01 · Starting Point & Bottleneck
Five- to ten-fold ticket spikes inside a shared inbox.
Reifen24 is a brand of Goodwheel GmbH in online tyre retail serving B2C and B2B customers. Each spring and autumn, inquiry volume jumps five- to ten-fold within days as the entire market switches to summer or winter tyres at once.
Previously, all customer inquiries arrived in one shared inbox with no fixed categorization, prioritization, or SLA tracking. Hiring temporary staff solely for two seasonal peaks per year was not economically viable. The goal was to relieve the existing support team so it handles peak volume calmly and reliably.
02 · Before → Today
From a shared inbox to pre-qualified peak-season support.
Before
Double-digit hourly waits during seasonal peaks
- Every spring and autumn, tyre-retail ticket volume surges by a factor of 5 to 10.
- All inquiries landed unsorted in one shared inbox handled on a “whoever sees it first” basis.
- Response times stretched into double-digit hours; escalations ran verbally with no metrics.
- Adding temporary headcount for a few weeks each year was not commercially viable.
Today
Pre-classified tickets, vetted drafts, and clear SLAs
- After pseudonymization, an EU AI model assigns incoming tickets across four categories at over 90% accuracy.
- Zoho Desk routes automatically to the right role and populates standard templates with order data.
- The support team reviews and sends; peak-season response times dropped by roughly 60%.
- Monthly SLA and category reports show management clearly where volume or training needs grow.
03 · Architectural Decisions
Automate the prep work, keep accountability with the team.
Reviewing historical inquiries revealed four clear categories: delivery status, claims, product advice, and invoicing or payment questions. Across the first three categories, more than 70 percent were recurring standard requests.
-
01
Zoho Desk and Zoho CRM as structured helpdesk
Clear ticket categories, role-based routing, per-category SLAs, escalation rules, and automated monthly statistics replace the shared inbox directly alongside customer context.
-
02
GDPR-compliant EU AI with upfront pseudonymization
Personal data is pseudonymized prior to every AI call. Classification runs exclusively on a model with EU data residency and a documented DPA, audited and signed off by Reifen24’s Data Protection Officer.
-
03
Human-in-the-Loop instead of blind auto-sending
In the tyre trade, the retailer carries liability for incorrect product advice. AI therefore produces only the classification and pre-filled draft; a specialist verifies variables such as order number, delivery date, and tyre model before sending.
04 · Rollout & Daily Operations
Live production in three phases within 6 to 8 weeks.
- Phase 1 · Zoho foundation
- Configuration of Zoho Desk and Zoho CRM with categories, routing rules, SLAs, escalation logic, and automated metrics, paired with two weeks of team training.
- Phase 2 · EU AI tuning
- Training and fine-tuning the EU model on the first few hundred real tickets over two weeks until classification accuracy stabilized just above 90 percent.
- Phase 3 · Templates & DPO
- Joint creation of reply templates with placeholders for order number, delivery date, and tyre model, plus formal sign-off on pseudonymization, EU residency, and DPA by the Data Protection Officer.
In the first seasonal peak after go-live, response times ran roughly 60 percent below the prior year at comparable inquiry volume and unchanged team size. Specialists now focus on complex product advice and claims while management receives precise monthly SLA data.