The claim is cheap. The capability is not.
Almost every outsourcing provider serving Australia and New Zealand lists a set of languages on its services page. Mandarin, Vietnamese, Arabic, Hindi, Cantonese, sometimes a dozen more. The list costs nothing to publish and tells a buyer almost nothing about what actually happens when a customer who speaks one of those languages calls in.
The gap between the list and the delivery is where most multilingual programmes disappoint. A provider can genuinely have an agent who speaks Vietnamese on staff, run a pilot with that agent, and still be unable to sustain a Vietnamese-language service once volume from Melbourne's or Sydney's Vietnamese-speaking customers actually arrives at scale, because one fluent agent is not a programme. The difference between a language on a list and a language the business can actually deliver against is depth, measurement and a specification that names what "capability" has to mean.
Australia and New Zealand are genuinely multilingual markets. Both countries have long-standing and growing communities who speak a language other than English at home, both have significant trade and tourism relationships with non-English-speaking markets across Asia and the Pacific, and New Zealand carries a specific obligation and opportunity around te reo Māori that most outsourcing relationships handle badly or not at all. A business that gets multilingual delivery right is not chasing a nice-to-have. It is serving the customer base it actually has.
This paper sets out why the common approach to multilingual delivery falls short, what a specification that actually works requires, and how a provider should staff and measure a non-English language so that capability holds up once real volume, not pilot volume, arrives.
Why the common approach falls short
Three shortcuts show up again and again in multilingual programmes that later disappoint, and each fails in a predictable, specific way.
The single-fluent-agent shortcut treats one bilingual staff member as proof the language is covered. It proves the opposite. One agent has no bench: they take leave, they leave the business, they cannot cover a shift outside their own hours, and the moment volume in that language exceeds what one person can handle, the business either queues those customers behind an English-first line or quietly reverts to a translated or machine-assisted response. Neither is the service the customer was promised.
The blended-measurement shortcut is the quietest failure and the hardest for a buyer to catch from outside. A provider reports one overall quality or satisfaction score across an entire programme. English volume, which is usually the majority, keeps that number healthy even while the smaller-volume, non-English-language interactions are performing poorly. The business cannot see a problem that never appears on the only dashboard it is given, and it has no way to know whether its Mandarin-speaking or Arabic-speaking customers are actually being served well or just being served quietly.
The proficiency-without-register shortcut hires for grammatical fluency and stops there. A candidate can pass a general language test and still sound wrong to a native speaker of that language in a customer service context, because formality, directness and the specific vocabulary a customer expects in a support conversation are not the same skill as classroom fluency. A fluent but mismatched register reads to the customer as a service that does not really understand them, even when every word used is technically correct.
Each of these shortcuts is a way of making a language claim cheaper to sustain than genuine capability actually costs. None of them survives contact with real customer volume in that language.
What a specification that works actually requires
A specification that produces real multilingual capability names the language precisely, requires an assessed level per agent, requires a bench sized for volume rather than for a pilot, and requires the language to be measured on its own rather than folded into an average.
Name the language and, where it matters, the variant. "Chinese" is not a specification; Mandarin and Cantonese are different languages with different writing conventions in some contexts, and a provider that treats them interchangeably will supply the wrong one to at least some customers. The same discipline applies wherever a language has meaningful regional variation relevant to the customer base being served.
Specify the level, per skill, against a recognised standard. Language capability should be assessed against a standard such as the Common European Framework of Reference, recorded per agent, and separated by skill, because speaking, listening, reading and writing do not move together and a voice role weights them differently to a chat or email role. The specification should set a minimum level for the role and require the provider to staff to it, not to a general claim of fluency.
Specify bench depth at the volume the business actually expects, not the volume of a three-week pilot. A provider that can field two fluent Vietnamese-speaking agents for a pilot needs to be able to state, before the contract is signed, how many it can field at the client's projected peak and what happens to that language's service levels if volume doubles.
