BUILDING
MASTER ROUTES
Visibility & Planning - Pilot Program
Get Slider Revolution 200+ Templates 25+ Addons Object Library And Much More!
Due to a Non-Disclosure Agreements, the information on this project is limited.
AI-powered frictionless offers
Introduction
Upwork’s Work Management team recognized a significant opportunity to enhance its offer management system as part of the larger Project Harmony initiative. . The existing process presented challenges for both clients and freelancers due to inefficiencies and communication breakdowns. This project aimed to create a more streamlined, user-friendly, and efficient workflow, leading to higher offer acceptance rates and stronger client-freelancer relationships.
Addressing key challenges & solutions
The core problem addressed by this project was the friction and miscommunication inherent in the existing offer negotiation process. Misaligned expectations on offer terms frequently resulted in delays, offer withdrawals, and ultimately, lost contracts. This negatively impacted user satisfaction, revenue generation, and Upwork’s overall platform efficiency.
To address these issues, the project concentrated on these core areas:
- In-platform counter-offer functionality:
Enabling freelancers to directly propose changes to offers within Upwork’s platform, reducing reliance on external communication channels.
- Streamlined communication flow:
Improving the communication and negotiation process to provide greater transparency and efficiency.
-
Enhanced user experience (UX):
Designing an intuitive user interface that guides users through the offer process seamlessly, providing clarity and reducing the risk of errors.
- Improved offer acceptance rates:
The primary goal of the project was to significantly increase the percentage of offers that result in signed contracts.
- AI-Driven system enhancements:
Leveraging AI capabilities to automate aspects of the offer management process and provide intelligent suggestions to users.
By focusing on these key areas, the project aimed to solve the core problems of the existing offer management workflow, significantly improving both the user experience and overall platform efficiency.
Responsibilities
Led user research, design, and usability testing, ensuring the final product was user-centric, effective, and aligned with Upwork’s brand.
Team members involved
The core team included UX designers, backend and frontend engineers, a CTO, and a product manager.
Timeline & structure summary
The project followed Upwork’s Amazon-inspired PDLC (Define, Plan, Build, Launch, Optimize). Key stages included PRD development, design, engineering, A/B testing, and post-launch optimization.
Skills and tools
The team utilized Figma, Miro, JIRA, Confluence, Slack, and Google Docs, employing design thinking methodologies and A/B testing.
A Data-Driven approach to AI integration
Aligning teams and integrating AI
Project Harmony embarked on a monumental journey: the ambitious integration of Generative AI into Upwork’s offer creation process. This initiative faced the daunting challenge of harmonizing the diverse perspectives of multiple teams—UX, Engineering, Growth, Payments, and Business Leadership—each with unique priorities and visions. The complexity of this undertaking demanded a proactive leadership approach from the very beginning.
”Project Harmony united diverse teams to integrate Generative AI into Upwork's offer creation.
Recognizing the importance of a shared vision, I established a clear direction that united our diverse teams. By setting expectations early and providing consistent guidance, we fostered an environment that maintained focus and momentum, crucial for navigating the complexities ahead.
At the heart of our success was a commitment to cross-functional collaboration. I championed teamwork through regular meetings and workshops with key stakeholders, using insights from documents like the Client Withdrawal Reasons research and the Offer Expiry research.
These discussions revealed clients’ struggles with defining offer terms and highlighted freelancers’ frustrations with ambiguous or incomplete offers. By prioritizing these conversations, we built consensus on core user problems, ensuring that every decision aligned with a unified understanding of user needs. This collaborative approach was essential in mitigating conflicts and forging a path forward.
Every phase of our design process was fueled by data-driven insights. We relied on rigorous quantitative evidence extracted from Upwork’s internal data repositories to inform every design decision, from subtle user interface adjustments to major architectural shifts.
For example:
- We uncovered that 18% of offers were retracted due to term adjustments, as detailed in the Work Management Group Research readout.
