Layla jo
The project aims to address the challenge of efficiently leveraging User-Generated Content (UGC) by developing a system to streamline the content curation process. The vast volume of UGC available poses a significant obstacle, leading to time-consuming workflows and delaying the delivery of value to clients.
For clients looking to leverage UGC, the total volume of UGC we can ingest is overwhelming. This leads to a time-consuming workflow to find content to power our displays, slowing our time to value. We aim to expand the volume of available content, potentially heightening the challenge for clients seeking workload reduction. Additionally, we're exploring automation and usability enhancements to complement this goal.
1. Increase customer engagement by addressing time concerns - Save clients time and make it easier to find content (with the goal of taking action on it) amongst the sea of content sourced.
2. Help our clients succeed by providing recommendations based on their past successes - Make targeted recommendations that speak to the KPIs our personas care most about.
- Increased the amount of displayable social content by 4% to unlock more revenue potential for clients.
- Save time curating content with content recommendations.
- Intelligent technology learns from the content preferences to surface content clients would likely publish without having to search.
As part of the project, we conducted a survey to gather feedback on clients' utilization of User-Generated Content (UGC) and its application. Out of 15 clients approached, 11 provided responses. Interestingly, 4 clients stated they do not utilize UGC through our platform.
The most common page clients leverage UGC is the Mentioned & Tagged page with all 11 clients saying they leveraged that page.
When asked to select just one data point that clients found most valuable, they chose onsite conversion, followed by social conversion. Some clients expressed interest in finding content for specific products, despite it ranking lower in the previous question.
After identifying the users' motivations and frustrations, my next step was to analyze direct and indirect competitors. I created the competitor analysis to clearly identify other direct and indirect competitors and to uncover similar existing patterns in other popular platforms.
🔍 Key Findings:
Then, I was thinking where in Bucky(Our persona)'s workflow and tasks he needs content. This can assist us in determining where and what kind of content to surface. I shared Bucky's potential workflow with a PM and asked for his thoughts. He guided me on which recommendations we should prioritize for initial release with clients. We decided to focus on 'Recommendations by conversion-driving content' and 'by engagement-driving content.'
I utilized Figjam stickies to differentiate between the MVP designs and non-MVP designs. I suggested implementing a feedback system, but a PM believed it would be beneficial to engage in discussions with ML engineers and create mid-fidelity mockups.
I was also considering the questions Bucky is trying to answer and his goals. The PM mentioned that there are some questions we need to confirm with engineers. Additionally, there was a sticky note indicating that we wanted to focus on future projects, as it posed a broader question.
Based on my competitive analysis and discussions with a PM, I created mid-fidelity wireframes to review with engineers and the ML team's PM. I provided a summary of the pros and cons for each option:
Option 1 - Carousels: Leveraging the familiarity users have with this format from platforms like Netflix and Sephora, I suggested this option. I proposed revising the filters at the bottom and adding a 'Refresh' button at the top, allowing users to generate different results if unsatisfied.
Option 2 - Category Navigation: This option involves organizing categories at the top, such as 'See All,' 'Engagement Driving,' 'You May Also Like,' etc. It offers scalability for adding more categories in the future.
Option 3 - Search Bar: Similar to Airbnb's interface, this option includes a search bar where users can select a date, category, and enter search queries.
After discussions with the ML team, we narrowed down the options to Option 1 and Option 2. Considering the ongoing experimentation with other models, we aimed to minimize changes at this stage. As the ML team was occupied with other tasks, we decided to maintain the current information architecture (IA) and format. Instead, we planned to introduce a minor change to gauge the acceptance of the change among our pilot clients.
I added an AI symbol to any of the AI-suggested tiles and introduced an additional filter. Furthermore, I set the default sorting to "most recommended" to ensure that this new feature is prominently showcased to our clients. I also worked with a techinical writer to work on the description and tooltips.
"Layla embodies a remarkable blend of positivity, collaborative spirit, and a genuine curiosity to understand the 'why' behind design challenges. Her infectious enthusiasm not only lifts team morale but also fosters an environment where creativity flourishes."
1. Collaboration with cross-functional teams
It involved collaboration with cross-functional teams, including ML engineers, PMs, and domain experts. I endeavored to effectively communicate design decisions and understand their technical constraints. Given that we hadn't met or worked together before, I made a point to check in with them regularly. Scheduling meetings with everyone was challenging, so I focused on using our time together efficiently to lead discussions.
2. Understanding ML concepts
I needed to have a basic understanding of machine learning concepts and techniques to effectively collaborate with ML engineers. This included understanding various algorithms and model evaluation metrics.. To achieve this, I enrolled in an online class and listened to podcasts to gain fundamental knowledge of machine learning.