On a recent Customer Experience Live UK panel I was on, something came up that surprised the room. When customers were given a choice between a live agent and an AI agent to resolve a service issue, many chose the AI one.
That seems to cut against everything we hear about customers distrusting AI. But it doesn’t, really. It tells us something more precise — and more useful — about what customers want. They don’t want a human or a bot on principle. They want their problem solved. And they’ll take whichever one does that well.
Hold onto that, because it’s the key to a risk building quietly across customer experience right now.
The gains are real, and they’re fragile
Forrester’s Customer Experience Index shows that in the US and Canada, consumers’ perception of CX quality has now declined for four straight years, reaching an all-time low. The drivers won’t surprise anyone who works in this space: weaker employee experience, a fading focus on customer obsession, disappointing technology implementations — and a persistent gap between how executives rate their own CX and how customers actually experience it.
In fact, a closer look at the data shows that stagnation is still the dominant reality for the vast majority of companies. While headlines celebrate that 26% of North American brands improved, 68% saw absolutely zero change to their scores. In Europe, 91% remained completely flatline. So while the market has moved from a downward spiral, it has improved somewhat into a heavy holding pattern.
After four years of decline that drove CX quality to an all-time low, 2026 brought early, fragile signs of recovery — Forrester’s CX Index found 20% of brands improved globally, with stronger gains in North America (26%).
But Forrester is careful to call this incremental and uneven, not broad progress. That hard-won, tentative recovery is exactly what’s now at risk.
That last one matters more than it looks. When leadership believes the experience is better than customers do, the organization stops focusing on the things that would improve it. That gap is not a measurement problem. It’s an alignment problem.
So CX quality is hard-won, slow to improve, and easy to erode. Which is exactly why the next part should give leaders pause.
The fastest way to undo that progress
Forrester’s 2026 predictions include a warning that should land on every CX and revenue leader’s desk: in 2026, a third of companies will harm customer experiences with frustrating AI self-service. The pressure to cut costs, Forrester says, will push companies to deploy customer-facing AI chatbots and virtual agents prematurely — in contexts where they’re unlikely to succeed — eroding both the brand and the customer relationship, and damaging acquisition and retention.
Read that alongside the four-year decline and the risk becomes clear. Organizations that have been slowly improving CX quality could undo years of progress in a single rushed AI rollout.
We’ve already seen what this looks like. Air Canada’s chatbot promised a customer a bereavement discount, and a tribunal held the airline responsible for what its bot promised. DPD’s chatbot was manipulated into swearing at a customer, and the screenshots spread across the internet within hours.
Neither of those failures was an AI problem. They were governance problems. The technology did exactly what it was told to do inside a system that hadn’t been properly designed, tested, or supervised.
These aren’t isolated incidents. According to data from the 2026 Sinch AI Production Paradox Report, 69% of retail organizations have already been forced to completely roll back a deployed AI agent. The primary drivers for these emergency rollbacks were severe governance failures, critical hallucinations, or data privacy leakages. Rushing a rollout is no longer a calculated risk; statistically, it is an operational liability.
Why customers embrace some AI and reject other AI
This is where the panel finding comes back in. Customers chose the AI agent because, in that context, it worked. It resolved their issue quickly and well. That’s not a contradiction of the trust research — it’s the resolution of it.
Edelman’s 2026 Trust Barometer places the growing use of generative AI among the forces shaping trust, though not the largest — about one in ten point to it as the biggest single influence on their trust. More tellingly, significant shares of lower- and middle-income people say they expect to be left behind by GenAI rather than benefit from it.
The dividing line isn’t human versus machine. It’s competent, governed AI versus careless, rushed AI. One earns trust. The other burns it — faster and more visibly than a broken human process ever could, because AI operates at speed and scale. Point it at a broken experience and it will deliver that broken experience to every customer, instantly.
Trust is built inside the organization before customers ever feel it
Here’s the part most cost-driven AI rollouts miss. A trustworthy customer experience isn’t something you bolt on at the customer-facing layer. It’s produced by how the organization is aligned behind the scenes.
Remember Forrester’s point about the gap between how executives and customers perceive CX. That gap is a symptom of misalignment — teams working from different definitions of “good,” owning disconnected pieces of the journey, optimizing for their own metrics rather than the customer’s outcome. When that’s the reality underneath, no amount of AI fixes it. AI just automates the disconnection and ships it to customers faster.
This is why journey orchestration matters, and why it’s more than a technology purchase. Customers don’t experience your org chart. They experience one continuous journey across teams that often don’t talk to each other. When those handoffs are broken, the customer feels the seams — and every seam is a small withdrawal from the trust account. Orchestrating the journey means aligning the teams, the data, and the decisions behind it so the experience holds together from the customer’s point of view, not the organization’s.
Forrester’s own research bears this out. As Keith Johnston, VP and group research director, puts it: “when companies align their brand promise with the experiences they deliver, they create a compelling total experience…” — one that generates higher revenue, retention, and customer value.
This isn’t just theory; it is backed by Forrester’s 2026 Global Total Experience Score Rankings. For the first time, Forrester mathematically combined its CX Index and BX Index with a brand-new Employee Experience Index (EX Index™). The data proved the power of internal alignment: while traditional standalone CX scores trickled up incrementally, 53% of US brands significantly elevated their unified Total Experience Scores. The brands winning the market are doing so by fixing employee tooling and brand alignment behind the scenes before turning on public-facing software.
Disconnected, brand and experience send conflicting signals. Orchestrated together, they multiply.
That alignment is also what makes AI safe to deploy. Which brings us to what to actually do.
What to do before you scale AI
1. Check whether behaviors match the metrics — and the outcomes.
Look honestly at whether your customer interactions match your tracking dashboard. Crucially, stop relying on legacy metrics like Average Handle Time (AHT) or basic CSAT to evaluate automated channels. These metrics are fundamentally broken when assessing AI self-service. If an AI bot leaves a customer in circles, your AHT might look excellent, but your retention is dying. Instead, shift your Center of Excellence KPIs to track AI Containment-to-Resolution Rates, Frictionless Handoff Speed, and Customer Journey Progression.
2. Align your data.
You can’t orchestrate a journey you can’t see. If customer data is scattered across teams, tools, and spreadsheets that don’t connect, you have no single view of the experience — and no way to know whether AI is helping or harming it. Getting the data aligned is a precondition, not a nice-to-have.
3. Align your teams — build a center of excellence.
Bring product, technology, operations, compliance, service, and CX leaders into one place where AI strategy, governance, and best practice live together. This is where human oversight gets built in before deployment, not bolted on after a failure. It’s how you make sure an AI rollout serves the customer’s outcome rather than just the cost line.
4. Ask customers directly — and check your own truth.
Run a trust survey. Ask customers what they actually experience, and compare it honestly against what your organization believes it delivers. Closing that executive-customer perception gap is where turning the trend around begins.
The advantage is there for whoever earns it
We are operating in an AI-first world, where everything happens at speed and scale. That cuts both ways. Deploy AI on top of a broken, misaligned experience and you scale the damage. Deploy it on top of an aligned organization with real human oversight and you scale the value.
Forrester’s own framing for 2026 is that trust and value will be the guiding beacons, and that superficial efforts won’t cut it anymore. That’s the opportunity. CX quality has been declining for four years, and a third of companies are about to make it worse by rushing AI to cut costs. The organizations that resist that pressure — that align first, orchestrate the journey, and put oversight in place before they scale — won’t just avoid the damage. They’ll pick up the customers everyone else is about to lose.
That’s not a technology advantage. It’s an operational one. And it’s available to whoever is disciplined enough to earn it.
No comments yet.