Designing a Scalable Multilingual Support Experience for Instacart's CareBot


Project Overview

Our goal was to improve the support experience for non-English speaking shoppers while reducing inefficiencies in internal support workflows. The challenge was to design a low-cost, scalable language solution that enhances the customer journey and positions the company for future international expansion.


The Problem

Non-English speaking shoppers faced significant barriers:

  • They were often routed to English-speaking agents first, increasing frustration and support handle time.
  • The process led to higher operational costs, as English agents couldn't assist but still took time to transfer shoppers to appropriate language specialists.
  • The company faced limitations in global expansion, where legal and service requirements demanded more inclusive language support.



Problem Statement

How might we design a low-cost solution that can grow into future plans of language expansion and improve
the experience for both external customers and internal agents?

 Competitive Audit & Prototype

We conducted a competitive audit to explore how others integrate language switching. Five UI concepts were proposed and evaluated for feasibility, usability, and scalability:

  • Slide title

    Write your caption here
    Button

  • Slide title

    Write your caption here
    Button

  • #1 - Language Icon Switcher

    ✅ Recognizable, compact

    ❌ Flags represent countries, not languages; limited screen space

  • #2 - Language Dropdown Selector (Text)

    ✅ Familiar UX

    ❌ Timing of language selection unclear; dropdown too large for chat context

  • #3 - Change Language in Settings

    ✅ Useful for Spanish (most common alt language)

    ❌ Not scalable; would require full app translation.

  • #4 - Language Selection in Chat Flow *Selected*

    ✅ Easy to implement using existing components

    ✅ Allows dynamic, on-the-fly switching

    ✅ Directly connects users to correct language agents

  • #5 - Pop-Up Before Chat Start

    ✅ Visually prominent, elegant UX

    ❌ Engineering limitations due to chatbot being not fully native

PHASE 1 —  UI/ Conversational solution

As part of Phase 1, we included  design option #4 as a language selection solution embedded the chat flow.

  • Minimal development effort
  • Seamless user experience
  • Leverages current chat architecture

 Example interaction—
"Choose preferred language" → Shopper selects preferred language → Direct transfer to appropriate agent


Results Post-Launch

🔻 Reduction in agent-to-agent transfers

🔺 Improved CSAT scores for non-English shoppers

💸 Lowered staffing and training costs

📊 Improved accuracy in language-based workforce forecasting


  • Slide title

    Write your caption here
    Button

  • Slide title

    Write your caption here
    Button

  • Slide title

    Write your caption here
    Button

 
PHASE 2 — AI Translation Integration

To further streamline support and prepare for global scalability, we began exploring AI-powered live translation within the chat flow. This would enable English-speaking agents to assist any customer, regardless of language.


Tools Explored:

  • Google Translate API
  • AWS Translate
  • Microsoft Azure Translator


Key Considerations:

🧪 Translation accuracy in conversational contexts

💵 API cost vs. business benefit

⚙️ Integration time with existing chat backend

🔐 Data privacy and compliance


Impact & Takeaways

This two-phase approach created a scalable foundation for multilingual support that:

  • Significantly improved shopper experience and inclusivity
  • Created cost-saving opportunities through operational efficiency
  • Aligned with future business goals for global market expansion


Key Lesson

Solving for language isn't just a UX problem — it's a business enabler.
Starting with simple UI changes can pave the way for sophisticated, AI-driven transformation.