OVERVIEW
AI-generated SEM ads
Integrating ChatGPT for SEM Creative Optimization
Search Engine Marketing (SEM) is a crucial component of Frequence's advertising strategy, driving significant value for our clients. Between January 2024 and April 2024, we conducted 4,773 SEM campaigns. Each of these campaigns required the Operations (Ops) team to manually group keywords and the Creative team to write ad copy, a process taking between 30 minutes to 1.5 hours per campaign.
To address the inefficiencies and high manual workload, we embarked on a project to integrate generative AI into our SEM campaign workflow. The goal was to automate keyword grouping and ad copy creation, significantly reducing the time and effort required by our Ops and Creative teams. This strategic move aims to free up valuable human resources, enhance operational efficiency, and maintain the high quality of our SEM campaigns.
To address the inefficiencies and high manual workload, we embarked on a project to integrate generative AI into our SEM campaign workflow. The goal was to automate keyword grouping and ad copy creation, significantly reducing the time and effort required by our Ops and Creative teams. This strategic move aims to free up valuable human resources, enhance operational efficiency, and maintain the high quality of our SEM campaigns.
Role
In this project, my role involves conducting user research to understand workflows and pain points, designing intuitive interfaces for the AI tool, and facilitating usability testing to refine the tool based on user feedback. The UX designer collaborates with developers to ensure accurate implementation and evaluates the tool's impact on efficiency and user satisfaction. This ensures the generative AI tool is user-friendly and effectively meets the needs of the Creative and Ops teams.
Time Line
03.2024-08.2024
Platform
Web
Goal
· Reduce Time Spent on Manual Tasks: Automate the ad copy creation process to significantly decrease the time Creative and Ops teams spend on these tasks.
· Enhance Efficiency: Achieve a 50% reduction in time spent on keyword grouping and ad copy writing.
· Prepare for Future Phases: Develop and train the AI model for future integration, allowing Account Executives to use the tool externally
· Enhance Efficiency: Achieve a 50% reduction in time spent on keyword grouping and ad copy writing.
· Prepare for Future Phases: Develop and train the AI model for future integration, allowing Account Executives to use the tool externally
Tools
Figma,Whimsical (Wireframe tool)
Industry
SaaS, Marketing, B2B
IMPACT
Impact
Integrating generative AI into Frequence's platform will significantly streamline the SEM creative process, saving time for both the Creative and Ops teams. Phase 1 focuses on reducing manual ad copy creation by 50%, preparing the AI for future external use by Account Executives. The project is expected to save approximately 71,595 minutes of work time over a four-month period. This strategic move positions Frequence as a competitive leader in the ad tech market.
BACKGROUND
The Challenge
This project marks the first implementation of AI on the Frequence platform. The main challenge was determining how to adapt the existing workflow of the Creative team to seamlessly incorporate the AI tool. The design needed to align with the Creative team's work habits, minimizing the learning curve and allowing them to quickly adapt to the AI tool without adding significant cognitive load. Additionally, since the AI model was still in its training phase, the design had to showcase the convenience of AI-generated content while also providing easy methods for manual intervention when the generated content did not meet SEM ad requirements.
"How might we create a design that helps the Creative team reduce cognitive load, improve work efficiency, and allow for a seamless integration of AI-generated content with manual operations?"
BACKGROUND
The Solution
I conducted in-depth interviews and observation sessions with the Creative team to map out their current workflows and identify key pain points. Detailed feedback on their expectations and concerns regarding the new AI tool was collected. The focus was on creating an intuitive interface that mirrored familiar processes to reduce the learning curve.
The AI tool
DISCOVERY
The Process
The UX process at Frequence is based on the Double Diamond Theory and Lean UX process. UX team aim to incorporate the key phases of Discovery, Definition, Ideation and Implementation in all of the projects.
DISCOVERY
Research
· In-depth interviews with the Creative team
To understand the current SEM ad creation workflow of the Creative team and ensure the new tool addresses their needs and pain points, I engaged in multiple discussions with them. Starting with understanding their existing processes, I conducted several rounds of wireframe reviews to communicate requirements and expectations. This led to high-fidelity demonstrations of the new product, where I received valuable suggestions and feedback, all of which were incorporated into the product design.
