Beyond The Dial Tone · Vendor Signal · Updated July 2026, Original April 2026
Five platforms. Sixty voice and contact-centre capabilities. No universal winner.
The contact-centre market is converging, but it is not converging around a single product model.
Microsoft and Salesforce are bringing voice, digital engagement, CRM, workflow, data and AI into their existing enterprise platforms. Genesys and NiCE continue to develop from the opposite direction, adding more AI, orchestration and enterprise data capability to mature contact-centre operating suites. Amazon Connect occupies a third position: a cloud-native service that has steadily added packaged outbound, workforce and quality capabilities without abandoning its composable AWS foundation.
All five can now cover much of the same requirements list. That does not make them interchangeable.
The important differences increasingly sit beneath the feature name: how deeply the capability works, which product or licence provides it, where the operational state lives, how much configuration it requires, and what happens at enterprise scale.
The buying question is no longer which platform has the longest feature list? It is: which platform should own the operating model for customer service — and which capabilities must be proven under our workload?
What the benchmark does — and does not — say
The companion benchmark examines 60 capabilities across ten areas: voice connectivity; routing and queue control; IVR and self-service; outbound engagement; agent experience; recording, security and compliance; analytics and quality management; workforce management; administration, resilience and extensibility; and AI and orchestration.
This is deliberately not a league table. A predictive dialler, for example, is not just present or absent. Buyers may need to test pacing controls, abandoned-call management, answering-machine detection, list governance, regional regulation and performance with a blended workforce. In the same way, a tick beside workforce management says little about forecast horizons, schedule constraints, intraday controls or the scale at which the planning engine remains usable.
Public documentation establishes breadth. Production validation establishes depth.
The full capability catalogue, evidence rules and provisional public-evidence matrix are published in full: CCaaS 60-Capability Benchmark — methodology and evidence matrix.
The market now contains three operating models
1. CRM-native service platforms
Microsoft Dynamics 365 Contact Center and Salesforce Agentforce Contact Center both treat the contact centre as part of a broader service platform. Their strategic advantage is not merely that CRM and telephony share a screen. It is the possibility of placing the customer record, service workflow, knowledge, automation and AI agent inside the same control environment.
Microsoft combines Dynamics 365 Contact Center with Dataverse, Teams, Copilot Studio, Power Platform and Azure. Its current documentation covers unified routing, native voice, IVR and AI agents, transcription, sentiment, proactive voice engagement, predictive and progressive dialling (generally available since February 2026), workforce management and quality evaluation. That is a materially broader operational proposition than Microsoft offered at the product’s launch. (Microsoft overview, dial modes, workforce management)
Salesforce is moving quickly along the same path. Agentforce Contact Center now brings native voice, digital channels, CRM context and AI agents into Agentforce Service. Salesforce also documents proactive outbound engagement, contextual continuity between channels, workforce management, quality management and Agentforce Voice integration with native or partner telephony. Its current packaging explicitly includes workforce and quality management in the Agentforce Contact Center Plus package, with a separate workforce management add-on also listed. (Agentforce Contact Center, packaging, workforce management)
It would be misleading to assign Microsoft a fixed maturity lead. The two products have different strengths, packaging and release histories, and Salesforce is closing previously visible gaps quickly. Any comparison made in July 2026 could become stale within a release cycle.
For buyers, the practical distinction is contextual:
- A Microsoft-centred enterprise may reduce integration boundaries through Dynamics, Teams, Copilot Studio, Power Platform and Azure.
- A Salesforce-centred enterprise may gain the same kind of compounding effect through Agentforce Service, Data Cloud, Flow and the wider Salesforce customer record.
- Organisations using another CRM should examine whether either platform still simplifies the architecture once connectors, data synchronisation and operational ownership are included.
Neither ecosystem should receive an automatic advantage simply because the CRM and contact-centre products share a logo. The seam closes only when identity, data, events, workflow and operational controls are genuinely unified.
2. Operational contact-centre suites
Genesys Cloud CX and NiCE CXone — now branded CXone Mpower — begin from decades of contact-centre operations. That history matters where requirements are shaped by complex routing, blended inbound and outbound work, recording policies, workforce optimisation, quality processes, regulated operations, large multi-site estates or BPO delivery.
Genesys documents preview, progressive, power, predictive and agentless outbound modes; call analysis; campaign controls; recording; screen recording; speech and text analytics; quality evaluation; forecasting; scheduling; adherence and coaching. (Genesys dialling modes, quality and compliance, workforce forecasting)
NiCE CXone similarly combines Personal Connection outbound dialling with mature recording, quality, analytics, forecasting, scheduling, adherence and longer-term workforce planning. Its public documentation gives buyers considerably more operational detail than a high-level feature grid. (NiCE proactive voice, forecasting, recording)
The specialists should not be treated as “legacy” choices. Their operational depth can be the differentiator, particularly when service performance depends on controls that have been refined over many years. But their position is no longer automatic either. Buyers should test how cleanly each platform participates in the enterprise CRM, data and AI architecture. A technically excellent contact-centre suite can still create duplicate customer state, competing workflow engines or another integration boundary.
3. Composable cloud contact centre
Amazon Connect is often described as the platform for organisations that want to build. That remains partly true, but it is no longer the whole story.
