scale
17,021 applications · 511 open roles · 5-person recruiting team
The Challenge
Recruiting teams in IT staffing face a volume problem that compounds faster than headcount can scale. The IT sector’s average time-to-hire sits at 40–45 days (LinkedIn Talent Solutions), recruiters spend 40–60% of their working hours on CV screening alone (Deloitte Human Capital), and a single active role can generate hundreds of applications. The team behind this dataset was managing 270 inbound CVs per week – peak months exceeded 2,700 applications – with no automated way to rank or filter. Without a scoring layer, recruiters spent most of their time reading CVs with no realistic path to hire, leaving less capacity for interviews, candidate relationship management, and sourcing.
Features:
A candidate applies through a role-specific link. The platform parses the CV, extracts structured data – skills, experience history, education, tenure – and scores it against the role’s requirements. High-scoring candidates move into a deeper automated analysis that maps their profile against core competencies and generates structured screening notes. The recruiter opens a ranked shortlist instead of a raw inbox, runs a structured screening call, and schedules an interview directly from the dashboard. From there, a tracked pipeline carries the candidate through screening, shortlisting, interview, selection, offer, and digital onboarding – every stage timestamped and auditable. For multi-location hiring, one job posting feeds candidate pools across branches without duplication.
Technology
The Solution:
The platform wraps the existing recruitment workflow in an AI evaluation layer that scores every inbound CV automatically, before any recruiter sees it. A multi-model scoring pipeline assigns a 0–100 job-fit score to candidates within hours of application, based on structured analysis of skills, experience depth, tenure patterns, and role-specific requirements. Candidates ranked above the screening threshold get a richer automated profile – competency mapping, skill gap identification, suggested screening questions – surfaced directly in the recruiter dashboard alongside a ranked shortlist. Human review concentrates on the 15% of applications the AI has already flagged as worth a recruiter’s time. Scheduling, pipeline progression, and onboarding documentation sit in the same platform, creating one tracked record from first application to hire.
Services Offered
- AI Product Design & Architecture
- Multi-Model Scoring Pipeline Development
- Recruiter Dashboard & UX
- API & Third-Party Integrations
- Digital Onboarding Workflow
- Ongoing AI Model Optimisation
How It Works
A candidate applies through a role-specific link. The platform parses the CV, extracts structured data – skills, experience history, education, tenure – and scores it against the role’s requirements. High-scoring candidates move into a deeper automated analysis that maps their profile against core competencies and generates structured screening notes. The recruiter opens a ranked shortlist instead of a raw inbox, runs a structured screening call, and schedules an interview directly from the dashboard. From there, a tracked pipeline carries the candidate through screening, shortlisting, interview, selection, offer, and digital onboarding – every stage timestamped and auditable. For multi-location hiring, one job posting feeds candidate pools across branches without duplication.
Results That Speak
All metrics are drawn from live production data across 14.5 months of operation. Industry benchmarks are sourced from LinkedIn Talent Solutions Annual Report, SHRM Talent Acquisition Benchmarking, and Deloitte Global Human Capital Trends.
CV Screening Throughput
| This platform | IT industry benchmark | |
|---|---|---|
| CVs processed per week | 270 | Manual: ~25–30 per recruiter/week¹ |
| AI scoring coverage | 99.99% of all applications | – |
| Peak intake absorbed | 2,710 in one month | Typically requires temporary headcount |
| Screening volume per recruiter | ~54 CVs/week with AI assistance | 25–30 CVs/week without AI |
¹ SHRM Talent Acquisition benchmark for manual CV review in IT roles.
The 5-person team processed a volume that would require 9–11 dedicated screeners under a fully manual model.
Recruiter Hours Saved
Using SHRM’s published estimate of 6 minutes per CV for a meaningful manual review:| Scenario | CVs reviewed | Time at 6 min/CV |
|---|---|---|
| Fully manual (no AI) | 17,021 | 1,702 recruiter-hours |
| With AI (human reviews AI-passed CVs only) | 2,583 | ~86 recruiter-hours² |
| Estimated saving | – | ~1,616 recruiter-hours over 14.5 months |
² AI-ranked, pre-summarized CVs require an estimated 2 minutes of human review rather than 6.
