Find Your Career

Find Your Career

ZipRecruiter, 2024

ZipRecruiter, 2024

ZipRecruiter, 2024

Empowering job seekers with personalized career recommendations using LLM technology

Empowering job seekers with personalized career recommendations using LLM technology

Empowering job seekers with personalized career recommendations using LLM technology

Overview

At ZipRecruiter, we discovered that many job seekers struggled to identify the right career paths. To tackle this, we developed an LLM-powered career recommendation tool that engages users in a conversation to uncover their preferences and potential. The tool provides personalized job title suggestions, helping users navigate their career options more effectively.

As the design lead, I led the development of this AI-driven conversational chat interface, collaborating closely with the VP of Product Design, the product designer, and cross-functional teams, including user research, data science, engineering, and marketing.

The tool had a significant impact, achieving 92% user satisfaction and an 85% conversion rate to registration.

Role

UX/UI design

Data review & analysis

Wireframing

Prototyping

Timeline

Nov 2023 — Feb 2024

Challenge

60% of job searches start blank at ZipRecruiter.

60% of job searches start blank at ZipRecruiter.

⚠️
Many job seekers on our platform are uncertain about their next career opportunities. This often led to broad, unfocused results, leaving job seekers feeling overwhelmed and unsure where to begin.

⚠️
Many job seekers on our platform are uncertain about their next career opportunities. This often led to broad, unfocused results, leaving job seekers feeling overwhelmed and unsure where to begin.

Research and Insights

Job seekers highlighted the need for a tool that could help them discover the best career options tailored to their unique skills, aspirations, and goals.

Job seekers highlighted the need for a tool that could help them discover the best career options tailored to their unique skills, aspirations, and goals.

Surveys

From 120 job seekers, 75% felt overwhelmed by unfocused results, and 68% struggled to identify suitable roles.

User Interviews

80% expressed frustration with generic tools.
They craved personalized, goal-aligned recommendations.

80% expressed frustration with generic tools. They craved personalized, goal-aligned recommendations.

Behavioral Data Analysis

60% of users started their searches with a blank query, and 42% abandoned their searches within minutes.

Goals

We believed generative AI could serve as a powerful companion to guide job seekers toward clarity and confidence. Utilizing LLM, our goal was to build a robust job title recommendation tool with these objectives:

We believed generative AI could serve as a powerful companion to guide job seekers toward clarity and confidence. Utilizing LLM, our goal was to build a robust job title recommendation tool with these objectives:

Empower users

Connect users' skills and goals with relevant career opportunities.

Boost engagement

Create dynamic, interactive flows that adapt to user inputs.

Drive conversions

Build trust early by delivering immediate and tangible value.

Design Iteration

To capture detailed and valuable user input without overwhelming or constraining them, we identified two potential directions for the user interface.

To capture detailed and valuable user input without overwhelming or constraining them, we identified two potential directions for the user interface.

Structured Wizard Flow

A structured, step-by-step flow asking users predefined questions about their work experience, education, and skills.

Free-text Input

A conversational interface prompting users with the question, "Tell me about yourself," allowing them to share their stories freely.

Structured Wizard Flow

Pros

Familiar and straightforward; ensured all required data points were collected.

Cons

Felt rigid for users without conventional credentials; limited flexibility and engagement.

Why we moved on

While clear and structured, the design didn’t accommodate diverse user needs, prompting us to explore more flexible options.

Free-text Input

Pros

Allowed users the freedom to share their story; provided richer, more personalized insights.

Cons

Users might feel overwhelmed by the open-ended nature; some hesitated to share personal data with AI.

Why we recommended

We believed this approach had the potential to unlock deeper user insights. However, concerns about usability led us to validate this design through testing.

To make an informed decision, we conducted sandbox testing with the free-text input design, collecting 800+ user responses in a single day.

Sandbox testing screen

We learned that users valued the flexibility and were eager to share detailed information about themselves. Many were excited by the idea of AI helping them find career options. However, feedback revealed a desire for more guidance to reduce overwhelm and indecision. Building on the insights from the smoke test, we developed a conversational, chat-style interface that combined structured and open-ended interactions.

Final Design Approach

Introduce an AI-powered career recommendation tool that leads users on a personalized journey to discover their ideal career match.

Introduce an AI-powered career recommendation tool that leads users on a personalized journey to discover their ideal career match.

An AI-powered chat interface that feels natural and intuitive, beginning with "Tell me about yourself" to encourage users to freely share their aspirations.

Questions evolve dynamically based on user input, with each response shaping the next question to create a more personalized experience.

Beyond exploring skills and interests, the tool gathers key preferences such as salary expectations to offer more tailored recommendations.

If the initial recommendations don’t meet users' expectations, additional questions are asked to improve the recommendations.

Each career recommendation includes brief descriptions and explanations, highlighting how the suggestion aligns with the user’s background and preferences.

Results

The project helped strengthen our mission to connect job seekers with suitable opportunities while raising the bar for user-centric design in the job marketplace product.

The project helped strengthen our mission to connect job seekers with suitable opportunities while raising the bar for user-centric design in the job marketplace product.

92%

92%

of users satisfied with results

85%

85%

conversion rate to registration

1.2x

1.2x

lift in Monthly Active User

Post MVP

After launching the MVP, we continue improving the feature by integrating toggleable topics.

After launching the MVP, we continue improving the feature by integrating toggleable topics.

This design will help users narrow their focus while keeping the flexibility of open-ended input. Currently in development, this improvement reflects our commitment to delivering an even better user experience.

This design will help users narrow their focus while keeping the flexibility of open-ended input. Currently in development, this improvement reflects our commitment to delivering an even better user experience.