Gen AI supports low-domain clients describing their needs.
Project Overview
My Role
Before
After
"Embracing the Gen AI Opportunity: Reinventing current business model at Upwork"
In November 2022, when chatGPT was first released, the software industry faced a crucial juncture - to harness the potential of AI or risk survival. As a world-renowned work marketplace, Upwork found itself pondering how to seize the Gen AI opportunity to cater to the unmet needs of both clients and talent by integration with Open AI. The challenge was set - testing hypotheses and delivering innovative solutions within a month.
We faced a significant question: how could we seize this opportunity and address the most pressing user need? Our answer was a resounding success, as we developed a tailored solution for our low-domain clients, resulting in an impressive 20% increase in adoption within just 6 weeks. By hitting the nail on the head with our design, we demonstrated the power of targeted solutions and their impact on our clients' engagement.
As a leading designer of Upwork Labs, my contribution to the development of the 0 to 1 Gen AI product was guiding its journey from ideation to rapid experimentation to launch as a revolutionary solution in the market. I led the creation of a comprehensive design strategy that laid the foundation for the product's development.
Conducting in-depth user research, I collaborated closely with cross-functional teams to identify unmet user needs and translate them into innovative product features. By leveraging data-driven insights and user feedback, I guided the product development process, ensuring that every design decision was rooted in a deep understanding of user preferences and market demands.
I played a key role in defining the product's user interface and experience, ensuring that the Gen AI product was intuitive and user-friendly by creating prototypes and wireframes for quick validation, and optimizing the product's functionality and usability. All of this was carried out in 6 weeks of span.
Before and After Flow
The Problem
It was the primary objective of this project to provide services to clients who have limited knowledge or none when it comes to hiring a candidate but still desire to hire the most suitable candidate for their job. These clients frequently encounter difficulties in crafting a precise job description that can accurately depict their requirements. Not just the job description but users also struggled with details like budget or any legal compliances. As a matter of fact, around 70% of Upwork's clients face similar challenges.
47% of the clients who couldn’t finish the job post description said “I don’t know what to write”
46% said “I don’t know how to describe what I need”
Upwork's ability to onboard clients and prompt them to post multiple jobs without relying on external dependencies is a crucial aspect of the platform's core functionalities. Unfortunately, a technical issue related to this feature has been causing negative impacts on the business.
Clients especially struggle to get started given that 72% of the prospective clients have never hired a freelancer before.
Current conversion rate of client visitors to registration within 7 days is 2.0% and performing at -34.9% YoY.
Goals
Challenges
User Goal
The users, who have limited knowledge of the hiring process, aim to find and hire the most suitable candidate for their job. They seek a platform that can assist them in crafting precise job descriptions and addressing other related details like skills, budget and legal compliance, ultimately facilitating a smoother hiring process despite their lack of expertise.
Business Goals
Increase job postings by 200%: Upwork aims to streamline the job posting process for clients, enabling them to post multiple jobs independently without the need for external support or dependencies.
Increase NPS score by 80%:The platform seeks to provide a seamless and user-friendly experience for clients, ensuring that they can easily navigate through the platform's functionalities and avail services efficiently.
Increase customer retention by 300%: By addressing the technical issues associated with the platform's core feature, Upwork aims to retain its existing clients and prevent them from seeking alternative solutions for their hiring needs.
Product Goals
Reduce # of steps for Job Posting: Leveraging Gen-AI, we managed to reduce multiple steps in job posting flow to auto-populate job details and make the UI more intuitive and user-friendly for our clients.
Streamline Job Posting Process: The platform aims to simplify the job posting flow as it’s a crucial part of the onboarding process for new clients, ensuring that they can quickly and seamlessly access the platform's services and post multiple jobs without encountering technical issues or dependencies.
User Education and Empowerment:
The challenge lies in effectively educating and empowering users with limited knowledge of the hiring process. Providing them with the necessary resources and guidance to craft precise job descriptions and address crucial details like skills, budget, and legal compliance might require a user-friendly and intuitive interface along with clear, accessible support resources.
Integration of OpenAI:
Integrating OpenAI's capabilities to enhance the job posting functionality can pose a technical challenge, especially if there are complexities in integrating the AI seamlessly into the platform without compromising the user experience.
Simplifying Job Posting Flow:
Streamlining the job posting process for new clients can be challenging, as it involves creating an intuitive and straightforward flow that guides users through the process step-by-step, while also ensuring that they have the necessary support and tools to address any technical or process-related issues that may arise.
Ensuring Smooth User Experience:
Achieving a seamless and user-friendly experience for clients involves addressing potential usability challenges, eliminating any friction points in the job posting process, and ensuring that the platform's functionalities are easily accessible and comprehensible for users with varying levels of technical expertise.
Technical Issue Resolution:
Overcoming technical issues associated with the platform's core feature requires a thorough understanding of the underlying technical infrastructure and potential complexities, as well as effective troubleshooting and debugging strategies to ensure a smooth and uninterrupted user experience.
