Multilingual Website Governance for International NGOs | Socialectric

Most international NGOs treat translation as the solution to their multilingual website problem. Install Weglot. Translate the pages. Publish. Job done.

Except it is not done. Six months later, the English site has been updated with new programme descriptions, a revised governance section, and a current annual report. The French and Spanish versions still show the previous programme structure, the old governance language, and last year's report. Nobody was assigned to keep the translations current. Nobody defined which pages must exist in all languages and which are English-only. Nobody established what happens when a translated page contradicts the English original because one was updated and the other was not.

Translation is a technical problem. Governing a multilingual website is an institutional one. And most international NGOs solve the first and ignore the second.

Where multilingual governance breaks down

The governance failure follows a predictable pattern. The organisation decides to offer its website in multiple languages, usually because a funder requires it, a programme region demands it, or a new strategic plan includes international visibility as a goal. A translation tool or service is engaged. The existing English content is translated. The multilingual site goes live.

Then reality intervenes. The English content is updated by the Communications Director as part of normal operations. Nobody notifies the translation team. Nobody checks whether the translated versions need updating. Nobody has been assigned to monitor content parity across languages. The result is that the English site is current and the other language versions gradually drift into inaccuracy.

For an organisation serving stakeholders in multiple countries, this is a credibility problem. A programme partner in Senegal reading the French version of your site sees different information than your London-based funder reading the English version. An institutional funder reviewing both will notice the inconsistency and question which version is accurate. Neither outcome serves the organisation. The scale of the opportunity is significant: according to Weglot's multilingual statistics, English reaches only approximately 19% of the world's population, yet 49.4% of websites exist only in English. For international NGOs, this means the decision to go multilingual is almost always the right one. The failure is not in making that decision but in governing what follows.

What multilingual governance requires

Effective multilingual governance addresses four areas that translation tools do not cover.

The first is content parity policy. Not every page needs to exist in every language. A governance policy should define which content must be maintained in all supported languages (mission, programmes, governance documents, contact information) and which is English-only (blog posts, detailed technical resources, internal-facing content). This prevents the expectation that everything will be translated while ensuring that the content stakeholders in each language actually need is always current.

The second is update workflows. When an English page is updated, who is responsible for triggering the translation update? Is it the Communications Director? The translation service? An automated system? Without a defined workflow, translation updates depend on someone remembering. And memory is not a governance mechanism.

The third is quality assurance. Machine translation, including the tools integrated into platforms like Weglot, produces output that is functional but not always accurate in context. The translation landscape has shifted dramatically: the 2025 Enterprise Content and AI Translation Benchmark Report found that machine translation and hybrid workflows now account for approximately 65% of all translation volume, a reversal from 2021 when human-only translation dominated at 72%. This means most organisations are already relying on machine translation to some degree. The governance question is not whether to use it but how to manage quality. Legal language, programme-specific terminology, cultural references, and institutional tone all require human review. Who reviews translated content before it goes live, and how often?

The fourth is legal and regulatory compliance. Different jurisdictions have different requirements for what must appear on an organisation's website in the local language. Data protection notices, cookie consent, charitable registration disclosures, and accessibility statements may all need to exist in the languages your site supports. A GDPR compliance framework that only exists in English does not serve a French-speaking user whose data you are processing.

The stakeholder dimension

Multilingual governance is ultimately a stakeholder prioritisation question. Which stakeholders need content in which languages, and what content do they need?

An international NGO whose primary funders are English-speaking foundations but whose programme beneficiaries are in francophone West Africa has a clear priority: governance and accountability content in English for funders, programme and service information in French for beneficiaries. Translating the entire site equally treats all content as equally important, which means the resources available for translation are spread thinly rather than concentrated where they matter most.

The stakeholder salience framework helps here. If your primary stakeholders in a particular language are beneficiaries seeking service information, the priority pages in that language are programme descriptions, eligibility criteria, contact information, and referral pathways. If the primary stakeholders are funders who read both English and French but expect to see French content as evidence of in-country presence, the priority is a professional French-language section that demonstrates commitment, not a machine-translated mirror of every English page.

Common mistakes in multilingual implementation

The first is translating everything and maintaining nothing. A comprehensive multilingual launch followed by no ongoing maintenance is worse than a selective, maintained multilingual presence. Three well-maintained programme pages in French communicate institutional seriousness. Forty machine-translated pages that are six months out of date communicate neglect.

The second is treating translation as a one-time cost. Translation is an ongoing operational commitment. Every time the English site is updated, a governance question follows: does this update need to be reflected in other languages? If the answer is usually yes, the translation budget needs to be recurring, not project-based. With neural machine translation now accounting for 85% of enterprise deployments, the per-word cost of initial translation has dropped significantly. But the governance cost of maintaining parity, reviewing quality, and managing workflows has not decreased at all. Organisations that budget for translation tools but not for translation governance will always end up with stale content in non-English languages.

The third is ignoring right-to-left languages and non-Latin scripts. If your organisation serves Arabic-speaking or other RTL-language communities, the website must support right-to-left text rendering, which requires design and development consideration beyond translation. This is an accessibility and usability requirement, not an optional enhancement.

The fourth is assuming the editorial workflow is the same for all languages. In practice, different languages may have different approval chains. A French translation of a governance document might need review by the country director. An Arabic translation of a safeguarding policy might need review by the regional programme lead. The workflow must accommodate these differences without creating bottlenecks.

Making it governable

The most practical approach is to start small and maintain well. Define the minimum viable multilingual presence: the pages that must exist in each supported language, the update frequency, and the person responsible for triggering translation updates. Document this in the website governance policy as a section on multilingual content.

Then build from there. Add languages and pages only when the governance framework can support maintaining them. A multilingual website that expands faster than its governance can sustain will always end up in the same place: an English site that works and translated versions that do not.

Question 1: Should we translate our entire website or just key pages?

Key pages only, maintained well, is almost always better than a full translation maintained poorly. Define which pages serve stakeholders in each language and focus your translation resources there. Blog posts, detailed technical guides, and English-focused content rarely need translation unless you have a specific audience need.

Question 2: Can machine translation tools like Weglot replace human translators?

For initial translation of straightforward content, machine translation is efficient and often good enough. For governance documents, legal text, safeguarding policies, and programme-specific terminology, human review is essential. The governance question is not which tool to use but who is responsible for reviewing quality and how often.

Question 3: How do we keep translations current when the English site changes frequently?

Build a trigger into your content update workflow. When an English page on the priority list is updated, the person making the update flags the translation team. Some tools, including Weglot, can flag content changes automatically. The key is having an assigned person who monitors these flags and ensures translations are updated within a defined timeframe, not left indefinitely.

If your organisation operates internationally and is unsure whether its multilingual website presence is serving stakeholders effectively, a Blueprint Audit includes a stakeholder journey analysis across all language versions and identifies where content parity is failing.