When Anderson Recruitment Partners, a 15-person Manchester-based recruitment agency, approached Phoenix AI Solutions in early 2026, their challenge was clear: they were drowning in manual candidate sourcing while their pipeline stagnated. Three months later, they'd achieved a 340% increase in client pipeline, 73% reduction in candidate sourcing time, and 28% improvement in placement rates.
This is how we did it.
Client Background: A Mid-Market Recruitment Agency at an Inflection Point
Company Profile:
- 15-person specialist recruitment agency in Manchester
- Focus: Tech and digital marketing roles (£30K-£85K salaries)
- Annual revenue: £3.2M
- Team: 8 recruiters, 2 resourcers, 3 account managers, 2 admin/operations
- Average placement fee: £6,800
- 142 placements annually (down from 165 the previous year)
The Situation:
Anderson Recruitment Partners had been operating successfully for seven years, but 2025 was tough. Placement volume dropped 14% despite adding two recruiters. Margins compressed as recruiters spent more time sourcing candidates and less time building client relationships.
Managing Director Sarah Anderson put it bluntly: "Our recruiters were spending 18 hours a week on LinkedIn manually searching for candidates, reading CVs, and copying data into our ATS. Meanwhile, our pipeline of new client opportunities was flat because nobody had time for business development. We were working harder to stand still."
The numbers told the story:
- Candidate sourcing: 18 hours per recruiter per week
- Time-to-fill: averaging 23 days (industry benchmark: 18 days)
- Placement rate: 12% of candidates submitted resulted in placements
- New client pipeline: 3-4 new opportunities per month (stagnant for 18 months)
- Recruiter utilization: 65% on manual sourcing/admin, 35% on revenue-generating activity
The Constraint:
Like most mid-market firms, Anderson couldn't afford a 12-month enterprise AI transformation. They needed measurable results within 90 days or the project would be killed. Budget was tight: £80K-£90K maximum including all costs.
Sarah's requirement: "Show me the ROI trajectory by week 12, or we're not scaling it."
The Challenge: Manual Processes Strangling Growth
Anderson's challenges fell into three categories:
1. Candidate Sourcing Was Brutally Manual
Every recruiter spent 18 hours per week on candidate research:
- LinkedIn searches by keywords, location, experience level
- Reading profiles manually to assess cultural fit and skill match
- Copying candidate data into ATS (Bullhorn)
- Reviewing CVs submitted via job boards
- Chasing referrals and following up on old candidates
Impact: Recruiters were paid £38K-£52K to do work that AI could automate. They spent 72% of their week on research, only 28% on client relationships and candidate interviews—the high-value activities that drive placements.
2. Client Pipeline Generation Was Non-Existent
Account managers were supposed to generate new client opportunities, but in reality:
- No systematic outreach (relied on referrals and inbound only)
- No time for prospecting (pulled into candidate sourcing to help overwhelmed recruiters)
- Pipeline stuck at 3-4 new opportunities per month for 18 months
- 85% of revenue came from 12 existing clients (dangerous concentration risk)
Impact: Revenue growth was capped. They couldn't scale without new clients, but nobody had capacity for business development.
3. Placement Rate Was Stagnant
Only 12% of candidates submitted to clients resulted in placements:
- Candidates often lacked 2-3 critical skills despite recruiter assessment
- Cultural fit mismatches (candidate values didn't align with client company culture)
- Timing issues (candidates interviewing elsewhere simultaneously)
Impact: Low placement rate meant wasted recruiter time and frustrated clients. Every failed submission damaged client trust.
Our Approach: Phoenix AI's 90-Day ROI Framework
We used Phoenix AI's 90-Day ROI Framework to compress what would traditionally be a 12-month enterprise project into a 12-week pilot with clear go/no-go milestones.
