What is the best way to translate a Salesforce Experience Cloud site?

The best approach combines Salesforce's native Translation Workbench for declarative metadata with a Salesforce-native translation platform that handles runtime content, Knowledge articles, CRM data, and messaging. Pure browser overlays break authenticated journeys and Lightning Web Runtime sites, while custom builds are expensive to govern. A native architecture with translation memory and glossaries scales across person touchpoints.

How does Agentforce support multiple languages?

Agentforce can respond in multiple languages using the underlying LLM, but production deployments need language detection, governed glossaries, translated Knowledge sources, and a fallback path to human agents who can converse in the person's language. Without translation memory and review workflows, Agentforce drifts in tone and terminology across conversations.

Can Salesforce Translation Workbench translate CRM data?

No. Translation Workbench covers metadata such as picklist values, field labels, custom labels, and validation messages, but not the row-level CRM data that people actually read. Translating case descriptions, Knowledge bodies, account notes, or Agentforce replies requires a runtime translation layer integrated with the Salesforce platform.

Can WhatsApp conversations be translated in Salesforce?

Yes, when WhatsApp is integrated through Salesforce messaging channels, inbound messages can be detected and translated to the agent's working language and outbound replies translated to the person's preferred language. This requires a translation layer with conversation context, glossary enforcement, and the ability to route to multilingual human agents on escalation.

How do nonprofits support multilingual people?

Effective nonprofits treat language as a person engagement capability rather than a website project. That means translating Experience Cloud portals, Knowledge, Agentforce conversations, mobile apps, WhatsApp and SMS messages, and giving human service teams real-time translation tools — all governed by shared translation memory and glossaries.

Can Salesforce Mobile Publisher support multiple languages?

Mobile Publisher inherits language settings from the underlying Experience Cloud site and platform translations, but app-store metadata, push notifications, and onboarding screens need separate localization. Teams also need to plan for offline content translation, dynamic Knowledge updates, and translated in-app Agentforce conversations.

What are translation memories?

A translation memory is a database of previously approved translations stored as source-target segment pairs. When new content is sent for translation, matching segments are reused automatically, which improves consistency, reduces cost, and shortens turnaround. For Salesforce, translation memory should span Knowledge, Experience Cloud, Agentforce prompts, and messaging templates.

What are translation glossaries?

A glossary is a controlled list of terms with approved translations and usage rules. For nonprofits this includes program names, legal terminology, housing concepts, and sensitive vocabulary. Glossaries enforce terminology across human translators, AI translation engines, and Agentforce so people see the same language in every channel.

How do AI agents support multiple languages?

AI agents detect the language of the inbound message, retrieve translated Knowledge, generate a response in the person's language, and apply governed glossaries so terminology stays consistent. Production deployments add quality monitoring, human review for sensitive cases, and seamless handoff to multilingual service teams.

What is the difference between native Salesforce translation and browser translation overlays?

Native Salesforce translation works inside the platform — covering metadata, Knowledge, CRM data, Agentforce, and messaging — and supports governance, memory, and reuse. Browser overlays inject client-side translations only for what the user sees in the page, and frequently break authenticated experiences, LWR components, forms, and analytics.

Salesforce Nonprofit London Community Sprint

Language Should Never Be A Barrier To Essential Services

Hyde Housing had been working to solve this multilingual challenge for nine months. Ahead of Salesforce World Tour London the week of June 15th, Salesforce community members came together in London for a Nonprofit Community Sprint to tackle it, mapping how to deliver multilingual experiences across Experience Cloud, Agentforce, and Mobile Publisher.

Read the community sprint findings

Experience Cloud
Agentforce
WhatsApp
Apple Messages
Mobile Publisher
Knowledge
Translation Studio

Built by the Salesforce community

A collaborative problem-solving sprint

Practitioners from across affordable housing, the Salesforce community, and multilingual engagement collaborated on the architecture and findings.

Amit Sahu

Affordable Housing Technology Leader

Hyde Housing

Drove Hyde Housing's nine-month effort to solve multilingual person engagement across Salesforce Experience Cloud, Agentforce, Mobile Publisher, Knowledge, and messaging channels — and brought the challenge to the London Community Sprint.

