Figma, Figma Mirror
Tools
UX/UI designer
My role
March 2022
Dates

Voice Assistant "Samokat"

2

Project Overview

The assistant simplifies situational orders, such as missing ingredients while cooking, allowing users to order groceries by voice without unnecessary actions.
This project was completed as part of my studies at Yandex Practicum.
1

About the "Samokat" Service

The company’s main growth occurred at the beginning of the pandemic. Samokat was the first to provide people with what they needed: fast and free delivery.
Samokat is a grocery delivery service. Its key feature is fast delivery (10−15 minutes). Delivery is free, and the app is simple and user-friendly.

About the Project

Purchase Journey Analysis

Before development, I analyzed the checkout process in "Samokat" — how many steps it takes, what challenges may arise, and how to improve it.

Solution: Voice assistant

4. Smart search: the assistant understands variations in product names and corrects inaccurate queries, simplifying the selection process.
3. Convenient for people with special needs: voice commands and responses make the interface more accessible, and voice order confirmation reduces the number of errors.
2. Assists with errors by offering alternative products and payment methods.
1. Reduces the number of steps and simplifies order management by allowing all actions — from placing an order to cancellation — in a single window.
I developed a voice assistant concept that:
4. Multiple steps: long shopping lists increase the number of actions, especially when searching through the catalog.
3. Limited inclusivity: users with disabilities may find it difficult to navigate the app.
2. No error resolution: in case of failed payment or out-of-stock items, no alternative solutions are offered.
1. Search difficulties: the search function does not recognize typos or inaccurate queries.
The purchase flow analysis helped identify areas for improvement.

Issues with the Current Experience

Problems and Solutions

Competitors already use voice assistants, which simplify the product ordering process and increase user loyalty. However, their solutions are often limited to standard scenarios, and there are still areas for improvement.

Yandex "Alice"

3. A useful reminder function for the contents of the shopping cart.
2. A universal assistant button that combines recording, request processing, and speech stopping functions.
2. In case of speech recognition issues, there is not always access to the keyboard for manual input of the query.
1. Not all user query errors are handled with solution recommendations.
1. The voice assistant reading the text aloud and the ability to stop this function.
Cons:
Pros:

Sber "Salyut"

3. The pricing details are not displayed, which violates Nielsen’s heuristic "Recognition Rather than Recall".
2. On some steps, the ability to manually enter data disappears, limiting the flexibility of use.
1. The lack of onboarding and hints makes the voice assistant feature less noticeable and understandable for the user.
Cons:
2. The animation of the voice assistant button during speech recognition, response processing, and voice output helps the user understand that the assistant is working and processing the request.
1. A useful feature of labeling products with numbers to simplify selection.
Pros:

Competitor analysis with voice assistants

I studied the solutions of competitors that have already implemented voice assistants for product orders: Sber "Salyut" and Yandex "Alice" (test version).

Interaction scenarios

The main interaction scenarios between users and the voice assistant have been defined, ranging from standard situations to critical cases. This helped account for diverse needs and usage contexts.
3. Critical scenario: An error occurs during payment, and the assistant offers solutions to resolve the issue.
2. Additional scenario: The user is unsure of what they need and asks the voice assistant to recommend products.
1. Main scenario: The user orders a product through the voice assistant and completes the purchase by paying for the order.
Based on the collected data, wireframes were created to define the preliminary design and optimize the user journey.

Wireframes

4. Order assistance. The voice assistant helps users choose products, suggests alternatives, and streamlines the ordering process.
3. Product ordering. During the ordering process, products are numbered to simplify input—users only need to say the number to select an item. Before completing the order, the voice assistant displays the cart and asks for confirmation, reducing the risk of errors.
2. Voice assistant states. The assistant goes through the following states: listening (pulsing button), processing (button with timer), reading aloud ("Stop" button), and switching to text input.
1. Onboarding. Helps users learn about the voice assistant feature and understand how to use it.

Design

Respondents found the voice assistant useful and helped identify key areas for improving usability and accessibility.
3. Keyboard button: Improve visual design (icons, size) for better intuitiveness.
2. Alternative products: Offer alternatives immediately instead of asking the user if they need them.
1. Onboarding reduction: Reduce the number of screens from three to one to simplify the process and increase engagement.
Proposed improvements after testing:
To evaluate the usability of the developed solution, testing was conducted with active users of delivery services. Participants interacted with a prototype in Figma Mirror, going through three scenarios: standard, additional, and critical.

Testing

  • Tested the interface with users using Figma Mirror to validate or refute hypotheses.
  • Integrated the new solution into the existing design system.
  • Applied mobile development principles and Human Interface Design guidelines to create the interface.
The project resulted in the development of a voice assistant for grocery ordering.
Throughout the process, I:

Conclusions

Up
vasilchenko.daria@yandex.ru
+972 50-755-9062
Haifa, Israel

Contacts