Product Photo AI Generating App: Concept Testing & Optimization

‍Role: Senior Product Manager
Duration: August 2023 – November 2023

Goal Overview

The objective of this project was to launch a Product Photo AI Generating App from scratch, designed to help users generate professional eCommerce product photos using AI technology. Users could create images either through pre-designed templates or by inputting custom guided queries. The app's core system included:

  • A landing page
  • Signup and login functionalities
  • An AI-powered product photo generator based on user input

The main goal was to assess the market viability and user adoption through product testing and targeted ads, aiming for satisfactory user engagement and conversion to a paid plan.

Challenge

Initial User Experience

Initially, the app allowed users to input only custom prompts to generate eCommerce product images. However, many users struggled to create effective prompts, leading to unsatisfactory results. This hindered the app’s primary goal of helping users achieve their job-to-be-done, as most users were unable to generate the desired product images (e.g., watches, perfumes, electronics).

Due to the improper prompt input, users failed to complete their tasks successfully, which resulted in a low engagement rate and poor retention.

Environment

The app was developed in collaboration with a cross-functional team, operating remotely. I led the product management and coordinated closely with design, development, marketing, and customer success teams. Jira was used to manage project tasks, track progress, and ensure timely feedback and iteration.

Key Metrics

  • Signups: 4% conversion rate from paid ads to signups.
  • Retention: Only 1% of users converted to the paid plan after the free trial.
  • Revenue: Generated $600 from 3 paid users after an investment of $1,500 in ad expenses.

Product Management Process

1. Research & Discovery

Initial User Feedback & Insights

After analyzing user behavior, we identified that the custom prompt system was the main challenge. Users were often unable to write effective prompts, leading to poor image outputs, which caused frustration. As a result, almost no user successfully completed their job-to-be-done with the product.

Research Conducted
  • User Feedback: Conducted in-depth interviews with users to understand where they were struggling with the custom prompt feature.
  • Ad Performance Analysis: Evaluated the conversion funnel and user acquisition costs, noting that while 4% of users signed up, very few transitioned to the paid plan.
  • Competitor Analysis: Researched how other AI-powered image generation tools structured their input systems, focusing on user guidance during prompt creation.

2. Solution: Adding Templates and Guided Prompts

Template-Based Guidance

To address the challenges with custom prompts, I introduced a template-based system. Users could select from pre-built templates for product categories (e.g., watches, perfumes, electronics), providing them with more structured options. This gave users clear, guided steps to create professional images without needing to formulate custom prompts from scratch.

Guided Prompts for Custom Input

For those who still preferred to use custom prompts, we added a guided prompt system. This feature provided real-time suggestions and prompts as the user typed, helping them structure their queries effectively. With this improvement, 80% of users were able to achieve a much higher success rate in generating the desired images.

Prototyping & Testing

I collaborated with the design team to create interactive prototypes for both the template-based and guided custom prompt systems. These prototypes were tested internally, and feedback was gathered from stakeholders and selected users to refine the designs further.

Cross-Functional Collaboration & Communication

Collaboration with Development and QA Teams

Working closely with the development team, I ensured the guided prompt and template-based systems were implemented efficiently. We tracked progress using Jira, focusing on addressing AI performance issues and integrating user feedback.

Marketing & Stakeholder Communication

I worked with the marketing team to assess the performance of paid ad campaigns, ensuring that key insights from user behavior were incorporated into the ongoing development. I also provided regular updates to stakeholders, discussing key challenges, user engagement metrics, and recommendations for improvements.

User Testing & Iteration

User Testing

The initial system (custom prompts only) saw minimal success. After the template and guided prompt systems were added, I conducted additional user testing sessions. These sessions revealed a significant improvement in user satisfaction, as most users were now able to generate images that met their expectations.

Iteration

  • Continued to refine the guided prompts based on real-time feedback, adjusting the suggestions provided to users for better outcomes.
  • Further optimized the template system, adding more categories and refining the steps for each to streamline the process.

Live Testing & Ad Campaign Review

Post-launch, we monitored user engagement via Hotjar and continued to analyze Google Analytics to track conversion rates and user drop-offs. The addition of templates and guided prompts led to higher user satisfaction, though retention remained a challenge beyond the free trial period.

Outcome

Impact Metrics

  • Signups: 4% conversion from paid ads.
  • Guided Prompt Success: After implementing the template and guided prompt systems, 80% of users successfully achieved their desired outcome, significantly improving the initial experience.
  • Paid Plan Retention: The retention rate for paid users remained at 1%, highlighting the need for further optimization in user engagement and value proposition.
  • Revenue: The app generated $600 from 3 paying users after spending $1,500 on ads.

Key Learnings & Recommendations

The initial product launch revealed that while interest in the concept was high, users needed more guidance to complete tasks successfully. The key takeaway is that without adequate support for users (through templates and guided prompts), their ability to use complex AI systems is limited.

Next Steps & Recommendations

  • Improve User Guidance: Continue refining the template and guided prompt system to increase user success rates further.
  • AI Technology Enhancement: Address the limitations of the AI model or transition to more advanced models, such as Stable Diffusion, to improve product image quality.
  • Retention Focus: Enhance the onboarding and user experience, ensuring users feel more confident using the app and are more likely to convert after the free trial.

These recommendations aim to increase user engagement, drive higher conversion rates to the paid plan, and ensure that users are satisfied with the results of the AI tool.

My Contribution

  • Led the product development from concept to market, including research, testing, and iteration.
  • Implemented a guided prompt system and template-based approach to address user struggles with custom inputs.
  • Managed all phases of the project using Jira, ensuring smooth coordination between development, design, marketing, and stakeholders.
  • Analyzed user behavior and performance metrics to drive data-informed product decisions.
  • Conducted user testing sessions to refine the product experience and increase user satisfaction.

‍