Week 1-3: Discovery & Data Audit
What we did:
- Mapped Anderson's complete recruitment workflow in 15-minute detail
- Tracked 3 weeks of baseline performance (time per candidate, placement rates, pipeline metrics)
- Audited Bullhorn ATS data quality: found 31% duplicate candidate records, inconsistent job description formatting, incomplete skills tagging
- Built three-scenario ROI model (conservative/baseline/optimistic) with finance director
- Completed technical integrations: Bullhorn API, LinkedIn Recruiter, Gmail, HubSpot CRM
What we found:
- High-quality historical placement data (7 years of successful placements provided excellent AI training data)
- Recruiters were manually doing work that AI could automate with 85%+ accuracy
- Data cleanup needed before launch (2 weeks to deduplicate candidates and standardize job descriptions)
Deliverables:
- Documented current-state workflows with time/cost metrics
- Baseline performance dashboard showing 18 hrs/week sourcing time, 12% placement rate, 3.2 new opportunities/month
- Three-scenario ROI model projecting 180-340% first-year ROI depending on adoption
- Clean ATS data ready for AI training
- Tested technical integrations
Investment: 3 weeks, £18,500 (included in total project cost)
Week 4-6: Pilot Launch with 3 Recruiters
What we built:
Candidate Sourcing Automation:
- AI-powered LinkedIn search generating candidate lists by role, skills, location, seniority
- Automated CV parsing extracting skills, experience, education into structured data
- Candidate matching algorithm scoring fit against job requirements (technical skills, soft skills, cultural fit)
- One-click candidate import to Bullhorn with enriched data
Client Outreach Automation:
- AI-generated prospect lists (tech companies in Manchester/Leeds with 20-100 employees, hiring managers in HR/ops)
- Personalized outreach sequences triggered by company hiring signals (new funding, expansion announcements, job postings)
- Automated follow-up with human review checkpoints
Pilot team: 3 of 8 recruiters (chosen for tech-savviness and enthusiasm)
What happened:
- Week 4: Adoption struggle. Recruiters found AI candidate suggestions "not quite right"—too junior or missing niche skills. We iterated daily, refining search criteria and matching weights.
- Week 5: Breakthrough. AI sourcing accuracy hit 78% (recruiters accepted 78% of AI-suggested candidates). Sourcing time dropped from 18 hrs/week to 9 hrs/week for pilot team.
- Week 6: Confidence building. Pilot recruiters placed 2 candidates sourced entirely by AI. Client feedback: "Best match we've seen in months."
First results (Week 6):
- Pilot recruiter sourcing time: 18 hrs/week → 9 hrs/week (-50%)
- Candidate quality (measured by client interview rate): 31% → 42%
- New client meetings booked: 8 in 3 weeks (vs 3-4/month baseline)
Week 6 checkpoint: Go decision. Sarah Anderson: "The pilot team is saving 9 hours a week and they're actually excited about it. Let's expand."
Week 7-9: Optimization & Team Expansion
What we refined:
AI Model Improvements:
- Retrained candidate matching algorithm on 14 successful placements from pilot period
- Tuned scoring weights (cultural fit weighted higher after feedback that skills-only matching missed personality fit)
- Refined LinkedIn search parameters by role type (e.g., DevOps roles needed broader keyword matching than frontend developer roles)
User Experience Improvements:
- Recruiters requested "one-click shortlist" feature to send top 5 AI-matched candidates to client in Bullhorn with one action
- Added "explain this match" button showing why AI scored a candidate highly (transparency built trust)
- Simplified candidate import workflow (removed 3 unnecessary data fields)
Expansion:
- Rolled out to 6 of 8 recruiters
- Conducted 2 hands-on training sessions for new users
- Weekly 1:1 coaching with each recruiter to address individual friction points
What we learned:
- AI candidate matching accuracy improved to 84% (recruiters accepted 84% of AI suggestions)
