The Challenge
A UK travel agency group with 120 agents across 8 locations was struggling with a persistent problem: new agent onboarding took too long and cost too much.
The baseline numbers:
| Metric | Before AI Training |
|---|---|
| Time to first unassisted booking | 3-4 weeks |
| Time to 80% productivity | 10-14 weeks |
| New agent attrition (first 6 months) | 35% |
| Onboarding cost per agent | £4,500-£6,000 |
| Training completion rate | 28% |
| New agent satisfaction with onboarding | 3.2/5 |
The onboarding programme consisted of a 3-day classroom induction, buddy pairing with an experienced agent, supplier webinar attendance, and access to a traditional LMS with product knowledge courses.
The problems were interconnected:
Slow productivity: New agents couldn't book confidently because they didn't know the product range well enough. They spent weeks observing experienced agents, absorbing information through osmosis rather than structured learning.
High attrition: Agents who felt unsupported and unproductive in their first months left. CIPD data indicates that inadequate onboarding is among the top three reasons for early-career turnover. Each departure cost the business an estimated £8,000-£12,000 in recruitment, training, and lost productivity.
Low training engagement: The LMS courses had an average completion rate of 28%. New agents found the content generic, too long, and disconnected from their daily work. They completed the mandatory compliance modules and largely ignored the rest.
Inconsistent knowledge: After 12 weeks, assessment testing revealed that new agent product knowledge varied wildly — from 35% to 78% — depending on which location they joined and which buddy they were paired with.
The Approach
The agency deployed an AI-powered training and coaching platform to replace their LMS-based onboarding programme. The implementation followed a structured timeline:
Weeks 1-2: Platform Configuration
- Uploaded product information for their top 50 products across key destinations
- Created role-specific onboarding pathways for agency advisors, contact centre agents, and homeworkers
- Built assessment checkpoints at key milestones (week 1, week 2, week 4, week 8)
- Configured AI coaching rubrics aligned to their selling methodology
- Set up performance dashboards for managers at each location
Weeks 3-4: Pilot with 8 New Starters
- Deployed to the next cohort of 8 new agents across 3 locations
- New agents completed adaptive training modules from day 1, alongside shortened classroom induction (1 day instead of 3)
- AI roleplay scenarios introduced from week 1 — practising basic enquiry handling conversations
- Daily AI coaching feedback on roleplay performance
- Weekly check-ins between location managers and pilot group
Weeks 5-8: Iteration and Full Deployment
- Refined content based on pilot feedback (shortened modules, added more roleplay scenarios, adjusted assessment difficulty)
- Extended programme to all new starters across all 8 locations
- Existing agents given access to the platform for ongoing development (voluntary initially)
- Manager training on using performance data for coaching conversations
The Onboarding Programme Structure
Week 1: Foundations
Day 1: Classroom induction (company values, systems, compliance basics) Days 2-5: AI training pathway — core product knowledge across the agency's top 10 destinations. Each module: 5-7 minutes, adaptive to the agent's responses. Two roleplay scenarios per day — basic enquiry handling with AI coaching feedback.
Week 1 assessment checkpoint: Test on foundational product knowledge and basic selling process.
Week 2: Building Confidence
AI training pathway intensifies — covering the next 20 products with deeper detail. Roleplay scenarios increase in complexity — now including customer with specific requirements, budget constraints, and mild objections.
Agents handle their first customer enquiries under buddy supervision, with AI training targeting any knowledge gaps revealed during live conversations.
Week 2 assessment checkpoint: Test on expanded product range and basic objection handling.
Weeks 3-4: Supported Independence
Agents handle enquiries independently with buddy available for consultation. AI training shifts to gap-filling — the platform identifies specific knowledge weaknesses from assessment data and targets those areas. Roleplay scenarios now include full-complexity selling situations — upselling, complex objections, multi-destination requests.
Week 4 assessment checkpoint: Comprehensive product knowledge and selling skills assessment. Agents scoring above 70% are cleared for fully independent selling.