This finding highlighted a critical aspect of our strategy: the need to improve both the quality of experience and speed to start in our core areas of focus. Understanding this issue revealed a significant pain point for clients, as an alarming 75% of them returned to the beginning of the hiring funnel to rehire the same freelancer.
18% of survey respondents withdrew offers because they needed to make edits.
75% of them went on to hire the same freelancer
Image: Research shows 18% of offers are retracted for term adjustments, impacting “Quality of Experience” and “Speed to Start Work,” underscoring the need for improved efficiency.
This cyclical behavior underscored the urgency for us to streamline the offer process and address the underlying frustrations, creating a more efficient pathway for clients and freelancers alike. By focusing on these insights, we aimed to enhance user satisfaction and minimize redundancies in the hiring experience.
To achieve this, our team identified two key areas for focus:
-
AI integration: Innovating responsibly
Integrating GenAI and Uma demanded careful planning and collaboration with Upwork’s technical teams. I established clear guidelines to ensure responsible and ethical AI usage, which not only mitigated risks but also fostered innovation and enhanced workflow efficiency in the offer management system.
-
Agile refinement: Iterating for success An iterative design process was essential for delivering a user-centric product. Regular feedback loops allowed for continuous design refinement based on data and stakeholder input. This agile approach enabled us to adapt quickly to changes while aligning with user needs and business priorities.
This project’s success highlights the importance of strong leadership in product design. By focusing on teamwork, making decisions based on data, using responsible AI, and embracing an agile approach, we created a user-friendly product that really enhanced Upwork’s offer management system. This method worked well in navigating the challenges of large-scale projects with advanced technology.
scenarios & collaboration
Enhancing communication and efficiency:
The counter-offer feature was strategically positioned to improve metrics like offer acceptance and contract start rates. This approach prioritized transparency and efficient communication between freelancers and clients.
Target audience and research alignment:
Targeting all freelancers with active offers (excluding Enterprise users) allows us to focus on fostering mutually acceptable contract terms. Our strategic positioning involved detailed A/B tests to evaluate impacts, guiding our research and planning activities. We also created visuals, including pie charts like this that highlighted contract type modifications (fixed price vs. hourly) and scenarios illustrating client and freelancer modifications, to help align our team and partners.
Image: “Concurrent activity edge cases” for understanding modification scenarios.
offer experience
User-Centric design shaping the Counter-Offer feature:
As Product Design Manager, I led a cross-functional team (designers, engineers, product managers, and researchers) in a three-week sprint to research user perceptions of a proposed counter-offer feature. This was critical to ensure the feature’s functionality and usability met the needs of both clients and freelancers and aligned with Upwork’s business objectives. Our established weekly rituals maintained momentum and facilitated clear communication throughout the fast-paced process.
Our research focused on two key areas:
-
Evaluate user understanding: counter-offers versus standard offers.
- Understand initial impressions of the counter-offer feature.
- Gauge understanding of related terminology.
-
Assess user (client and freelancer): perception of the counter-offer feature’s functionality and usability.
Testing & validation
Validating the Counter-Offer designs
Our research plan incorporated qualitative and quantitative methods, including user interviews, surveys, and A/B testing. The data informed our design decisions, ensuring a user-centric solution. This involved unmoderated user testing with UserZoom, with separate tests for freelancers and clients. Each participant completed 6-7 tasks focused on navigating contract change requests within low-fidelity prototypes built using Figma.
Image: Collaboration in action, brainstorming, prototyping, and refining for a better Upwork.
The participant groups were:
-
Freelancer test: UserTesting unmoderated (5 participants)
-
Client test: UserTesting unmoderated (5 participants)
Define to execute
The project’s success hinged on a clear, iterative process grounded in Upwork’s Amazon-inspired Product Development Life Cycle (PDLC). This framework emphasized working backward from the customer’s needs, ensuring we addressed user pain points directly. By meticulously following the PDLC, we successfully managed the project within the demanding three-week timeframe.