Annualized, that’s roughly 1,340 recruiter-hours per year returned from screening – equivalent to 0.64 FTE redirected to interviews, sourcing, and candidate management. Per recruiter on the 5-person team, that’s about 5 hours per week reclaimed.
Deloitte Human Capital research finds recruiters who spend less than 40% of their time on screening close roles 30% faster and report significantly higher hiring manager satisfaction.
AI Filter Accuracy – Score Correlation by Outcome
| Pipeline stage | Avg AI fit score | n |
|---|---|---|
| Hired | 68.1 | 43 |
| Interview stage | 62.5 | 115 |
| Active pipeline / screened | 61.1 | 1,895 |
| Shortlisted | 50.9 | 540 |
| Rejected | 49.5 | 7,000 |
Hired candidates scored an average of 18.6 points higher than those rejected at screening – a consistent signal that the ranking captures genuine role fit, not keyword matching. The secondary analysis layer reinforces the same gradient: hired candidates averaged 65.7 on the deeper analysis score versus 46.3 for screened-out candidates.
Shortlist Quality vs. Industry Benchmark
| This platform | IT industry benchmark¹ | |
|---|---|---|
| Shortlist-to-interview conversion | 21.3% (115 of 540) | 15–20% |
| Interview-to-hire rate | 37.4% (43 of 115) | 30–35% |
| One shortlisted candidate in every… | 12 was hired | ~20–25 (industry est.) |
¹ LinkedIn Talent Solutions Annual Report; SHRM Talent Acquisition Benchmarking.
Time-to-Interview
| This platform | IT industry benchmark | |
|---|---|---|
| Application to interview scheduled | 8.0 days | 40–45 days full time-to-hire¹ |
| For candidates who were hired | 7.0 days to interview | – |
¹ Full time-to-hire includes offer and notice period. The 8-day figure covers the recruiter-controlled portion of the cycle: screening, shortlisting, scheduling.
Rejection Handling at Scale
7,000 applications were filtered out at the AI screening stage (41.1% of total inflow), reviewed by no recruiter in depth. Under a manual process at 6 minutes per CV, those alone would have consumed 700 recruiter-hours. The system handled them entirely automatically.Offer and Onboarding
Digital offer letters and onboarding document collection are integrated in the same platform – no paper, no separate system. Offer volume in the current period is early-stage (3 formal offers issued); conversion and time-to-sign rates require additional pipeline maturity to report reliably.Why It Matters
Every unfilled role carries a daily cost: lost productivity, management distraction, and candidates who accept competing offers while a slow process catches up. Compressing time-to-interview to under eight days – against an industry average of 40-plus days for a full hire cycle – means the team is consistently faster to shortlist and schedule than peers running manual processes. A shortlist that converts to interviews at 21% (vs. a 15–20% industry benchmark) and interviews that convert to hires at 37% (vs. 30–35%) reduces the number of interview rounds needed per hire, directly lowering cost-per-hire and improving hiring manager confidence. The recruiter hours recovered from screening – roughly five hours per person per week – go toward the relationship work and strategic sourcing AI can’t do.
About the Solution
The platform is built on the principle that AI should concentrate human attention, not compete with it. Every score is visible, every ranking decision is explainable, and every output is overridable by the recruiter. The scoring architecture is model-agnostic – the underlying AI components can be updated as better-performing models become available without restructuring the surrounding workflow. The system is in active production use: the metrics above reflect 14.5 months of live hiring operations across a real recruiting team.
What This Demonstrates
This case study isn’t a recruitment product we’re trying to sell you — it’s evidence of how we build AI-implemented applications: a multi-model scoring pipeline, wired into a live operational workflow, validated against 14.5 months of real production data. If you’re evaluating an AI partner for a different problem entirely — not hiring, not staffing — this is the engineering and AI-implementation depth you’d be working with.
Book a 30-minute Digital Transformation & Product Roadmap Review with Carmatec’s team to talk through what an AI-implemented application could look like for your own operations. We’ve been building software and AI systems for 23 years, across teams in Bangalore, Doha, Dubai, New York, and London.
English 
Spanish (Mexico) 
Italian 
French 
German 
Japanese 
Dutch 
Swedish 
Finnish 
Spanish (Spain) 