Weeks 5-8: Performance Acceleration
Continued AI training on specialist products, advanced selling techniques, and competitive positioning. AI coaching focuses on selling technique refinement based on actual customer interaction patterns. Managers use performance data to target 1-to-1 coaching at each agent's specific development areas.
Week 8 assessment checkpoint: Full competency assessment against the agency's agent competency framework.
The Results
Headline Metrics
| Metric | Before AI Training | After AI Training | Change |
|---|---|---|---|
| Time to first unassisted booking | 3-4 weeks | 5-7 days | 75% faster |
| Time to 80% productivity | 10-14 weeks | 3-4 weeks | 70% faster |
| New agent attrition (first 6 months) | 35% | 14% | 60% reduction |
| Onboarding cost per agent | £4,500-£6,000 | £1,800-£2,500 | 58% reduction |
| Training completion rate | 28% | 91% | 225% improvement |
| New agent satisfaction with onboarding | 3.2/5 | 4.6/5 | 44% improvement |
| Week 4 product knowledge score | 52% average | 76% average | 46% improvement |
| Week 4 knowledge score variation | 35%-78% range | 68%-84% range | Far more consistent |
Revenue Impact
The revenue impact of faster onboarding is the most commercially significant outcome:
| Calculation | Detail |
|---|---|
| Average new hires per year | 30 |
| Time saved per agent (weeks at reduced productivity) | 8 weeks |
| Revenue per fully productive agent per week | £6,250 |
| Productivity during ramp-up (old method) | ~40% |
| Productivity during ramp-up (new method) | ~70% |
| Revenue gain per agent (8 weeks × £6,250 × 30% improvement) | £15,000 |
| Annual revenue gain (30 agents) | £450,000 |
| Annual platform cost (120 agents) | £24,000 |
| First-year ROI | 1,775% |
Retention Impact
The reduction in attrition from 35% to 14% generated additional savings:
| Calculation | Detail |
|---|---|
| Previous attrition: 35% of 30 new hires = 10.5 departures | |
| New attrition: 14% of 30 new hires = 4.2 departures | |
| Departures avoided | 6.3 |
| Cost per departure (recruitment + training + lost productivity) | £10,000 |
| Annual retention saving | £63,000 |
Key Learnings
What Worked
Day-one active learning. New agents started learning actively from day one rather than observing passively. This maintained the energy and motivation of the hiring decision.
Adaptive pathways. Agents with prior travel experience (career changers, returning agents) progressed faster because the AI skipped content they already knew. First-career agents received additional foundational content without slowing down experienced starters.
Roleplay from week one. Even basic roleplay scenarios in the first week built confidence and established the expectation of practice-based learning. By week 3, agents were comfortable with roleplay and actively sought practice opportunities.
Manager dashboards. Location managers could see exactly where each new agent was in their development — which knowledge areas were strong, which needed attention, and what coaching conversations to prioritise. This replaced the previous approach of asking buddies "how's the new person doing?"
What Required Adjustment
Initial content was too detailed. The first version of the training modules tried to cover everything about each destination. Feedback from the pilot revealed that new agents needed breadth first (awareness of many destinations) and depth later (detailed knowledge of key products). Content was restructured accordingly.
Assessment difficulty needed calibration. Week 1 assessments were initially too hard — creating anxiety rather than confidence. Difficulty was reduced for early checkpoints and increased progressively.
Buddy programme wasn't eliminated — it evolved. Rather than removing buddies entirely, the programme evolved. Buddies shifted from "teach the new person everything" to "be available when the AI training surfaces questions that need human context." This was a better use of experienced agents' time.
Applicability
This case study's results are consistent with broader industry data. ABTA member agencies implementing AI onboarding report 50-75% reductions in time-to-productivity, with the specific improvement depending on the starting position and the comprehensiveness of implementation.
The approach is applicable to:
- Travel agencies of any size (the platform scales from 5 to 5,000 agents)
- Tour operator trade teams onboarding new BDMs or trade support staff
- Hotel groups onboarding front desk, reservations, and sales staff
- Contact centres where rapid onboarding directly affects capacity and revenue
Transform your onboarding with TravAI →
This article is part of our AI in Travel & Tourism series. Related reading:
