Zum Inhalt springen
2 / 5 seats · Q3 2026
Slot
Case Study № 02 · Support Agent

200 tickets in the Monday-morning jam — today it’s 12.

The support inbox filled up over the weekend with standard requests — shipping, returns, invoice PDFs. Monday morning the team faced 200+ open tickets. Today the agent resolves 94% of them before the team comes online.

Year
2026
Industry
E-commerce · DACH
Stack
LangChain · RAG · Zendesk
Duration
4 weeks
Status
Live · production
§ 02 · The problem
What didn’t
work.

The client — a German e-commerce mid-sized company — had ~280 tickets/workday. 78% were Tier-1 standard requests (shipping status, return, invoice re-issue). The support team spent 65% of its time on questions whose answer was in the FAQ — which customers didn’t read.

  • 280 tickets/workday · 78% Tier-1 · 22% complex
  • 65% support time on FAQ answers (shipping · returns · invoices)
  • 14 min avg handling time Tier-1 → 28 h/day staff effort
  • CSAT 3.4 / 5 — because complex tickets waited too long
Pipeline · Live

Support Agent · ticket resolution

Ø Zeit-Ersparnis

Avg. 13.8 min saved per Tier-1 ticket

01
Request
Email · chat · form
02
KB search
RAG · 1,200 articles · confidence
03
Classification
Tier-1 / Tier-2 · intent
04
Answer draft
Brand voice · DE/EN · personalized
05
Confidence check
> 85% autonomous · < 85% human
06
Resolution
Auto-answer or escalation
§ 03 · Result
What changed after 6 months.

Sechs harte Zahlen. Aus den Live-Dashboards des Kunden, gemessen über die ersten Monate nach Go-Live. Kein Marketing-Number-Massaging — Roh-Output aus dem Agent-Log.

KPI № 01
94 %

Tier-1 resolved autonomously

of 78% of all tickets

KPI № 02
12 s

avg. Tier-1 response time

was 14 min

KPI № 03
4.7 / 5

CSAT

was 3.4

KPI № 04
−28 h/day

support time saved

available for complex tickets

KPI № 05
0

Monday-morning hotline jam

first in 3 years

KPI № 06
€45K/yr

avg. savings

secured staff costs

§ 04 · Voice from the workshop
My team had a 200-ticket backlog every Monday morning. Today the dashboard shows 12 tickets in the queue — all Tier-2, because the agent already handled the rest. My team isn’t just more relaxed — our CSAT is high too.
Head of Customer ServiceE-commerce · 150 employees