LinkedIn profile

Lucas Burke Ruiz

Senior Salesforce Business Analyst — Helping NGOs Boost Their Impact — 8X Certified — Trilingual

Salesforce Community Sprint

Salesforce community contributor focused on nonprofit innovation, architecture patterns, and language accessibility across person journeys.

LinkedIn profile

Jon Jessup

Multilingual Customer Engagement & Agentforce Specialist

1440.io

CEO of 1440.io. Specialist in Salesforce-native multilingual architectures, omnichannel person engagement, and Agentforce translation governance.

LinkedIn profile

Read full contributor profiles →

The problem

The 9-month challenge Hyde Housing brought to the sprint

Hyde Housing had spent nine months looking for a way to provide multilingual support for residents and applicants across Salesforce-powered digital services. The challenge was much larger than translating a website — so the community gathered in London to work it through together.

Person touchpoints

Every channel needs translation

Experience Cloud

Agentforce

SMS / RCS

Mobile Publisher

Knowledge

Web Chat

CRM Data

Human Agents

Notifications

Translating one of these is a website project. Translating all of them is a person engagement program.

Read the full problem narrative →

Why this matters for nonprofits

Language access is mission access

For nonprofits, language is not a feature — it is the difference between a person receiving service and being turned away.

68M+

People in the US speak a language other than English at home

U.S. Census Bureau, American Community Survey

2B+

Active WhatsApp users worldwide, primarily on mobile

WhatsApp / Meta

76%

Of consumers prefer to buy products with information in their own language

CSA Research

Affordable Housing

Refugee Services

Community Health

Food Assistance

Education

Volunteer Management

Donor Engagement

See the full nonprofit context →

The architecture options

Three approaches we evaluated

Each option has a place. The right answer depends on how much of the person journey lives inside Salesforce, and how seriously the organization treats language as a service capability.

Dimension	Native Salesforce TranslationTranslation Workbench, translated Knowledge versions, custom labels.	Browser Translation OverlaysClient-side scripts that translate the rendered page after load.	Salesforce-Native Translation PlatformsAppExchange-installed platforms (e.g. Translation Studio) governing memory, glossaries, AI, and human review.
Strengths	• Built into the platform — no new vendor • Covers metadata, custom labels, picklist values • Knowledge supports translated article versions • Predictable Salesforce licensing	• Fast to deploy for marketing pages • Low up-front cost • Covers visual page text without touching the platform	• Covers Experience Cloud, Knowledge, CRM data, Agentforce, Mobile Publisher, and messaging • Shared translation memory and glossaries across teams • AI translation engines with human review queues • Governance, audit trail, and quality metrics • Produces real translated HTML for SEO and AI answer engines
Limitations	• Does not translate CRM data or runtime content • No translation memory or glossary enforcement • No AI translation or human review workflows • Agentforce, messaging, and Mobile Publisher mostly uncovered	• Breaks Lightning Web Runtime and authenticated journeys • Does not translate Knowledge, CRM data, Agentforce, or messaging • No glossary or memory reuse • Produces no indexable HTML — hurts SEO and AI answer engines • Hides translation problems from analytics	• Higher platform investment • Requires operating model — reviewers, glossary owners, governance forum • Longer initial setup than overlays
Best fit	• Small orgs with one or two languages • Metadata-only translation needs • Internal user UI localization	• Static marketing pages only • Prototyping or short-term workarounds	• Nonprofits operating across multiple languages • Orgs deploying Agentforce in multilingual communities • Organizations where language access is a mission requirement

Deep dive into the three architectures →

Lessons learned

The hidden challenges most teams discover too late

A summary of the 13 lessons the sprint surfaced. Each one came up repeatedly across the working sessions.

See all 13 lessons →

Agentforce changes everything

Organizations are no longer translating pages. They are translating conversations.

Agentforce shifts the unit of translation from the page to the conversation turn. That changes every assumption about content, governance, and quality.

Conversation flow

Translation layer applied throughout

1
Person
2
Agentforce
3
Knowledge
4
CRM Data
5
Human Agent

↕ Runtime translation, glossary enforcement, language detection, human escalation

Explore Agentforce deep dive

Omnichannel

People communicate everywhere

A multilingual program has to plan for every channel a person might choose, not just the website.