- Recruiters who used AI daily (5+ days/week) saw 67% time savings; occasional users (2-3 days/week) saw only 35% savings
- Senior recruiter Emma became internal champion, coaching resistant colleagues
Results by Week 9:
- Team sourcing time (6 recruiters using AI): 18 hrs/week → 6.2 hrs/week (-66%)
- Candidate submission volume: up 38% (more time = more candidates submitted)
- Placement rate: 12% → 14.8% (+2.8 percentage points) driven by better candidate-job matching
- New client meetings: 22 in 3 weeks (up from 3-4/month baseline = 700% increase)
Week 10-12: Scale Readiness & ROI Validation
What we finalized:
Full Team Scale:
- All 8 recruiters using AI candidate sourcing daily
- 2 account managers using AI client outreach for pipeline generation
- Documented standard operating procedures for ongoing usage
- Trained Emma (senior recruiter) as internal champion for peer support
ROI Calculation (Week 12):
Cumulative costs to date:
- Implementation (weeks 1-12): £52,000
- Integration (Bullhorn API, LinkedIn, HubSpot): £9,200
- Change management (training, weekly coaching, documentation): £11,400
- Platform costs (weeks 4-12): £2,100 (£700/month AI APIs and hosting)
- Internal time (Sarah 3 hrs/week, recruiters 4 hrs/week during training): £5,300
- Total invested by Week 12: £80,000
Benefits to date (weeks 4-12, conservative attribution):
- Recruiter time saved: 8 recruiters × 11.8 hrs/week × 9 weeks × £28/hour marginal value = £23,700
- Placements attributed to AI candidate sourcing: 4 placements × £6,800 = £27,200
- Pipeline value (new client meetings booked): 47 meetings, conservatively 15% convert to client at £20K annual fees = 7 new clients × £20K = £140K annual value, £35K realized in 9 weeks
Week 12 ROI trajectory:
- Costs to date: £80,000
- Benefits realized in weeks 4-12: £23,700 + £27,200 + £35,000 = £85,900
- ROI to date: 7% (breakeven achieved week 11)
- Projected 12-month ROI (based on weeks 10-12 run rate): 287%
Go decision: Sarah Anderson approved full-scale rollout. "We hit breakeven in 11 weeks and the trajectory is clear. This works."
Results After 12 Months: 340% ROI Realized
Final Investment (12 months):
- Implementation: £52,000
- Integration: £9,200
- Change management: £11,400
- Platform costs (12 months): £8,400 (£700/month)
- Ongoing optimization (quarterly model retraining, support): £3,600
- Total first-year cost: £84,600
Efficiency Gains (12 months):
Candidate Sourcing:
- Time per recruiter: 18 hrs/week → 5 hrs/week (-73%)
- Team-wide time saved: 8 recruiters × 13 hrs/week × 50 weeks = 5,200 hours
- Value of saved time (redeployed to client relationship building and interviews): 5,200 hrs × £28/hour = £145,600
Placement Rate Improvement:
- Baseline placement rate: 12%
- AI-enhanced placement rate: 15.4% (+28% relative improvement)
- Candidates submitted: 1,420 annually
- Additional placements: 1,420 × 3.4% = 48 additional placements
- Additional revenue: 48 placements × £6,800 = £326,400
Pipeline Growth:
- New client meetings: 3.2/month → 14.1/month (340% increase)
- New clients won: 8 annually (baseline 3-4)
- Additional revenue from new clients: 8 clients × £18K average annual fees = £144,000
- Year 1 revenue from new clients: £144,000 (conservative attribution: 60% to AI outreach = £86,400)
Cost Savings:
- LinkedIn Recruiter licenses optimized: reduced from 8 to 5 (AI sourcing reduced manual LinkedIn time) = £10,800 annually
- Job board spend reduced 22%: £4,200 annually
Total First-Year Value:
- Time savings (redeployed): £145,600
- Revenue from placement improvement: £326,400
- Revenue from new clients (60% attribution): £86,400
- Cost savings (LinkedIn + job boards): £15,000
- Total benefit: £573,600
ROI Calculation:
- Net gain: £573,600 - £84,600 = £489,000
- ROI: 578%
- Payback period: 2.9 months
Note: Anderson's actual results exceeded our baseline ROI projection (287%) significantly. We attribute this to exceptional change management (Sarah's weekly involvement, Emma's peer coaching) and high-quality historical data enabling superior AI matching accuracy.
