Unified Contact Centre for a State Government Agency

The Challenge

The agency operated twelve separate departments, each with its own telephony platform, queue configuration, and agent desktop. Citizens calling about complex matters — such as licensing that crossed departmental boundaries — were transferred repeatedly, sometimes four or five times before reaching the right person. There was no unified view of citizen history, no shared knowledge base, and no real-time reporting across departments.

This created measurable pain on both sides: citizens experienced long wait times and had to repeat their story at every transfer, while agents lacked context and had no visibility into what other departments had already done. Leadership had no consolidated data to drive service improvement decisions.

• 12 siloed telephony systems with no interoperability
• No single citizen profile — each department maintained its own records
• Citizens transferred an average of 3.1 times per complex enquiry
• Zero cross-department data sharing or case visibility
• Reporting was manual, delayed, and department-scoped only

Our Approach

DISquare began with a structured discovery phase, conducting stakeholder workshops across all twelve departments to map call flows, capture agent pain points, and identify the most common citizen journey patterns. This surfaced a clear architectural requirement: a unified telephony layer feeding into a single CRM, with the integration layer sitting between them.

From discovery, we moved into architecture design — aligning on Amazon Connect for cloud telephony, Salesforce Service Cloud for case and agent management, and Azure API Management as the integration backbone. The rollout was structured in three phases, starting with the four highest-volume departments to prove value quickly before scaling.

• Phase 1 — Discovery & Architecture (weeks 1–4): Stakeholder workshops, current-state mapping, architecture sign-off
• Phase 2 — Foundation Build (weeks 5–20): Amazon Connect configuration, Service Cloud Omni-Channel setup, Azure APIM integration layer, pilot with 4 departments
• Phase 3 — Full Rollout & Optimisation (weeks 21–32): Remaining 8 departments onboarded, skill-based routing tuned, queue management optimised, agent training delivered

Omni-Channel routing rules were designed around skill-based routing — agents were tagged with department expertise and service category proficiency, allowing the system to match citizens to the right agent on first contact. Queue management policies were configured to handle peak loads with overflow rules and callback options.

The Solution

The production architecture centres on three core platforms working in concert:

Amazon Connect replaced all twelve legacy telephony systems with a single cloud contact centre platform. IVR flows were designed to capture citizen intent upfront, routing based on both topic and the citizen's history with the agency. Screen pop on answer delivered the citizen's full profile to the agent before the conversation began.

Salesforce Service Cloud provides the agent desktop and case management backbone. OmniStudio FlexCards surface the right citizen data contextually — licence status, open cases, previous interactions — without agents navigating multiple screens. OmniScripts guide agents through structured processes (change of address, licence renewal, complaints) with branching logic that adapts based on citizen responses. DataRaptors handle the read/write operations between Salesforce data and the integration layer. Flow automation handles case creation, SLA clock management, routing escalations, and follow-up task generation.

Azure API Management acts as the integration fabric, exposing a consistent REST API surface to Salesforce regardless of what sits behind it in the agency's back-office systems. This decoupling meant departments could modernise their back-end systems independently without requiring changes to the Salesforce integration.

A shared knowledge base was built within Service Cloud Knowledge, with articles tagged by department, service category, and citizen type — surfaced contextually via Einstein Search during active cases.

Technical Architecture

The integration pattern follows a clean three-tier design: Amazon Connect at the telephony edge, Azure APIM as the integration mesh, and Salesforce as the system of engagement.

When a call arrives, Amazon Connect executes a Contact Flow that invokes an Azure APIM endpoint to look up the citizen's profile. APIM orchestrates calls to Salesforce REST APIs and relevant back-office systems, returning a unified citizen payload. This triggers a screen pop in the Service Cloud CTI adapter, presenting the agent with a FlexCard showing open cases, recent interactions, and account status.

Post-call processing is event-driven: Amazon Connect publishes call completion events which flow through to Salesforce Platform Events, triggering Flow automation for case updates, SLA enforcement, and supervisor dashboards. The MCP server handles orchestration for complex multi-system writes — for example, updating both a licensing system and creating a Salesforce case atomically.

The Results

Within three months of full rollout, the outcomes were measurable and sustained. Average handle time dropped by 40% as agents no longer needed to manually look up information across multiple systems — the screen pop and FlexCard surfaces delivered everything needed before the first word of the conversation.

First-contact resolution improved dramatically. Citizens were routed correctly on the first attempt 94% of the time, compared to a baseline of 61%, eliminating the bulk of cross-department transfers. The 3.2 million annual citizen interactions are now handled through a single platform with a 99.7% uptime record.

Agent satisfaction scores increased by 28 points in the post-implementation survey. Agents reported that the OmniScript-guided workflows reduced cognitive load and made onboarding new staff significantly faster. The structured process design also brought consistency — service quality no longer depended on which agent happened to pick up.

Citizen NPS improved by 22 points over the first six months. The primary driver was reduction in transfers and wait times — both of which were consistently cited in pre-implementation feedback as the agency's biggest service failure points.

Key Takeaways

Invest in the integration layer first. The decision to route everything through Azure APIM rather than building point-to-point Salesforce integrations paid dividends during rollout: each new department's back-office system was onboarded as a new APIM backend, with no changes required to Salesforce configuration.

Skill-based routing needs continuous tuning. The initial routing configuration was based on organisational structure. Post-launch analysis revealed that citizen enquiries rarely matched departmental boundaries — skills were progressively redefined around service categories rather than departments, driving the first-contact resolution improvement.

OmniStudio accelerated agent adoption. Purpose-built FlexCards and OmniScripts meant agents didn't need to learn a new system — the system guided them. This is especially valuable in government environments where agent cohorts span a wide range of technical confidence.

Platform Events unlocked real-time supervisory visibility. Supervisors moved from reviewing daily Excel exports to monitoring live dashboards. The shift to event-driven data flow was the single change that most transformed how service management operated.