Experience Cloud

Translate LWR and Aura sites, including dynamic components, authenticated journeys, and forms — not just static page text.

Enhanced Web Chat

Detect language per session, route to the right Agentforce experience or multilingual queue, and translate transcripts in both directions.

Localize templates per language and region, translate inbound and outbound messages, and preserve consent and context on language switches.

SMS

Manage per-locale templates with length-aware translation and carrier-safe formatting. Memory and glossaries keep marketing and service consistent.

RCS

Rich card content and suggested replies need localized labels per language. Plan fallback paths to SMS when RCS is unavailable.

Apple Messages for Business

Localize business listings, interactive messages, and Agentforce responses while respecting Apple's content guidelines per region.

Instagram

Direct message translation for service inquiries, with glossary enforcement on brand and program terminology.

Mobile Publisher

App store metadata, push notifications, onboarding, offline content, and in-app Agentforce all need separate localization tracks.

Agentforce

Salesforce-native agents that detect language, consult translated Knowledge, apply glossaries, and hand off to multilingual humans when needed.

Knowledge

Translate articles with memory and human review so every language stays in sync as the source content changes.

Translation Studio

A Salesforce-native translation platform unifies memory, glossaries, AI translation, and review queues across every channel.

See per-channel multilingual requirements →

Translation platforms

What we learned about modern translation platforms

Modern Salesforce-native translation platforms — Translation Studio is one example — combine memory, glossaries, AI translation, and human review inside the Salesforce platform.

Platform architecture

Salesforce ↔ Translation Studio ↔ AI Engines ↔ Human Reviewers

Salesforce

Experience Cloud, Agentforce, Knowledge, CRM Data, Mobile Publisher, Messaging

Translation Platform

Memory, glossaries, governance, review workflows, audit trail

AI Translation Engines

Pluggable LLM and NMT engines selected per language and content type

Human Reviewers

In-context review queues that feed approved segments back into memory

Read the platform lessons →

Hyde Housing outcome

Key findings

The team concluded that multilingual support should be treated as a person engagement capability rather than a website localization project.

Supporting Experience Cloud alone is not sufficient.

Organizations must consider Agentforce, Knowledge, CRM data, messaging, mobile apps, and human service teams as part of a unified multilingual strategy.

Read the full findings →

FAQ

Multilingual Salesforce — common questions

Designed for both search engines and AI answer engines such as ChatGPT, Perplexity, Gemini, Claude, and Copilot.

See all 28 questions →

Multilingual person engagement

Build a multilingual Salesforce experience that serves every person

Learn from Hyde Housing's journey and accelerate your own multilingual transformation across Experience Cloud, Agentforce, Mobile Publisher, Knowledge, and messaging channels.

Schedule a community discussion Download the findings

Language Should Never Be A Barrier To Essential Services

A collaborative problem-solving sprint

Amit Sahu

Lucas Burke Ruiz

Jon Jessup

The 9-month challenge Hyde Housing brought to the sprint

Language access is mission access

Three approaches we evaluated

The hidden challenges most teams discover too late

Experience Cloud is just the entry point

Agentforce translation is a conversation problem

Mobile Publisher hides localization surfaces

Knowledge translation breaks without memory

CRM data is the hardest content to translate

Translation memory is the single biggest cost lever

Organizations are no longer translating pages. They are translating conversations.

People communicate everywhere

Experience Cloud

Enhanced Web Chat

WhatsApp

SMS

RCS

Apple Messages for Business

Instagram

Mobile Publisher

Agentforce

Knowledge

Translation Studio

What we learned about modern translation platforms

Key findings

Multilingual Salesforce — common questions

What is the best way to translate a Salesforce Experience Cloud site?

How does Agentforce support multiple languages?

Can Salesforce Translation Workbench translate CRM data?

Can WhatsApp conversations be translated in Salesforce?

How do nonprofits support multilingual people?

Can Salesforce Mobile Publisher support multiple languages?

What are translation memories?

What are translation glossaries?

How do AI agents support multiple languages?

What is the difference between native Salesforce translation and browser translation overlays?

Build a multilingual Salesforce experience that serves every person