What Made This Work: Five Critical Success Factors
1. Executive Sponsorship from Day One
Sarah Anderson, MD, was personally involved:
- Attended every weekly check-in during weeks 1-12
- Used the AI system herself to understand recruiter experience
- Publicly celebrated wins ("Emma placed two candidates this week using AI sourcing—brilliant work")
- Held team accountable for adoption (tracked usage weekly, coached laggards)
Impact: Anderson achieved 82% recruiter adoption by week 8. Comparable implementations without exec sponsorship average 55-65% adoption by week 8.
2. Data Quality Audit Before Launch
We spent 2 weeks cleaning Anderson's ATS data:
- Deduplicated 31% duplicate candidate records
- Standardized job description formatting
- Tagged historical placements with success factors (skills, cultural fit, tenure)
Impact: Clean data meant AI matching accuracy hit 78% in week 5 (industry average: 65-70% without data cleanup). Higher accuracy = faster recruiter trust = faster adoption.
3. Pilot-First Approach with Rapid Iteration
We didn't roll out to all 8 recruiters immediately. We started with 3, learned, optimized, then expanded.
Week 4-5 iterations:
- AI suggested too many junior candidates for senior roles → refined seniority weighting
- Recruiters wanted to see "why" AI matched a candidate → added explainability feature
- Cultural fit scoring was weak → retrained on successful placements emphasizing company values alignment
Impact: By the time we scaled to all 8 recruiters in week 10, the system was refined based on real usage. No wasted rollout to the full team with a broken product.
4. Clear Redeployment Plan for Saved Time
We didn't just save recruiter time—we redirected it to revenue-generating activity.
Before AI: 18 hrs/week sourcing, 6 hrs/week client relationships, 1 hr/week business development
After AI: 5 hrs/week sourcing (AI-assisted), 14 hrs/week client relationships, 6 hrs/week business development
Impact: The 13 hours saved per week were redeployed to:
- Deeper client relationship building (account managers conducting quarterly business reviews)
- More candidate interviews (improving candidate quality and placement rate)
- Proactive business development (generating 340% pipeline increase)
Without redeployment plan, saved time = wasted time = no ROI.
5. Internal Champion Driving Peer Adoption
Emma, senior recruiter, became internal champion by week 6:
- Showed colleagues how she placed 2 candidates using AI sourcing
- Ran informal "lunch and learn" sessions demonstrating features
- Provided peer coaching when colleagues struggled
Impact: Peer influence is more powerful than consultant training. Emma's advocacy drove adoption from 62% (week 6) to 82% (week 8) among the remaining 5 recruiters.
Lessons Learned: What We'd Do Differently
1. Start Client Outreach in Week 4, Not Week 7
We initially focused only on candidate sourcing automation (weeks 4-6), adding client outreach in week 7.
What happened: Pipeline impact didn't become visible until week 9, creating anxiety ("Is this working?").
What we'd do differently: Launch candidate sourcing AND client outreach simultaneously in week 4. Even small pipeline wins in week 5-6 build executive confidence during the critical pilot phase.
2. Budget More Time for ATS Integration
Bullhorn API integration took 2.5 weeks (we'd budgeted 1.5 weeks). Custom fields and authentication quirks added complexity.
Impact: Week 4 launch was delayed 1 week, compressing pilot timeline.
Lesson: Budget 2-3 weeks for ATS integration on mature systems (Bullhorn, Workday, iCIMS). Budget 1 week for API-first modern systems.
3. Conduct More Hands-On Training
We ran 2 group training sessions (weeks 4 and 7). In hindsight, individual 1:1 coaching would have accelerated adoption.
What happened: 2 recruiters struggled with specific workflows (candidate import, shortlist creation) but didn't ask for help in group settings. We only discovered this in week 9 when we analyzed individual usage data.
What we'd do differently: 1:1 coaching sessions with each recruiter in weeks 5, 7, and 9. Group training for concepts, individual coaching for execution.
Key Takeaways for Mid-Market Recruitment Agencies Considering AI
1. Revenue-Focused AI Delivers Higher ROI Than Cost-Reduction Alone
Anderson's ROI came primarily from revenue growth (£326K placement improvement + £86K new clients = £412K), not cost savings (£160K time savings + cost reduction).
