Conversational Intelligence Dashboard

Multilingual Intent Classification & Conversation Routing — Orbit Commerce

Markets: 34
Languages: 9
Period: Last 90 Days
North Star Metric
76%
Global FCR (Weighted)
-3.2% vs target
DE Intent Accuracy
81%
-7% vs EN
Parity Gap
14%
-4% vs Q1
Transfer Rate
11%
+2.1%
Translation Latency
2.3s
±0.1s
Misroute Rate
7.8%
+1.3%
Code-Switching Rate
4.2%
+0.4%

First-Contact Resolution by Locale

Resolution Parity Gap Trend (vs EN Baseline)

Intent Classification Accuracy by Locale × Intent Type

Billing
Technical
Account
Returns
Shipping
General
EN
92%
89%
91%
88%
90%
86%
DE
74%
82%
85%
79%
81%
83%
FR
84%
81%
87%
83%
80%
82%
ES
80%
78%
82%
86%
84%
81%
IT
83%
80%
85%
82%
79%
84%
PT
81%
76%
83%
80%
82%
78%
JA
87%
88%
90%
86%
89%
85%
≥86% (Target)
78–85%
<78% (Critical)

Top 10 Misclassified Intent Clusters (Last 30 Days)

Locale Intent (Actual) Predicted As Volume Accuracy Impact (CSAT)
DE Billing Issue Technical Support 1,247 68% -0.8
DE Account Access Password Reset 892 73% -0.5
FR Return Request Cancellation 734 76% -0.6
PT Technical Support Product Question 621 71% -0.7
ES Shipping Delay Order Status 589 78% -0.4
DE Refund Status Payment Failed 512 74% -0.5
IT Product Question General Inquiry 487 80% -0.3
FR Subscription Change Upgrade Request 453 82% -0.2
ES Billing Issue Invoice Request 412 79% -0.4
PT Order Modification Cancellation 389 81% -0.3

Translation Latency vs. CSAT Impact (by Locale)

Alert: German (DE) billing-intent misclassification spiked to 19% in Week 7 following NLU model update. Resolution parity gap vs EN baseline widened to 18%, triggering executive review. Rollback initiated; retraining in progress with dialect-specific corpus augmentation.