Most international brands now have AI somewhere in their customer service: a chatbot handling first contact, an AI copilot drafting agent replies, a voice system triaging calls. It typically rolls out as one platform, designed centrally and pushed out market by market. What doesn’t travel with it is any guarantee that the experience lands the same way in every market.
The trust problem is real, and it’s getting worse
Recent consumer research paints a consistent picture. In the US, the large majority of consumers say they would rather speak to a person than an AI agent, and half say they would cancel a service that relied on AI alone. Most believe AI in customer service exists mainly to cut costs rather than to improve it. Overall trust in AI has fallen since 2023, and nearly all consumers now expect a clear explanation when an AI system makes a decision that affects them, a standard most organisations can’t yet meet. Barely one in ten consumers say they trust AI completely.
None of this means AI-assisted service is a mistake. It means the gap between the experience a brand designs at head office and the experience a customer actually receives, market by market, matters more than it did two years ago, and it’s harder to see from head office than ever.
Why this matters once you operate across markets
An NPS dashboard or CSAT survey tells you whether a customer was satisfied. It doesn’t tell you whether the AI disclosed itself when it was supposed to, whether the customer could actually reach a human when they asked, whether the answer given in Warsaw matched the policy given in São Paulo, or whether the local language handling was good enough to be trusted. Those are execution questions, not sentiment questions, and they only get answered by someone actually going through the journey.
What to check, market by market
- Is AI use disclosed to the customer, where local rules or your own policy require it
- Can the customer reach a human quickly when they ask to, or when the issue needs one
- Does the answer given match your brand’s policy and script, consistently across markets
- Is escalation seamless, or does the customer have to repeat themselves from the start
- Is language and tone good enough in each market to be trusted, not just technically correct
The gap between the experience a brand designs at head office and the experience a customer actually receives, market by market, matters more than it did two years ago.
One platform, five different experiences
A global retail bank rolled out the same AI-assisted service platform across five markets in a single quarter, expecting a broadly consistent customer experience wherever it landed. A global experience audit run across those markets during the rollout found the platform performing quite differently market to market: in two markets the AI didn’t disclose itself as local rules required; in one, customers asking for a human were routed through three extra prompts before reaching one; in another, the tone read as brusque rather than reassuring once properly translated. None of this showed up in the CSAT scores, which looked broadly similar everywhere, because the customers who gave up before resolution weren’t the ones filling in a survey afterwards. The findings let the bank fix the disclosure and escalation issues before a regulator, or a competitor, found them first.
This example is anonymised in keeping with our client confidentiality agreement.
Where this usually goes wrong
Brands assume that because the platform is the same everywhere, the experience is the same everywhere. It isn’t. Local language quality, local escalation paths and local compliance requirements all vary by market, and none of that shows up in a global dashboard. The only way to know what’s actually happening is to check it directly.
