Grocery Voice | sameerux

Flipkart

Sr. Director &

Head of Design, Bangalore

Grocery Voice Design

Role: UI Architect, Manager
Team: Product Designer, Conversational Designer, Content designer, User Researcher.
Stakeholders: VP of Products, VP of Strategy, VP of new business, Engineering, CTO, Brand Marketing
Time: 15 weeks.
Process: Business Strategy, Primary & Secondary User research, Design Interactions, Branding, A/B Testing

Problem Statement

Design a simple, inclusive, and engaging grocery shopping experience for India’s diverse user base by:

Reducing friction caused by language barriers and varying levels of digital literacy.
Making product discovery, selection, and basket completion simple and intuitive.
Using Voice as a personal shopping assistant to guide and educate users through product discovery, cart building, and checkout.
Enabling voice to provide step-by-step assistance, helping users learn how to navigate and use the app independently.
Exploring new interaction models beyond traditional tap-based navigation

Hypothesis

Voice as a Shopping Assistant: Users may perceive voice as a trusted store salesperson, guiding them through their shopping journey.
Build Trust: A helpful, conversational voice experience can create confidence, familiarity, and trust, especially for less tech-savvy users.
Beyond Search: Users can use voice not just to search, but to navigate, read, type, compare, and complete tasks.
Expert Guidance: Voice can act as a shopping expert, answering questions and recommending products based on user needs.
Grocery as a Testbed: Grocery provides a focused and relatively predictable environment to experiment with conversational voice interactions before expanding to other categories.

User Research Approach

The research team conducted multi-city studies across India to understand user behaviors, needs, and perceptions around mobile shopping and voice interactions.

Context & Behavior: Studied mobile usage patterns, shopping habits, and user contexts.
Shopping Friction: Identified pain points and barriers while shopping on Flipkart.
Voice Usage: Observed how participants currently use voice interfaces across different apps.
Role of Voice: Explored how voice fits into users’ daily lives and the value and trust they associate with it.
E-commerce Expectations: Understood how users would like to use voice for shopping and assistance.
Use-Case Validation: Evaluated potential voice-enabled Flipkart use cases with users.

N200mn User Persona

Age: 25–35 years
Location & Connectivity: Primarily Tier 2–3 cities; mix of 3G/4G users, with limited personal Wi-Fi access.
Education & Language: High-school to graduate-level education; may not be fluent in English and primarily uses local languages.
Occupation: Clerical, sales, driving, reception, domestic work, lower management, or homemaking.
Income: Approximately ₹8,000–₹20,000/month.
Device: Entry-level Android smartphone, typically under ₹9,000.
Digital Experience: Has used smartphones for 1+ year and regularly uses Google, WhatsApp, and YouTube.
E-commerce Awareness: Familiar with e-commerce but may have limited personal experience or infrequent usage.
Key Need: A simple, trustworthy, and guided shopping experience that reduces language and technology barriers.

Commerce User Environment

Mobile-First: Smartphones are an integral part of daily life, primarily used for social communication and entertainment.
Shared & Connected: Devices may be shared within families, with users often relying on dual-SIM connectivity and inconsistent networks.
Device Constraints: Predominantly entry-level Android phones with limited storage and RAM.
Value-Driven Adoption: New apps are installed when they offer clear, practical value.
Familiar Interactions: Users tend to rely on learned interaction patterns and are less likely to explore unfamiliar UI behaviors.
Growing Voice Adoption: Voice is increasingly used as an alternative input method, and voice-enabled experiences are perceived as useful and convenient.

Voice Reseach Findings

Language vs. Proficiency: Participants predominantly used English-language interfaces, despite limited English fluency and persistent spelling challenges.
Native-Language Input: Typing in regional languages was perceived as effortful, creating a gap between language preference and input behavior.
Cognitive Load: Participants frequently skimmed or skipped long text, indicating reduced comprehension and confidence when information density increased.
Voice as a Workaround: Voice input emerged as a compensatory behavior, helping participants bypass typing effort, spelling uncertainty, and language-related friction.
Implication: Voice has potential to function beyond search—as an accessible interaction and assistance layer for users with lower digital and language proficiency.

Voice can play an important role

Voice can play a critical role in making e-commerce more accessible, trustworthy, and effortless:

Build Trust: Create a conversational experience that feels like a trusted shopping companion.
Guide & Assist: Act as a perceived expert, handholding users through unfamiliar tasks.
Simplify Understanding: Explain products, terminology, and processes in a clear, conversational way.
Reduce Effort: Make reading, searching, entering, and retrieving information faster and more effortless.
Beyond Search: Position voice as an assistive layer throughout the shopping journey, not just an input mechanism.

Compition Benchmark

Benchmarked Alexa Shopping, Google Assistant, and JioSaarthi to understand existing voice interaction patterns and commerce capabilities.
Compared key capabilities across product discovery, search, recommendations, navigation, assistance, and task completion.
Studied how voice assistants build trust, conversational context, and guidance during user journeys.
Identified opportunities to differentiate Flipkart’s Voice Assistant as a shopping companion—not just a voice search tool. Amazon’s current Alexa for Shopping direction reinforces the potential for conversational recommendations, comparisons, cart building, and personalized assistance.

