Grocery Voice | sameerux

Flipkart

2. 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

Improve the shopping experience by reducing the friction caused by limited experience in language and technology.
Make it accessible through alternate technology
Embracing a new interaction method other than tap gestures

Educate consumers on how to use the app to buy and checkout using voice as personal assistant

Hypothesis

User look at the voice as a shopping assistant like an offline store salesperson
Users relay on trusted shoppers. Voice might help to earn user’s trust & create bonds.
Users don’t want to just search with voice but also do other tasks. Voice can help with tasks of reading, typing, and searching for information easy.
Voice can be the expert to seek advice
Experimenting voice interactions in grocery section will be easy as it has fewer categories and variables.

User Research for Voice

The research team conducted multiple research across India to understand following parameters

To understand participant’s context, behaviors related to mobile usage and shopping
To understand shopping related frictions esp while using Flipkart.
To notice current usage and behaviors for existing voice interfaces in variety of apps.
To understand role played by voice interfaces in participants’ life.
To hear participants’ thoughts and feelings about voice interfaces. Values attached.
To understand in what form participants want to leverage voice interface for ecomm.
To evaluate voice interface based use cases of Flipkart.

N200mn User Persona

25-35 years old,
A mix of 3G, 4G users. Mostly not wifi personally.
Graduation or below. School drop as well. Not fluent in Eng. Uses local language.
Contract jobs (Clerical, lower management,, salesboy, driver, receptionist, maids, etc). Housewife as well but who has her own phone and not using other’s phone.
Income Rs 8000-20000 per month. Owns an Android smartphone costing less than Rs 9000.
Using smartphones for at least more than a year.
Using apps like Google Search, WhatsApp, Youtube actively.
Aware of eCommerce apps however they might have used/not used eCommerce apps personally.
Not an active user of eCommerce apps but is aware about shopping online

Commerce User Environment

Mobile Phone has become part of life
It is used more for social and may be shared with family
There is a poor network. They use dual sims
Low-end Android phones mostly. Storage & RAM is limited. Struggle to optimize phone space
New apps are installed based on value for money
Exploration of apps w.r.t new interactions is limited. Users mostly repeat the interactions that they learnt
Voice is being used as a tool to input information.
Apps that have voice feature are being preferred

Voice Findings

English is preferred over the native language for using apps. Typing is perceived to be difficult in the native language
However, using English does not mean being an expert of English. Although users use English but they are not comfortable with it. Struggle with spellings persists.
Long sentences are difficult to read and often read partially/entirely skipped. This behavior results in lack of clarity and confidence in their actions on the app.
Voice as a medium is being used to overcome inability and frustration of typing correct spelling

Voice can play an important role

To create a bond and earn the trust of potential users w.r.t online shopping?
To play the role of a perceived expert and handhold a user?
To impart clarity on what various terminologies mean?
To make tasks of reading, entering, and retrieving information feel effortless?

Competitor Study

Amazon Alexa

Google Assistant

Other assistants studied were - Jio Saarthi

Flipkart - Voice & Tone, Branding

My team worked with an external design agency to define brand characters of voice assistant for Flipkart.
Design Thinking workshop was conducted to agree on Voice Assistant's voice and tone.
Optimistic, Reassuring, and dependable - guide users to find products
Sincere and apologetic - when users cannot proceed due to our limitations
Rewarding as user complete buying goal
Respectful and Inclusive of all type of users
Self-aware and Accountable when engaging with a user about non-shopping related queries
A representative of the brand when engaging in shopping-related queries

Voice Assistant Workshop

Ideation

Various design options were evaluated on the position of voice discovery elements. The question to answer was - Should it be persistent on every page irrespective of the business or it should only appear on some business where it makes sense.

Failed Attempt

Target business for experiment

The first experiment was done in the grocery section as we know there are manageable item parameters. Unlike mobiles where there are multiple parameters like ram, storage, screen size, brand, etc.. the grocery items like the wheat floor have only brand and weight. (eg. 1 Kg Aashirvaad Wheat Atta) So voice results are better and easy to manage error responses.

For the experiment, the voice was promoted through a banner only in Grocery. Which when tapped showed the voice UI at the bottom like a chat

Failed Hypothesis

The assumption was to learn from Whatsapp's way of interacting with the mic icon. Being Asian country users, we assumed users are well familiar with how to use WhatsApp voice. So the design inspiration was a chat UI at the bottom. The text area was a voice message and an actionable voice icon to invoke speech next to it. The text area was not editable.

User Testing Findings

With this hypothesis of chat-like UI we tested the concept with a number of users and found that users were applying their WhatsApp way of interaction to this. That is they were tap and holding the mic icon to record the voice. But this was not the intended interactions. Plus they were trying to tap the non-editable text area to type. The text area was actually to display what user said.

Successful Redesign

Taking user feedback and observation findings we redesigned the voice icon position and conversation design.

Result and Impact

2% month and month increase in voice-based grocery search.
4.8% people attempted to use voice for grocery transactions.
Being not late to the party, we were able to prove that voice can also be used in commerce applications - transactions.
This is the beginning of the future. Survey results showed that lot of N200MN users preferred speaking than typing.