Having observed CX strategies collide with operational reality, a recurring issue keeps resurfacing: leaders ask for “always-on” service, in their modern contact center, but what they actually need is reliable, consistent, and safe service at any hour, delivered without quietly compromising the people and systems that make customer experience possible.
That tension sits at the heart of my interview with Nerys Corfield, Director at Injection Consulting, who has seen contact centers attempt to stretch daytime service models into the night using a patchwork of bots, email SLAs, and under-resourced escalation paths. The result is often a service that looks “24/7” on paper, but feels unpredictable in practice, both for customers and advisors.
Nerys put it plainly: always-on service is not a marketing statement, it is a commitment to parity. It means “offering customers the same thing at 1am as they would get at 1pm.” That definition is deceptively simple, and that’s exactly why it is so hard to deliver.
Always-On Is Not a Binary, It’s a Design Decision
A lot of organisations still treat always-on service like a switch: either you have it or you don’t. But Nerys argues the reality is messier. Some businesses are scaling back live advisor hours, particularly in retail and finance, and “supplementing with AI overnight.” That move can be smart, but only if the organisation is clear-eyed about what customers actually need out-of-hours.
Her challenge to leaders is direct: why do you want always-on service, and what evidence do you have customers demand it? In her words, it is about “getting under the skin” of the request:
“Why do you want an always on service? Why do you think your customers want or need it? What evidence have you got that they do?”
The uncomfortable truth is that some brands chase always-on because it sounds modern, not because it is materially improving outcomes. Always-on can absolutely be necessary, healthcare and transport are obvious examples, but in other sectors it can be expensive theatre.
The Retailer Reality Check: Always-On Is Not Everywhere, Even in Peak Season
One of the most telling moments from Nerys’ interview came from a simple experiment. Late one night in the week before Christmas, she called four major UK retailers: John Lewis, Marks & Spencer, Boots, and Next. She expected to find live support available across the board. Instead, only two out of the four offered access to an advisor at that hour, and in two cases there was not a strong digital alternative for urgent needs.
The point is not to criticize the brands; it is to illustrate the gap between what the market says it is doing and what customers actually experience.
The reality of service coverage was inconsistent, and that inconsistency shapes trust. If customers cannot predict how quickly they will get help, they tend to over-contact, repeat themselves, or escalate through other channels. Every one of those behaviors raises cost.
This is why always-on is less about being “open,” and more about being intelligently responsive.
AI Coverage Is Not the Same as Service Parity
AI is now central to the always-on story, and it is also where organisations risk the biggest credibility gap.
Nerys’ view is that many overnight AI deployments currently provide a diminished service, because excellent customer service still depends on the option to reach a human when needed. As she put it, “you aren’t getting a parity of service because… acceptable customer service is always supplemented… with a live advisor.”
That is not anti-AI, it is pro-design. You can improve containment, add voice AI, and strengthen knowledge grounding, but customers still need confidence that when the issue is emotional, urgent, or complex, the system will not trap them in loops. This is why Nerys keeps returning to one operational rule: an always-on model must include the ability to “pivot out to live.”
It is also why she insists organisations first understand their daytime experience, and then deliberately model how that experience should change out-of-hours.
If the daytime experience is not good enough, extending it simply spreads weakness over more hours. Her advice is practical: “think about what their day service looks like and then… how they’re going to emulate that and deliver it.”
The Frontline Truth: Always-On Fails First in the Eyes of Agents
Always-on strategies are often justified as customer-first. But when they are designed poorly, agents are the first to feel it, and customers feel it next.
Nerys described what this looks like on the frontline: “this horrible confusion in their eyes.” Overnight teams, in particular, operate with “scant resources for support,” and they often have to be staffed with “net new resources” because day teams rarely shift into overnight work sustainably.
Operationally, that creates additional risk, from safeguarding and transport considerations for in-office overnight staff, to the “hangover” effect on morning teams who have to deal with unresolved overnight issues and automation fallout. Workforce planners then inherit the mess, because forecasting becomes harder when AI containment is inconsistent and escalation volumes spike unpredictably.
This is the always-on pattern that quietly burns teams out: it is not the hours alone, it is the lack of clarity, training, and support at the exact moment an interaction becomes hard.
Always-On Is Accelerating Cloud Migration, But Not for the Reasons Vendors Prefer
There is a temptation to frame contact center modernization as a technology refresh. In reality, it is a business pressure release valve.
Enterprises are accelerating CCaaS adoption because legacy platforms struggle with scale and resilience, and because always-on expectations increase the cost of downtime, integration delays, and peak handling failures.
There are common blockers: high operating costs, slow digital integration, seasonal spikes, and talent retention, which are all amplified under a 24/7 expectation set.
What is changing in 2026 is that AI has become the forcing function. Nerys said it bluntly: “It is all about AI. It is all about the noise around AI.” Some organisations still do not know what they want AI for, but they are aware their existing platforms will not support modern capabilities, from auto-summarization to automated quality management.
That motivation is echoed in a recent interview with Steve Blood, VP of Market Intelligence at Five9, who framed legacy modernization as a disruptive but necessary step. His analogy is memorable because it captures why people delay:
“We’re going to the dentist… It’s good for me. I know I should go, but I don’t know what they’re going to find.”
Steve also referenced industry survey findings that the top drivers for changing platform include cost and a lack of AI capabilities, and he warned that “doing nothing” is not a realistic strategy for CX leaders being pressured to adopt AI.
The Hidden Cost of Legacy Is Not Licensing, It’s Lost Visibility
Steve Blood goes on to describes the hidden “leak” of staying on-prem. It is not only the known costs, it is the unknown costs.
In cloud environments, transcription and large language model capabilities increasingly come embedded, which makes it easier to surface insights about customer sentiment, operational effectiveness, and what is going wrong across journeys. Steve described this as “gold,” because it answers the questions contact centers have historically struggled to quantify: what are we doing wrong, and what is it costing us?