Skip to main content
TurnellaBeta
WFM guideTransformation

Contact centre transformation

Contact centre transformation is a multi-year programme. The operations that treat it as a technology project — buy a new CCaaS platform, go live in 12 months — typically achieve the technical migration but not the operational improvement. The operations that treat it as a business change programme — technology plus process plus people, staged over 3–5 years — achieve the cost reduction, the quality improvement, and the WFM accuracy gain the business case promised.

Why contact centres transform

Technology end-of-life

On-premise telephony platforms (Avaya Aura, Cisco Unified Communications, Mitel) have typical support lifecycles of 10–15 years. Many UK contact centres are running on platforms that are end-of-support or near it. The cost of maintaining unsupported on-premise infrastructure accelerates as hardware ages and skilled support engineers become scarce.

Urgency

High for any operation on hardware >8 years old or platforms announced end-of-support.

Cost reduction requirement

The TCO (total cost of ownership) of on-premise contact centre infrastructure — hardware, software licences, maintenance contracts, in-house engineering — is typically 30–50% higher than equivalent CCaaS at scale. The per-seat flexibility of CCaaS also eliminates the on-premise model's spare capacity requirement.

Urgency

Medium — the saving is real but the migration cost must be netted against it. Typical payback on CCaaS migration is 2–4 years.

Digital and omnichannel capability gap

Legacy on-premise platforms were designed for voice. Adding chat, email, social media, and AI capabilities typically requires expensive third-party integrations on top of an aging platform — or building custom middleware. CCaaS platforms have these capabilities native, maintained by the vendor, and updated automatically.

Urgency

High for any operation with significant digital channel demand or self-service ambitions.

The 5-stage transformation roadmap

1

Assess and baseline

1–3 months

Key activities

  • Map current-state technology (ACD, WFM, CRM, QA platforms and their versions)
  • Document current processes (routing, scheduling, quality monitoring, reporting)
  • Baseline key metrics (AHT, SL, FCR, CSAT, cost-per-contact, attrition)
  • Identify gaps and pain points through operations, agent, and management interviews
  • Build the transformation business case with quantified benefit projections

WFM benefit

Establishes the baseline against which transformation ROI will be measured. Without a documented baseline, it is impossible to demonstrate transformation value. The WFM analyst is a key contributor to this phase — current planning accuracy (WAPE, adherence %, scheduling efficiency) must be documented.

Primary risk

Underestimating the complexity of existing integrations. On-premise telephony platforms often have undocumented customisations (call flows, routing logic, reporting queries) that are invisible until the migration attempt.

2

Telephony modernisation (CCaaS migration)

6–18 months

Key activities

  • Select CCaaS platform and negotiate commercial terms
  • Design target-state call flows and routing logic
  • Migrate IVR and call routing from on-premise to cloud (parallel run first)
  • Migrate or replace WFM platform (most CCaaS vendors offer native or partner WFM integration)
  • Agent training on new platform (typically 4–8 hours for CCaaS desktop)
  • Cutover: phased by skill group or location, not big-bang

WFM benefit

Cloud WFM integration typically improves adherence monitoring accuracy (real-time agent state feeds improve), forecast data quality (more granular contact data available), and scheduling efficiency (better optimisation tools). Expected WFM improvement: WAPE −2 to −5pp, adherence monitoring coverage +10–20%.

Primary risk

Big-bang cutover. A 500-agent operation that moves all agents to a new platform on a single day risks 2–4 weeks of service level degradation as agents adjust, reporting reconfigures, and issues are resolved. Phased migration is significantly safer.

3

Digital channel expansion

6–12 months (often overlapping with stage 2)

Key activities

  • Implement live chat on website and app (or migrate from legacy chat tool to CCaaS-native chat)
  • Implement email/ticket management within the CCaaS or integrated CRM
  • Redesign IVR with modern natural language capability
  • Add social media DM handling within the CCaaS routing layer
  • Train and certify agents on new channels (written skills assessment for chat/email)

WFM benefit

Opening digital channels reduces voice volume through channel shift. The WFM model must be reconfigured for each new channel — concurrency model for chat (2–3 simultaneous sessions per agent), backlog flow for email, Erlang C remains for voice. Initial channel mix forecast error is typically high; expect 3–6 months to calibrate.

Primary risk

Underestimating training requirements. Agents competent on voice who are assigned to chat without a writing skills assessment and 2–4 weeks of channel-specific training produce poor-quality written interactions — CSAT for chat suffers and channel shift does not deliver its intended benefit.