-
Define: Based on the research outlined above, we defined the problem of freelancer offer expiration, focusing on the friction points in client-freelancer interactions and the negotiation process. The goal was clear: create a counter-offer system to improve offer acceptance rates.
-
Plan: Milestones, including the UserZoom testing described earlier, were outlined. Wireframes were created, and buy-in was secured from stakeholders across multiple teams (Growth, Messaging, Trust & Safety, Enterprise, and Emerging Business).
-
Build: Engineering implemented the design, maintaining ongoing collaboration to ensure feature functionality aligned with business requirements.
-
Launch (Ship & Go): Readiness reviews and user acceptance testing were meticulously conducted. Post-launch, data analysis will continue to inform further refinements and optimization of the counter-offer feature.
Define: Understand the customer experience, validate market fit, assess the opportunity, align with strategy, and secure funding.
Plan: Refine the customer experience, address dependencies, assign ownership, secure resources, and outline milestones.
Build: Develop, test, and iterate while tracking metrics and engaging stakeholders.
Ship: Stage, iterate, and release while monitoring metrics and ensuring communication.
Go: Enhance user delight, monitor impact, and align with long-term goals.
Our iterative design process, fueled by design thinking workshops (brainstorming, prototype testing, and continuous feedback incorporation), and strong cross-functional alignment (harmonizing engineering and design priorities), allowed us to deliver high-quality results efficiently within the tight three-week deadline. This success stemmed from our commitment to starting with a clear problem definition and working backward from user needs.
Three weeks, countless insights, one goal:
“Transform the offer experience and unlock higher success rates for both CL’s and FL’s.”
Key pain points from user research
Building on the foundational research from the “Why Do Freelancers Let Job Offers Expire?” report, our in-depth research plan revealed critical insights into the complexities of the offer acceptance process on Upwork.
This enhanced understanding enables a more comprehensive and informed approach to improving the platform.
These are some of the key pain points included:
Client unresponsiveness :
“The client sent me an offer without actually discussing the terms…and then disappeared without answering questions.”
– FL
Lack of pre-offer alignment :
“We had some mid-understanding over the work and it expired while I was waiting for the reply back from the client.”
– FL
Unexpected offers :
“The offer came out of the blue, I asked some questions before accepting. But never got an answer.”
– FL
The recurring issues of client unresponsiveness, lack of pre-offer alignment, and unexpected offers highlight a critical need for improved communication on Upwork. Freelancers experienced frustration with offers sent without prior discussion, misunderstandings leading to expired offers while awaiting clarification, and unexpected offers creating confusion.
These findings underscore the importance of proactive communication and alignment before offers are sent. Integrating AI and automation could facilitate this, ultimately improving offer acceptance rates and creating a more positive experience for both clients and freelancers.
To effectively address the issues uncovered in our research, and in alignment with the recommendations from the “Why do freelancers let job offers expire?” report, we suggest the following actionable steps:
-
Implement a counteroffer feature
-
Provide clear prompts for clients to supply project details
-
Add in-platform notifications to alert freelancers to new offers
-
Implement a system to flag and address scam offers
Fostering ownership and system thinking
As we continued to foster ownership and think through all the variables within our customer journey, we wanted to ensure these solutions were sustainable and scalable. I worked with my team to guide and drive these solutions before the build stage of our development cycle through cross-functional discussions with business partners. These conversations helped establish ownership models and align development priorities.
The outcomes provided a clear understanding of roles and expectations, allowing us to seamlessly transition from insights to action. By doing so, we not only resolved immediate challenges but also laid the foundation for AI-powered growth and system-wide innovation in the next phase of My Stats.
Refining
for Launch
Create fabulous animations and scenes with Slider Revolution that make your visitors want to click buttons.
Refining
for launch
Collaborative innovation in action
As we move from the define & plan phases to the build phase, the focus intensifies on the collaborative efforts of our UX design and engineering teams. This collaboration is vital to ensure seamless implementation, aligning design specifications with technical capabilities. Our joint efforts will be driven by core design principles and work management agreements that uphold our commitment to our users.