AWS now documents native global telephony, predictive, progressive, preview and agentless outbound modes, answering-machine detection, forecasting, capacity planning, scheduling, adherence, screen recording, conversational analytics and manual or automated quality evaluations. (Amazon Connect features, outbound guidance, workforce management, evaluations)
Its distinction is therefore not simply “features versus no features.” It is the operating responsibility that accompanies an AWS-native service: architecture, security, observability, integration, release control and service ownership can remain more directly in the customer’s hands. That can be an advantage for an engineering-led organisation. It can be an unnecessary burden for one seeking a highly packaged operating suite.
Where convergence is already real
At the level of broad capability categories, the platforms now overlap substantially. All five can credibly participate in an enterprise shortlist for inbound voice and digital service; skills- and queue-based routing; IVR and conversational self-service; call recording and transcription; agent assistance and knowledge; supervisor monitoring and quality evaluation; forecasting and scheduling; APIs, workflow and integration; and generative and increasingly agentic AI.
This is why a conventional request for proposal can produce a misleading result. If every requirement is written as “supports capability X: yes or no”, several platforms will appear almost identical.
The difference emerges when the requirement becomes a scenario.
- Do not ask whether the platform supports handover from an AI agent to a person. Ask it to preserve the customer’s authenticated identity, objective, transcript, evidence, actions already completed and open commitments during the handover.
- Do not ask whether it supports predictive dialling. Ask it to demonstrate pacing under your expected staffing level, show how abandoned-call thresholds are controlled, explain regional restrictions, prove how campaign consent is enforced and confirm any vendor-documented regulatory caveats on representative-led dial modes.
- Do not ask whether workforce management exists. Supply anonymised historical demand, planning constraints and employee rules, then ask the vendor to produce and adjust a schedule.
Capability parity on paper is not operational parity.
A neutral buying framework
The following is a starting point for a shortlist, not a recommendation to select a product without validation.
| Operating context | Platforms that merit early examination | What must be proved |
|---|---|---|
| Dynamics, Teams, Power Platform and Azure are strategic | Microsoft Dynamics 365 Contact Center, plus an operational specialist where requirements are demanding | Integration boundaries, outbound controls, workforce depth, regional availability and total licensing |
| Salesforce is the service system of record and Agentforce is strategic | Salesforce Agentforce Contact Center, plus an operational specialist where requirements are demanding | Current generally available voice scope, telephony model, outbound depth, workforce limits and production references |
| Complex voice, outbound collections or blended campaigns dominate | Genesys Cloud CX and NiCE CXone; include Microsoft, Salesforce or Amazon where the latest releases meet mandatory scenarios | Dialler behaviour, compliance controls, routing complexity, campaign operations and CRM/AI integration |
| Recording, quality, workforce and regulatory operations dominate | NiCE CXone and Genesys Cloud CX; test other platforms against the same evidence pack | Recording policies, screen capture, retention, evaluation coverage, forecasting and auditability |
| AWS is strategic and the organisation has strong engineering ownership | Amazon Connect; compare with packaged suites to quantify ownership trade-offs | Build effort, operational support model, integration responsibility, regional coverage and lifetime cost |
| AI-led service transformation is the primary objective | All five should be considered | Tool execution, guardrails, evaluation, journey state, human handover, model portability and cost controls |
This framework intentionally allows more than one answer. The correct shortlist depends on which capabilities are mandatory, which ecosystem already owns the customer record and how much operational responsibility the organisation wants to retain.
The questions that now matter
The next generation of contact-centre selection should focus on six questions.
- Where does journey state live? When a customer moves from voice to messaging, from an AI agent to a person, or from the contact centre to a back-office process, which platform retains the authoritative state?
- Which product owns the action? Does the AI merely propose a refund, booking or account change? Which policy layer authorises it, which system executes it and how is the action reversed?
- Which capabilities are native in the purchased edition? A capability may exist in the wider vendor portfolio while still requiring another licence, product, region, implementation pattern or partner.
- What is genuinely generally available? Rapid release cycles create a mixture of established functionality, limited availability, preview capability and roadmap intent. These must not receive the same score.
- Can it operate under the real workload? Use representative volumes, agent counts, languages, regions, campaign types, retention policies, schedule rules and integration constraints during evaluation.
- What happens when a component fails? The most revealing demonstration is often not the happy path. Interrupt an AI tool call, make an external system unavailable or fail a specialist workflow halfway through. Then observe whether the platform notices, preserves state and recovers the customer journey.
The real conclusion
CCaaS convergence is real because the five platforms increasingly address the same broad territory. Parity remains elusive because capability names conceal differences in depth, packaging, operational control and architecture.
Microsoft has expanded rapidly beyond its initial CRM-native proposition. Salesforce is closing gaps at pace — its workforce engagement capabilities launched inside this assessment’s snapshot window — and should be assessed against what is available now, not against an older view of Service Cloud Voice. Genesys and NiCE retain substantial operational experience while continuing to add AI and orchestration. Amazon Connect has accumulated more packaged contact-centre capability than the “build it yourself” label suggests.
There is no stable, universal leader across all 60 capabilities — and publishing one would give buyers false certainty.
The useful decision is not which vendor wins the market. It is which platform should own customer context, interaction control, operational evidence and AI orchestration for a particular organisation. That decision begins after the feature grid, not inside it.
About the methodology: The accompanying benchmark defines all 60 capabilities, the evidence hierarchy, the scoring scale and a provisional public-documentation matrix. It is designed to be challenged and updated as products change.
Disclosure: This is an independent practitioner assessment based on first-party public documentation available on 24 July 2026. It is not a substitute for contractual confirmation, a security review or a production proof of concept. No vendor paid for inclusion or placement.