Customer complaints arrive from email, phone, chat, social media, web forms and branches. Without a structured workflow, they scatter across inboxes and spreadsheets — generating SLA breaches, inconsistent responses, duplicated case files and regulatory exposure.

Worse, complaint patterns stay invisible: the same product defect or billing error can generate hundreds of complaints before anyone spots the trend. Utilyx models complaint management as an executable workflow with omnichannel intake, AI-assisted triage by sentiment and urgency, competency-based routing, investigation, resolution and post-resolution NPS.

The workflow is structured as a decision tree with 3 main branches (high-urgency negative, standard, low-urgency) and 7 steps end-to-end, plus two sub-branches (managerial review for detractors, testimonial request for promoters). Every step is executed by a named actor (Customer, AI Agent, Routing engine, Agent, Manager) and every action is appended to the timeline_complaint audit subtable. Unresolved complaints in 48h increase churn by 3x — omnichannel triage with sentiment AI prevents this.

• Step 1 (Start — Customer): the complaint is captured from email, phone, chat, social media, web form or in-branch into a single queue. Mandatory fields: customer_id, order_ref, product, channel, message. The AI copilot normalizes the message and extracts entities.

• Step 2 (AI Agent): the Copilot IA analyzes sentiment (positive/neutral/negative) and urgency (low/medium/high), categorizes the complaint (billing, product, delivery, service, legal), and deduplicates against open cases by customer_id + product + 30-day window.

• Step 3 (Routing): three conditional branches split the tree — Branch A (negative sentiment + high urgency): immediate escalation with email to customer service manager, Slack notification via call-connector and SLA timer 2h; Branch B (standard): competency-based routing via visual routing rules — no manual dispatch; Branch C (low urgency + neutral sentiment): suggested KB self-service, agent contact only if customer replies.

• Step 4 (Investigation): the agent documents root cause, links related orders and contacts, and records evidence. sql-query pulls the customer history (last 5 orders, prior complaints, NPS score). The findings are logged in the timeline_complaint subtable.

• Step 5 (Resolution): the proposed resolution (refund, exchange, credit, gesture) follows an approval chain based on amount thresholds — under €50 auto-approved, €50-€500 requires team lead, above €500 requires manager. The customer is notified in their preferred language — 7 languages supported natively. Sub-branch: if resolution_time > SLA_threshold, escalation to manager + re-assignment.

• Step 6 (usePDF): a formal complaint report is auto-generated from a Word template with merge fields (case number, timeline, findings, resolution) and converted to PDF for the case file.

• Step 7 (Closure & NPS): after resolution, an NPS survey is auto-triggered. Sub-branch A (detractor): the case reopens for managerial review within 24h. Sub-branch B (promoter): routed to a testimonial request. The complete case file (report, evidence, correspondence) is sent to useArchive for legal retention with index fields case_id, customer_id, category, channel, date.

Teams using Utilyx cut complaint processing time from days to hours, with first-response time reduced by 65%, a 45% drop in duplicate cases thanks to AI deduplication, and a 35% improvement in NPS within 6 months of deployment.

The 7-language KB search lets a single workflow serve a multilingual customer base, the visual routing designer lets a customer service manager re-prioritize branches without code, and the no-code workflow builder turns a plain-language description into a runnable CSM process — 60% faster deployment, 70% lower TCO, no consultants required.

Every action is timestamped and attributed, producing a full audit trail ready for consumer-protection regulators. useArchive enforces retention periods per complaint category and channel.

The 7-language support means a single workflow serves multilingual customer bases without duplicated configurations.