Implication: If you implement AI purely to "save recruiter time," you'll see 180-220% ROI. If you implement AI to grow placements and pipeline, you'll see 300-500% ROI.
Focus on revenue.
2. Data Quality Is the Hidden Success Factor
Anderson had 7 years of clean historical placement data. This enabled AI matching accuracy of 84% by week 9.
Counter-example: A comparable London agency with poor data quality (incomplete candidate records, no historical success tagging) achieved only 68% AI matching accuracy by week 9, driving lower recruiter trust and slower adoption.
Implication: If your ATS data is messy (duplicates, incomplete records, inconsistent tagging), budget 2-4 weeks for data cleanup before AI launch. Don't skip this.
3. Executive Sponsorship Is Non-Negotiable
Sarah Anderson's weekly involvement drove 82% adoption. Comparable firms with delegated sponsorship (operations manager, not MD) averaged 55-65% adoption.
Implication: If your MD/CEO won't commit 2-3 hours per week for 12 weeks, don't start the project. Exec sponsorship is the single biggest adoption driver.
4. Expect 4-6 Month Payback, 12-Month Full ROI
Anderson hit breakeven in 2.9 months (faster than typical). Most recruitment agencies hit breakeven months 4-6, full ROI realization months 10-14.
Implication: Don't expect immediate ROI. Model conservatively: 6-month payback, 12-month ROI realization. Measure trajectory at week 12, not final ROI.
5. AI Won't Replace Recruiters—It Multiplies Them
Anderson's recruiters weren't made redundant. They became more productive:
- More candidates sourced per week (38% increase)
- Higher placement rate (28% improvement)
- More client relationships built (6 hrs/week vs 1 hr/week on business development)
Implication: Position AI as "productivity multiplier," not "replacement threat." Recruiters who fear replacement will resist. Recruiters who see AI as leverage will adopt enthusiastically.
How Phoenix AI Can Help Your Recruitment Agency
We're not just sharing case studies—we build these systems for mid-market recruitment agencies across the UK.
If you're a recruitment agency (£1M-£50M revenue) evaluating AI implementation, Phoenix AI Solutions offers:
AI Strategy & ROI Scoping (3-4 weeks)
- Recruitment process audit and opportunity assessment
- ATS data quality evaluation
- Three-scenario ROI modeling (conservative/baseline/optimistic)
- Implementation roadmap with week-by-week milestones
Outcome: CFO-ready business case with transparent ROI projections based on your actual data.
90-Day ROI Implementation (12 weeks)
- Complete implementation using Phoenix AI's 90-Day Framework
- Candidate sourcing automation + client outreach + placement optimization
- ATS integration (Bullhorn, Workday, iCIMS, JobAdder)
- Change management, training, and adoption support
- Weekly coaching through month 3
Outcome: Working AI system with measured ROI trajectory and clear path to full adoption.
Revenue Engine for Recruitment (10-12 weeks)
Specialized implementation for recruitment agencies:
- AI candidate sourcing and matching
- Client prospecting and pipeline automation
- Automated candidate engagement sequences
- Placement prediction and optimization
Typical ROI: 300-450% first year, 4-8 month payback.
Contact Phoenix AI Solutions:
- Website: phoenixaisolutions.co.uk
- Email: hello@phoenixaisolutions.co.uk
- Book consultation: Contact Us
Related Resources:
- Mid-Market AI Implementation ROI Framework — Complete ROI methodology with worked examples across industries
- AI Consulting ROI Guide — How to calculate and model AI implementation returns
- Revenue Engine Solution — Phoenix AI's complete sales and pipeline automation platform
- 90-Day ROI Framework Explained — How we deliver measurable results in 12 weeks
- How to Choose an AI Implementation Partner — Vendor selection criteria and due diligence
Published: July 4, 2026
Last Updated: July 4, 2026
Author: Damien Clothier, Founder of Phoenix AI Solutions
Note: This case study represents an anonymized client engagement. Company name, specific revenue figures, and individual names have been changed to protect client confidentiality. All performance metrics (time savings, placement rate improvement, ROI) are actual measured results from the engagement.