4

Self-service and AI integration

12–24 months (ongoing)

Key activities

  • Implement or upgrade IVR self-service for high-containment contact types
  • Deploy chatbot for digital self-service on top contact types
  • Integrate AI agent-assist (real-time knowledge, suggested responses, next best action)
  • Implement predictive routing using CRM contact history and customer segment data
  • Introduce automated QA/QC (AI-reviewed contacts alongside human QA sample)

WFM benefit

Self-service and AI should be reflected in updated WFM assumptions: lower arrival rate to agent queue (after IVR/chatbot containment), higher AHT on agent-handled contacts (complexity shift), reduced AHT for AI-assisted contacts (suggested next actions, knowledge pushed to agent in-call). These changes must be measured and validated before updating the planning model.

Primary risk

Vendor containment rate overstatement. AI chatbot vendors report containment (any contact that does not escalate to an agent). True containment (customer intent resolved) is always lower. Using vendor-reported containment for WFM planning overstates the staffing saving and creates recurring understaffing.

5

Data and analytics maturity

Ongoing from stage 2

Key activities

  • Build unified data platform connecting ACD, WFM, CRM, QA, and HR systems
  • Implement real-time operational analytics dashboard (beyond WFM platform native reporting)
  • Develop predictive workforce planning (machine learning volume forecast, attrition prediction)
  • Implement voice analytics / speech analytics for root-cause analysis at scale
  • Customer journey analytics: track cross-channel paths, identify friction and redesign

WFM benefit

Predictive analytics for WFM reduces WAPE from a typical 8–12% at weekly level to 4–7%, which directly reduces overstaffing buffer required. Attrition prediction allows proactive recruitment before attrition events rather than reactive hiring. Voice analytics identifies AHT drivers and knowledge gaps at scale, enabling faster AHT improvement than manual QA sample.

Primary risk

Data integration underestimated. Connecting ACD, WFM, CRM, and QA systems from different vendors through a unified data platform requires integration work that is consistently underestimated in transformation business cases. Plan for 12–18 months of data infrastructure work before predictive analytics are operational.

Common transformation failure modes

Big-bang cutover

What it looks like

The entire operation switches to the new platform on a single date. All agents, all routing, all reporting — live simultaneously.

Why it happens

Appealing in the business case (single cutover date, lower dual-running cost) and in the programme plan (milestone clarity). Extremely high risk in practice — defects, agent confusion, and data issues all hit simultaneously.

Better approach

Phased migration: one skill group or one site at a time, over 4–12 weeks. Run parallel on the old platform for the first 2 weeks on the new platform. Dual-running cost is almost always cheaper than a failed cutover.

Technology before process

What it looks like

Buy and implement the CCaaS platform first; redesign processes to fit the technology later. Or: implement AI chatbot on top of existing broken processes and expect the AI to compensate.

Why it happens

Technology decisions are faster to make than process redesign. Procurement timelines and vendor pressure create a sequencing problem — the platform is live before the processes are ready.

Better approach

Process redesign begins during stage 1 (assessment). Target-state processes are defined before the technology is selected. The technology is then configured to support the designed processes, not the other way around.

No change management

What it looks like

Transformation is communicated to agents and team leaders as a technology upgrade. The human factors — fear of job loss from AI, resistance to new systems, disruption of familiar workflows — are not addressed.

Why it happens

Change management is often the first casualty of programme budget pressure. Technology spend is visible and specified; change management is perceived as overhead.

Better approach

Treat agent and TL communication as a programme workstream from stage 1. Explain what is changing and why. Address AI displacement fears honestly (AI handles simple contacts; agents handle more complex, more interesting work). Involve frontline staff in the design of new processes and routing logic.

Business case on vendor containment rates

What it looks like

The AI chatbot ROI is built on the vendor's quoted containment rate (40–60%). This containment rate is applied to voice volume, producing a headcount saving that justifies the investment. After go-live, actual net deflection is 10–20%, the headcount saving does not materialise, and the transformation ROI case fails.

Why it happens

Vendor demonstrations use best-case containment scenarios. The gap between a controlled demo environment and a live operation handling full contact complexity is never visible in the sales process.

Better approach

Build the business case on independently verified containment rates from comparable operations (not vendor-provided). Model conservatively: 15–20% net deflection for a new chatbot on a mixed contact type portfolio. Capture upside separately as the chatbot matures.

Transformation questions

How long does contact centre transformation take?

3–5 years for a meaningful transformation in a 100–300 seat operation. Stage 1 (assessment): 1–3 months. Stage 2 (CCaaS migration): 6–18 months. Stage 3 (digital channels): 6–12 months. Stage 4 (self-service/AI): 12–24 months ongoing. Stage 5 (data/analytics): ongoing from stage 2. Big-bang 12-month programmes have high failure rates.

What is CCaaS and why are contact centres migrating to it?

Contact Centre as a Service — cloud-based contact centre software delivered as a subscription. Key vendors: Genesys Cloud CX, NICE CXone, Salesforce Service Cloud Voice, Amazon Connect, Twilio Flex. Migration drivers: no hardware, pay-per-seat flexibility, native omnichannel, faster feature access (AI/analytics deployed by vendor), native remote working, better WFM data quality.

Related guides