No Jitter is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

The Enterprise CCaaS Challenge: Architecting for the Cloud

A frequently asked question at Sytel is, “Contact center as a service (CCaaS) has arrived, but at what point should I if at all, plan to move away from legacy systems and adopt it?”
We try to use questions like this as a teaching moment. CCaaS has arrived. But for the most part, the service offerings available deliver a simple offer. For example, if you want basic call and media queueing, CCaaS can help you achieve this. If you want to run an IVR/chatbot campaign, you can do that too.
Delivering multisession and multi-channel that integrates seamlessly with your business processes from the cloud is a much bigger ask if you want an enterprise-grade blended automatic call distribution (ACD) that includes dialer and social media. Most major brand CCaaS offers (and we’re talking about genuine CCaaS offers, not legacy ACD installed on a few VMs in Amazon Web Services (AWS), follow the stateless microservice pattern. An offering like this enables linear scale, but at the cost of not doing everything expected of an enterprise solution.
Is there a workable enterprise CCaaS solution?
Sytel has been grappling with this for nearly two decades; we were early to cloud working with partners like CosmoCom, an Internet service provider, and have spent many hands-on years on the problem since.
It isn’t enough to have a microservice design; you must look at how everything fits together and deploys, what bits of the solution must be in the core platform, and what interfaces get exposed for business integration. The following are startling conclusions to our research:
  • The ACD can’t do everything expected of an enterprise ACD and still be stateless. You have to use an actor model for simplicity, but the actors that make up the ACD/dialer ‘brain’ have to be in process – more on why in a follow-up article. A full-blown ACD, therefore, has to be stateful. This flies in the face of cloud thinking and presents some challenges for resilience and scale. Solving this problem requires a different approach.
  • Real-time service level agreement (SLA) adherence and load-balancing have to be part of the core ACD. Interfacing workforce optimization over the public cloud in real-time is not practical.
  • You must design multisession behaviors into the ACD. It’s easy enough to have an agent desktop environment that allows you to interact with chat and email while handling voice. But, if you don’t have an orthogonal approach for managing multisession across all media types as part of the ACD, this negates the ability to do automated blend and service level management.
  • Not only does the cloud ACD need to do everything a mature ACD solution does, but it also must deal with multisession agents and media with different behaviors. Because of this level of complexity, supervisors are no longer capable of managing resources in real-time. The ACD must evolve to become the Automatic Session Distributor (ASD).
  • Data-intensive operations such as list management and customer data integration need a different approach to cloud orthodoxy. You must include caching and TP monitor behaviors in the design. The design has to work for both customer data held in the cloud and enterprise data centers.
  • APIs need to be designed around business integration touchpoints and continue operating when these integrations go offline or don’t respond promptly.
  • Orchestration needs to be intelligent to take account of communication and processing load requirements of all the different components that make up a full-service CCaaS offer. It isn’t enough that the tenant portal is the native web. Landlord and system administration also has to be native web.
Despite all of the above, the short answer to the question is ‘yes, there is a workable enterprise CCaaS solution.’ It requires bold decisions on how to architect for the cloud and a DevOps mindset that delivers continuous improvement over what for any vendor is a multi-year research and development cycle. Sytel’s first global enterprise customer went live with our cloud platform in 2020 and rolled it out across 42 countries and counting. This rollout came on the back of many successful deployments with smaller (300-2000 concurrent agents) CCaaS providers and the learning experiences this gave our company.
Automation is the key to success, and this means blending across all media services provided by the ASD. This combination ensures that whatever the shape of your workloads, the ASD does the best possible job of keeping to SLAs and delivering maximum human agent utilization. The benefit of automation is supervisors can spend their time managing and mentoring their teams to raise output quality instead of juggling agent allocations.
As a final thought, perhaps a question to ask any CCaaS provider is ‘Do your enterprise customers use your cloud offering to deliver their contact center requirements end-to-end?’ Many enterprise customers use CCaaS to deliver point solutions for short-term projects, but the leap to a comprehensive cloud contact center backbone is still more of a promise than an achievement!