Specify measurement per language. Quality, customer satisfaction and first-contact resolution should be reported separately for every language in the programme, each held to the same target as English, not folded into a single blended number. This single requirement is what keeps a weak language from hiding inside a healthy-looking average.
New Zealand's specific case: te reo Māori
New Zealand carries an obligation that most outsourcing relationships are not built to meet, and it is worth addressing directly rather than glossing over. Te reo Māori is an official language of New Zealand, and a growing number of New Zealand organisations, particularly in the public sector and increasingly in the private sector, want at minimum basic greetings and cultural acknowledgement handled with genuine care, and in some cases substantive service delivery in te reo.
Corpshore's own current language coverage for the New Zealand market lists Māori on a consultation basis rather than as a core, always-staffed delivery language, and that distinction is stated plainly here rather than rounded up into a bigger claim than the current bench actually supports. A business with a genuine te reo Māori delivery requirement, beyond acknowledgement and greeting-level use, should raise that requirement explicitly during scoping so it can be assessed and resourced honestly, rather than assumed to be covered by a general "New Zealand languages" claim the way it too often is elsewhere in the industry.
This is the same discipline the rest of this paper argues for, applied to the case where getting it wrong carries the most cultural weight: name the requirement precisely, state current capability honestly, and build the bench deliberately rather than claiming a capability that is not there yet.
How staffing and quality assurance actually hold this together
Specification sets the bar. Staffing and ongoing quality assurance are what keep a programme at that bar once it is live.
Assess at hire, against the specific language and register the role needs, not a general aptitude test. Every candidate for a language-dependent role should be assessed before an offer across speaking, listening, reading and writing, plus a role-relevant scenario, and recorded at the level actually demonstrated rather than rounded up because the team needs the seat filled.
Build the bench ahead of demand. A language capability that is recruited reactively, after a client signs and asks where the promised coverage is, is always behind, because qualified candidates in a specific language and register take time to source and assess properly. Recruitment for a language the business has committed to should begin before volume in that language actually arrives, not after service levels start slipping.
Maintain capability in role. Language proficiency is not a one-time hiring gate. It improves with structured use and drifts without it, so it should be reviewed periodically for agents already in role, with targeted coaching where it has slipped and recognition where it has improved.
Score language and process separately. A quality scorecard that conflates "did the agent follow the right process" with "did the agent's language use serve the customer well" cannot tell a business which problem it actually has, and the two need different fixes. Separating them is what makes the per-language reporting from the previous section actually actionable rather than just a number on a report.
What good looks like, and where Corpshore fits
A genuine multilingual operation is identifiable by specific, checkable things rather than by the length of its language list. It names each language precisely, including variant where that matters. It assesses every language-dependent hire against a recognised standard, records a level per skill, and re-tests in role rather than assuming a hire stays at the level they tested at two years earlier. It builds bench depth deliberately, ahead of a client's actual projected volume, and states current bench size honestly rather than inferring capability from a single strong hire. It reports quality, satisfaction and resolution separately by language, at the same standard as English. And where a specific requirement, such as te reo Māori beyond greeting-level use, exceeds current bench depth, it says so rather than claiming a coverage it cannot actually deliver.
The single most useful question a buyer can ask a prospective provider is a direct one: for this specific language, how many agents can you field at our expected peak volume, at what assessed level, and how is that language's quality measured separately from the rest of the programme. A provider with a real answer has a real programme. A provider that redirects to its overall satisfaction score does not.
Corpshore Australia delivers in English and a defined set of additional languages, sourced from the group's wider delivery footprint of 35 or more languages, staffed and assessed to the language and role a specific engagement actually needs rather than offered as a blanket claim. Reported client satisfaction across Corpshore's Australian and New Zealand engagements stands at 97 per cent, measured by independent survey. For a specific language requirement, current bench size and coverage are confirmed directly during scoping, the same discipline this paper argues every buyer should require of any provider, Corpshore included. A discovery call is the right place to have that conversation.