Customer Retention Strategies:
Retaining existing clients and preventing them from seeking alternative solutions necessitates a comprehensive understanding of client needs and preferences, as well as the implementation of targeted strategies, such as personalized support, continuous product improvements, and value-added features, to enhance the overall customer experience and satisfaction.
Time to market:
we’ve to build this experiment in the sandbox, where we can test and validate the integrated functionality of Open AI within a month.
“HMW minimize the friction and time taken by the users to post the job and support the user to complete the job post making them feel confident.
When a user posts a job, we must understand their needs. We will inform them of every step in the process and use Open AI functionality throughout.”
Research
To understand the unmet needs, I kicked of the project by consolidating all the research we have done so far on job posting functionality across the platform with the user interviews, usability testing, market insights, NPS and customer support tickets with the help of lead UXR.
Research outcome
Focus now was
Existing job post flow
Concurrently, I conducted a comprehensive analysis of the current job posting workflow to gain a deeper understanding of the specific pain points and areas for improvement. This thorough assessment aimed to identify the primary unmet needs of our users, allowing me to prioritize the most critical aspects for enhancement.
Moreover, I explored the potential opportunities for leveraging the integration of Open AI, aligning the existing data set with the platform's capabilities. This strategic approach sought to maximize the effectiveness of the Open AI integration, ensuring that it not only addresses the identified user needs but also adds significant value to the overall job posting experience.
Screen 1: Craft a Compelling Job Title
At this initial stage, users often struggle with generating an engaging and accurate title that effectively communicates the essence of the job post, making it challenging for them to attract suitable candidates.
Screen 2: Specify Key Required Skills
Given the users' limited expertise in the field, articulating the main skills necessary for the job can be a daunting task, as they may not possess in-depth knowledge about the specific skill sets required.
Screen 3: Determine Project Scope and Duration
For users with limited industry knowledge, estimating the scale and duration of a project can be a complex endeavor, as they may lack the expertise to accurately assess the time and resources required for successful project completion.
Screen 4: Specify Job Location
While users generally have an understanding of the job's physical location, this step often requires manual input, posing a potential challenge for those seeking a seamless and efficient posting experience.
Screen 5: Set a Realistic Budget
Owing to uncertainties surrounding project timelines, users may encounter difficulties in defining a suitable budget, as they may not have a clear idea of the resources and costs involved in the project.
Screen 6: Craft an Engaging Title and Comprehensive Job Description
Formulating an impactful and comprehensive job title and description can prove to be a complex task for users with limited domain knowledge. Expressing their requirements in a clear and concise manner may present a considerable challenge, as it demands a nuanced understanding of the role's responsibilities and necessary qualifications.
Design
Iteration 1
To expedite the integration of the AI functionality due to the high demand for shipping, we implemented the AI feature in the final step while maintaining the existing flow for iteration one. Consequently, we successfully launched the beta version within a swift two-week timeframe.
Iteration 2
In the second iteration, I revisited the workflow to specifically target the most significant challenge faced by low-domain users - the difficulty in articulating their requirements accurately when creating job postings.
To optimize the process, I restructured the workflow, leveraging the Gen AI functionality to enable users to initiate the job posting by first expressing their requirements in a concise sentence or two. This strategic approach empowers users to confidently articulate their needs right from the outset, facilitating a more precise and effective job posting experience.
Screen 1: Briefly explain the need
Screen 2: Review the generated job description and add the scope
Subsequently, the system utilizes Open AI integration to generate the job description, automatically selecting the job category and requisite skills. It further provides users with the flexibility to modify the pre-selected fields and infuse their personal touch into the generated job description.
Usability testing:
Following the implementation of iteration 2, I conducted a comprehensive validation process, engaging with 20 users possessing limited domain expertise. Through this rigorous validation, we ascertained that the new workflow was well-received, significantly enhancing user navigation and comprehension. However, users expressed a specific need for an 'undo' and 'redo' function, particularly within the context of editing job descriptions. This additional feature would enable users to seamlessly review and select the most suitable description, enhancing their overall experience and ensuring precise articulation of their job requirements.
This resulted in an impressive 20% increase in adoption within just 6 weeks after shipping the functionality within a month. By hitting the nail on the head with our design, we demonstrated the power of targeted solutions and their impact on our clients' engagement.
Iteration 3
In the third iteration, our designs underwent significant enhancements, resulting in a remarkably streamlined user experience, particularly during the job posting process, which was notably expedited. This version introduced a novel feature that prompts users to articulate their specific requirements, subsequently generating a tailored job description that precisely aligns with their needs.
Additionally, the system implemented an automated data scraping functionality, pre-populating essential job details such as location, budget, and scope, thereby simplifying the process for users. This intelligent automation empowers users to effortlessly edit these details to ensure a seamless match with their unique job requirements, fostering a more intuitive and efficient job posting experience.
