AI Search Engine in Panel
Company
Hostinger
My Role
Product Designer
Date
December 2023
Project Brief
To enhance the usability of a hosting platform’s control panel by integrating an AI-powered search function that simplifies navigation and accelerates task completion for users.
The AI Search tool is designed to intelligently interpret user queries, provide direct access to relevant pages, and execute specific actions, thereby streamlining user workflows.
Challenge
The platform offers a wide range of products and services, which makes the control panel complex and difficult to navigate. Users often struggled to locate specific settings or actions, resulting in frustration and wasted time.
Traditional search and navigation methods were insufficient, necessitating a more intuitive solution.
Process/
Research
Hypothesis validation approach:
Support Ticket Analysis
Exploratory Session
Competitor Research
Problem in a nutshell:
Users frequently encountered difficulties when trying to navigate the platform’s extensive control panel. Common tasks, such as configuring particular domain settings, setting up/ configuring email settings, reviewing particular hosting plan limitations, were buried under layers of menus and options.
Solution:
An AI-powered search tool that will utilize natural language processing (NLP) and machine learning to interpret user intent, deliver relevant results, and enable task execution directly from the search interface. This implementation will minimize user navigation time across the platform and enhance overall satisfaction by providing a more streamlined and efficient way to manage their services.
Process/
Information architecture
Solid information architecture is crucial before starting the design as it organizes content logically, enhancing usability, search efficiency, and reducing complexity for both designers and developers.
Ensure Search Result Relevance:
Implement machine learning algorithms to guarantee accurate and relevant search results.
Enable Dynamic Indexing:
Set up automatic updates to the search indexes as the platform evolves.
Implement Natural Language Processing (NLP):
Develop the system to understand and interpret natural language queries for intuitive and user-friendly searches.
Prioritize Actions in Hierarchy:
Structure the interface to prioritize frequently used actions to improve overall user efficiency.
Improve Search with Metadata Tagging:
Associate relevant metadata with products and actions to improve search accuracy.
Customize User Permissions:
Tailor search results based on the specific access levels of different users.
Apply Access Control Filters:
Use filters to reduce clutter in search results by controlling what is shown based on access.
Provide Real-Time Suggestions:
Implement dynamic suggestions that appear as users type their search queries.
Enable Direct Action Execution:
Allow users to complete tasks directly from the search results without additional navigation steps.
Process/
Design decisions
After gathering key takeaways from the research, our team refined the search’s information architecture, which will help me, as a designer, define how users interact with it and guide the structure of the search interface and essential interaction points.
User Education for Actionable Search
Users has to be “familiarized” with the new search, to teach them, that it can also take actions.
Categorization
Actions and pages are grouped logically (e.g., “Hosting,” “Domains”) to enable quick access to relevant areas.
Contextual Suggestions
The search bar surfaces related knowledge base articles, actions, and links as users type, speeding up task completion.
Search History
Users can quickly resume previous tasks or find information they searched for earlier, reducing the need to start from scratch.
Design Trade-offs
We decided not to include filtering, subcategories, pagination, and a most-used categories list in the AI search design to maintain simplicity and focus on a streamlined user experience. These components were deemed unnecessary due to the AI’s ability to deliver highly relevant, dynamic results without the need for additional navigation or filtering complexity.
Testings/
Task-based usability testing
The focus were on tasks that directly impact the user’s business operations and the company’s revenue stream, such as upgrading hosting plans, adding payment method, enable-ing auto-renewal.
Regular Search Bar:
Average time was +-4.5 minutes per task.
AI-Search:
Average time reduced to 1.5 minutes per task.
Conclusion:
The implementation of the AI-powered search bar has significantly streamlined critical business-driven tasks, leading to faster completion times, reduced user frustration, and potentially lower support costs.