AI Lead Scoring for a European Dealer Group — +28% Conversion in 60 Days
Automotive
Visual-AI-Labs added an AI lead-scoring and follow-up layer on top of the dealer group’s existing CRM in two 30-day cycles, lifting conversion by 28%.
- +28% — Showroom-visit conversion
- 11 h → 9 min — Time-to-first-contact (leads)
- −40% — Sales-rep time on cold leads
- +34% — Test-drive bookings / lead
The problem
Leads came in across web forms, marketplaces, brand-portal hand-offs and walk-ins, and landed in a shared CRM queue. Sales reps worked the queue top-down, which meant hot leads waited next to cold tyre-kickers, and the first response often arrived more than 10 hours after a customer submitted a form. The group did not want to replace its CRM — it wanted leads ranked and followed up faster.
What Visual-AI-Labs built
Visual-AI-Labs delivered the project in two 30-day cycles. Cycle 1 shipped lead scoring: a model trained on the group’s historical CRM data plus inbound signal (source, vehicle interest, financing intent, message content) that re-ranks the queue in real time and surfaces the top-quartile leads to the right rep at the right showroom. Cycle 2 shipped an AI follow-up assistant that drafts a personalised first response within minutes — the rep reviews and sends, or lets it auto-send on simple inquiries.
- Lead-scoring model trained on the group’s own CRM history
- Real-time re-ranking and routing across 11 showrooms, 4 brands
- AI follow-up assistant — first response within minutes, in the customer’s language
- Connector to the existing CRM (no replacement, no double data entry)
- Ops dashboard for sales managers: model performance, override rate, conversion by score band
Results
Within 60 days of launch, showroom-visit conversion rose 28% and time-to-first-contact dropped from a median of 11 hours to 9 minutes. Sales reps spent 40% less time chasing cold leads, and test-drive bookings per lead grew 34%. The model’s top score band now converts at more than 3× the bottom band, giving sales managers a clean prioritisation signal.
Why this worked in 60 days
The group had clean historical CRM data, a real bottleneck (response latency), and a willing pilot showroom. Visual-AI-Labs scoped the first cycle to scoring only, proved the lift in one showroom, then rolled out the follow-up assistant in cycle 2 across the group — without forcing a CRM replacement.
Discuss a similar project with Visual-AI-Labs →
FAQ
Did the dealer group have to replace its CRM?
No. Visual-AI-Labs added scoring and follow-up on top of the existing CRM via its API.
How much historical data was needed?
About 18 months of CRM history was enough to train a useful first model; the model improves with each subsequent month.
What about other brands joining the group?
The scoring model is brand-aware; adding a new brand is a configuration change, not a retraining from scratch.
Does the AI auto-send messages?
Only on simple inquiries (price availability, opening hours, test-drive booking). Anything sales-sensitive is rep-reviewed before send.
GDPR?
EU-only processing; lead PII never leaves EU infrastructure. Visual-AI-Labs builds against the group’s DPA.
Maintenance?
Quarterly model review by Visual-AI-Labs; ops tunes routing without engineering.
Can a single-showroom dealer use this?
Yes — the architecture scales down. The scoring uplift on a single showroom is typically smaller in absolute volume but proportionally similar.