Amazon Alexa

Google Assistant

Voice Assistant — Brand Personality

Partnered with an external design agency to define the brand personality, character, voice, and tone for Flipkart’s Voice Assistant.
Conducted a Design Thinking workshop with cross-functional stakeholders to align on the assistant’s personality and behavioral principles.
Optimistic & Reassuring: Guide users confidently through product discovery and shopping.
Sincere & Apologetic: Acknowledge limitations honestly when users cannot proceed.
Rewarding: Celebrate progress and completion of the user’s shopping goals.
Respectful & Inclusive: Design conversations that respect users across languages, abilities, and backgrounds.
Self-aware & Accountable: Clearly distinguish between shopping assistance and non-shopping queries.
Brand Representative: Consistently reflect Flipkart’s brand values throughout the shopping experience.

Ideation Strategy

Explored multiple design options for the placement and discoverability of voice entry points across the Flipkart ecosystem. The key question was whether voice should remain persistent across every page and business vertical, or appear selectively where it could provide meaningful value. The goal was to make voice discoverable without feeling intrusive—positioning the assistant as a helpful companion that appears when relevant, rather than another UI element competing for attention.

Failed 1st Iteration Concept

Target Business Vertical for Voice Experiment

The Grocery vertical was selected as the initial test bed for the voice experience because product attributes are relatively simple and predictable. Unlike categories such as Mobiles, where products can have multiple parameters—RAM, storage, screen size, brand, processor, and more—grocery products typically have fewer attributes. For example, a wheat flour product may primarily be defined by brand and weight (e.g., 1 kg Aashirvaad Wheat Atta).

This simpler product structure made it easier to generate relevant voice results, manage conversational flows, and handle error scenarios during the initial experiment.

For the experiment, voice discovery was promoted through a contextual banner within Grocery rather than being persistent across the app. Tapping the banner opened the voice interface as a conversational panel at the bottom of the screen, creating a chat-like experience while keeping the user within the shopping journey.

Failed Hypothesis

The initial hypothesis was inspired by WhatsApp’s familiar voice interaction model. We assumed that, given the widespread adoption of WhatsApp among Indian users, people would naturally understand and adopt a similar microphone-based interaction pattern.

Based on this assumption, we designed the experience as a chat interface anchored at the bottom of the screen. The input area displayed the captured voice message, with an actionable microphone icon positioned alongside it to initiate the next voice interaction. The text field was intentionally non-editable, keeping the experience focused entirely on voice-based conversation.

However, this interaction model did not perform as expected, revealing that familiarity with WhatsApp’s voice messaging did not necessarily translate into familiarity with conversational voice shopping.

User Testing Findings

Testing revealed a clear mismatch between the interaction model we intended and the mental model users brought from WhatsApp.

Press-and-hold behavior: Users instinctively tapped and held the microphone icon to record, following the familiar WhatsApp voice-message pattern rather than the intended tap-to-speak interaction.
Unexpected text input: Users attempted to tap the text area and type, assuming it was an editable input field.
Display vs. input confusion: The text area was designed to display the user’s spoken words, but its visual treatment led users to perceive it as a conventional text-entry field.
Key learning: Familiar interaction patterns can create unintended expectations. The experience needed to establish a clearer mental model for conversational voice interaction, rather than borrowing directly from voice messaging.

Successful Redesign Concept

Based on user feedback and observed interaction patterns, we redesigned the voice entry point and conversational states to create a clearer and more intuitive experience.

Floating Voice Entry Point: Moved the microphone from the chat input area to a floating action button (FAB), clearly separating voice activation from text input.
Bottom-Sheet Conversation: Tapping the FAB opens a bottom sheet that visually communicates the different states of the voice interaction.
Clearer Mental Model: Removed the editable text field to eliminate confusion and clearly presented the spoken words as conversation output rather than user input.
Simplified Interaction: The new design established a more predictable voice interaction model, reducing reliance on users’ existing messaging-app behaviors.

Results & Impact

The experiment demonstrated that voice could move beyond search and become a viable interaction model for commerce, particularly for users who find typing and navigating difficult.

2% MoM growth in voice-based grocery searches, indicating sustained adoption beyond initial experimentation.
4.8% of users attempted voice-based grocery transactions, validating interest in using voice for an end-to-end commerce journey—not just product discovery.
Voice beyond search: Users successfully explored, discovered, and interacted with products using conversational input, demonstrating the potential for voice across the shopping journey.
Validated a new interaction model: The redesigned FAB + conversational bottom-sheet model reduced ambiguity and created a clearer mental model for voice interaction.
Strong relevance for Bharat users: Research indicated that many N200M users preferred speaking over typing, reinforcing voice as an accessibility and inclusion opportunity.
Strategic opportunity: The experiment established voice as a potential foundation for a broader conversational commerce experience across Flipkart’s ecosystem.

Market Context

The experiment was aligned with a broader shift toward voice in India. Amazon reported that Alexa interactions in India grew 67% in 2020 and another 68% in 2021; its shopping app was handling more than 8.6 lakh Alexa requests per day in 2021.

Industry research has also highlighted India as a strong market for voice commerce: one global study reported that 71% of Indian shoppers owned a smart assistant, while 22% of shoppers globally had used voice commerce to purchase a product.

The key takeaway: voice was not simply an alternative to typing—it represented an opportunity to make commerce more conversational, accessible, and inclusive, particularly for the next billion users.