· Competitive analysis
After meeting with the creative team and understanding their workflows and needs, I started testing various AIGC products online. I analyzed six or seven different software tools, such as AdCreative AI, Copy AI, and Notion AI, examining the features and applicability of each one. Based on this analysis, I decided on the features that our AI product design will include.
DISCOVERY
Insight
From the interviews and competitive analysis, I learned the insights below:
A creative user spent over 30 minutes writing an SEM ad
Each SEM ad consists of several different segments. A creative user needs over 30 minutes to create an SEM ad. However, using AI, ads can be generated in just a few minutes. The creative user only needs to review and adjust the AI-generated content, reducing the creation time to just 15 minutes.
· AI-generated content and manually edited content need to coexist simultaneously
When AI can rapidly generate SEM ads, creative users contribute the nuanced touch and critical thinking needed for refinement. This collaboration harnesses the strengths of both AI and human creativity, leading to more effective and engaging content.
· Allow for partial regeneration
When creative users find that AI-generated SEM ads need partial regeneration, while some meet the requirements, we need to handle the partial generation in a way that ensures consistency and quality.
EXPLORATION
Persona
Personas enable designers to empathize with users and guiding the creation of user-centered solutions by providing a clear understanding of who the users are, what they want, and how they might use the product. In this design, the Creative persona of the Frequence was the primary user and I created this persona to help understanding their needs and behaviors.
EXPLORATION
User Flow
Based on the existing workflows of the creative team, the new SEM ad builder page will automatically generate different pages containing six major ad segments, including Responsive Search Ads, Call Ads, Structured Snippets, and more. The creative user will review all the ad pages, make partial modifications and adjustments, and then submit the ads to Google Ads.
EXPLORATION
Wireframes
Taking a highly collaborative approach, I involved engineers, creative team, and leadership at every step. This ensured that everything we produced was technically viable, satisfied the creative users' needs, and aligned with the company's direction and market trends. A significant part of my approach included sketching, prototyping, and incorporating feedback from meetings.
IDEATION
Prototype
After all these research, I utilized the new design system to visualize the design and create a prototype. I then presented this prototype to all stakeholders and executives, aiming to persuade them to adopt this new design.
DESIGN DECISIONS
Wizard Navigation or Side Navigation
I created different solutions for the wireframes and met with the creative users to discuss the pros and cons of each. For example, the "All in One" solution combines the side navigation with all ad forms on the same page, allowing the creative team to have an overview of everything. In contrast, the "Wizard Navigation" guides them step-by-step, simplifying their cognitive load. Based on the creative team's feedback, efficiency is a higher priority for their work. As a result, I proceeded with the "All in One" solution.
DESIGN DECISIONS
Inline regenerate or whole regenerate
During the process i did multiple design reviews with stakeholders. For the first round, I designed an inline AI regenerate feature, which allowed creative users to adjust the generated ad content for each line on the fly. However, their feedback indicated that they preferred to quickly review all content before deciding whether to make local adjustments. As a result, the design evolved from the initial inline regenerate to a global regenerate feature with a lock function.
MEASUREMENT
The SEM ad creation time was reduced by 15 minutes.
Between January and April 2024, Frequence conducted 4,773 SEM campaigns, each requiring the Creative team to write ad copy, taking 30 minutes to 1.5 hours per campaign. SEM AI Builder significantly reduces the time the Creative team spends on generating SEM ads, cutting down the process from over 30 minutes per ad to less than 15 minutes. The Creative team can focus more on strategic and innovative work. This not only boosts efficiency but also ensures a diverse and creative range of ad content.
MEASUREMENT
takeaways
This project took the engineering team 4 sprints and is currently in the implementation stage. We can measure the results by evaluating how many minutes it saves for each SEM ad creation. Additionally, for the regeneration process, we should assess the accuracy by tracking how many times the "Regenerate" button is clicked per ad. We also need to determine whether the "Lock" functionality should be locked by default or initially unlocked. These product decisions should be made based on further data collection.
yuwangdesign@gmail.com