Design principles & agreements
Revenue:
Our aim is to continuously improve equity among users while maximizing talent earnings within client budgets.
Big Bets:
We will delve into AI and GenAI innovations, partnering with AI/ML teams to creatively address user needs.
Contract Flywheel:
This is our primary focus within work management, ensuring we address the drivers that impact our Weekly Business Review (WBR) roadmap.
Invest in Work Management:
We are committed to enhancing the quality of user experience by alleviating pain points and investing in a robust platform, as the “heart” of Upwork.
Our process is grounded in user insights derived from research sessions, reinforcing a commitment to simplicity, functionality, and empowerment. These values guide our decisions, allowing for a design that is accessible and actionable, meeting user needs effectively.
Image: Diagram depicts the process flow between clients and freelancers regarding job offers and proposals.
Collaborative commitments in crafting experiences
During the build phase, we exemplified our commitment to interactive processes that transcended departmental boundaries. This collective engagement was essential for crafting an experience that resonated with our users.
To illustrate our vision for the new experience, we worked closely with our architecture engineers to re-engineer the flow, encompassing all user requirements collected during the design phase. This meant jumping into Miro and utilizing design tools alongside our engineering partners to create these new experiences on both a data-driven and strategic engineering level.
As part of our ongoing effort to enhance the user experience, we initiated open dialogue around edge cases identified in our user research. These discussions highlighted specific user needs that informed the design changes, which were then integrated into the larger user flow as represented in the accompanying journey map.
This journey map provides valuable context for understanding our users’ stories—where they came from, what they wanted to accomplish, and how we could facilitate their goals in a mindful manner. The edge cases covered the entire workflow, revealing opportunities to represent both users and their narratives effectively.
Navigating user needs
Addressing edge cases for enhanced experiences
In refining the offering process, we acknowledged various edge cases that would further enhance the user experience, such as:
Freelancers (FL) reverting to recent offer versions
We discussed restructuring the page for better discoverability, specifically focusing on the edit contract forms page, outdated offers page, and the active and previous contract page flows. These changes involved a comprehensive rearchitecture of the page layouts and included a new slider feature to house all these improvements.
Major kudos to the team for redesigning this experience in an adaptive approach to meet users where they are, especially on mobile platforms and devices.
Image: The image illustrates the redesigned outdated offers page and the active and previous contract page flows. It features a dropdown to highlight any changes made, providing clear visibility and enhancing user navigation
We also examined other edge cases to iterate our approach, particularly focusing on the interactions during concurrent active edits. This edge case served as a vital fix incorporated into the key requirements, enabling us to minimize confusion and enhance real-time communication between clients and freelancers.
Additionally, we reviewed the messaging and notification systems, implementing inline alerts and clarifying content structure across all notifications. This further simplified user actions, reducing any unwanted pauses in getting work done and minimizing the chances of “ghosting” from the process.
Embedded AI for change visibility between users
This feature enhanced transparency and clarity during negotiations by integrating our AI model, “UMA”, into the experience, creating a seamless and mindful embedded interface. We drew inspiration from Apple’s product tech documentation tables, implementing a simple horizontal scroll to effectively manage and compare large amounts of information.
Additionally, we conducted low-fidelity internal testing to assess the intuitiveness of this interaction, particularly on an adaptive layout. This enhancement improved the notification flow, allowing users to easily showcase changes and compare various nuances.
Image: Mindful AI assistant: Streamlining contract editing and enhancing user guidance with a seamless, embedded interface.
These steps within the build phase represented a pivotal transition where collaboration and user insights materialized into an innovative product. Together, we ensured that our work management tools not only met user needs but also exceeded expectations, creating a superior experience and elevating the platform’s craft to new heights of opportunity for growth.
This iterative process was key as we evaluated and refined our designs based on feedback from internal stakeholders. The team consistently incorporated insights from leadership, stakeholders, and user testing throughout the project, resulting in a product that effectively met the needs of both clients and freelancers.
From vision
to refinement
Create fabulous animations and scenes with Slider Revolution that make your visitors want to click buttons.
As we entered the launch phase, it was crucial to reflect on our journey and the significant strides we made in leadership and user experience. The implementation of “UMA,” our Mindful AI, enabled us to create a more streamlined offering, effectively addressing the evolving needs of both clients and freelancers. Our collaborative efforts were centered on refining processes and enhancing platform usability, setting the stage for a successful launch.
This phase prioritized continuous testing, feedback incorporation, leadership buy-in for launch readiness, and data analysis, all of which drove refinements to the UX and overall functionality. By understanding the key metrics influencing user satisfaction post-launch, we established a foundation for ongoing improvements that resonate with users’ needs, paving the way for the next evolution of our platform and streamlining it for enterprise readiness.
AI-Driven system enhancements
By leveraging our Mindful AI for intelligent assistance, we integrated real-time messaging and offer management, ensuring that clients and freelancers receive prompts at critical decision points, which reduces friction in negotiations and alleviates concerns about communication missteps.
Key improvements include:
- Real-Time assistance: Immediate prompts during negotiations
- Friction reduction: Simplified communication processes
- Enhanced relationship management: Supportive AI suggestions
Image: Our AI integration created real-time messaging and prompts to streamline negotiations, reduce friction, and support a working relationship.
Image: Automated alerts and enhanced notifications ensure immediate updates and clear communication across platforms.
Streamlined communication flow
To improve transparency and efficiency in communication and negotiation, we expanded the visibility of marketplace communications through automated alerts on the offer landing page for quick updates, included inline notifications in the requested offers changes drawer, enhanced messaging in our SMF alerts and emails, and addressed outdated pages or withdrawn scenarios for clearer communication.
Key improvements include:
- Automated alerts: Immediate updates for changes
- Enhanced notification system: Clear messaging across platforms
- Communication clarity: Addressed outdated pages and scenarios
Image: An adaptive design approach optimized the offer page with seamless navigation, tailored for global accessibility.
Redesigned requested offer changes drawer
We enhanced our in-platform counter-offer functionality by implementing a responsive fly platform for both hourly and fixed-price contract options, making it mobile-friendly and adaptive while considering international user requirements to ensure easy navigation of the updated offer page across all screen sizes.
Key improvements include:
- Mobile-Optimized adaptive design: Tailored for seamless usability across various screen sizes
- User-Friendly navigation: Streamlined for easier access and enhanced usability
- International considerations: Designed to meet the needs of global users
Testing the content
Image: The final design embodies a user-centered, intuitive solution achieved through strategic alignment, cross-functional collaboration, and iterative refinement.
• Freelancers: Offering services through hourly and fixed-price contracts.
acceptance rates:
Throughout this project, we discovered just how crucial it is to provide clarity and direction to effectively champion our ideas. I’m genuinely grateful for our amazing team’s dedication and hard work—an incredible group committed to making a real difference.
We learned the importance of thorough user research to understand our users’ needs for developing a successful product, embraced an iterative design process to create a user-friendly and effective solution, and recognized that cross-functional collaboration with diverse teams was key to successful product development. Additionally, data-driven decision making allowed us to leverage analytics to ensure we made a measurable impact throughout the process.
I’m deeply thankful for the team’s dedication to owning their destiny and working together as a cohesive unit, paying attention to every detail to get things right. By fostering collaboration, encouraging ownership, and embracing iteration, we turned a critical challenge into a remarkable success. The improvements in contract start rates and user satisfaction truly highlighted how user-centered design can drive meaningful business outcomes.
This case study reflects our successful product development effort, driven by user needs and supported by data-informed solutions and iterative practices. The integration of AI significantly boosted system effectiveness, positively impacting outcomes for both clients and freelancers.