My Product Owner is a Robot: Using ChatGPT to Create User Stories
An experiment with generative AI to create User Stories for a mobile game feature from a simple design brief.
We often encounter a familiar adage: “It’s not AI that will replace you; it’s the individual who harnesses AI that will.” This statement carries a profound truth, especially in the context of generative AI and its burgeoning role in various sectors beyond mere content creation and analysis, notably lacking in production and project management domains.
My foray into the AI landscape has been mainly through the lens of risk assessment and crafting risk statements (I will be posing on that shortly). The results were intriguing - while AI is an excellent catalyst for initiating thoughtful discussions, its lack of intricate organisational context limits its ability to pinpoint critical, nuanced edge cases.
Turning our attention to User Story generation, an exciting experiment unfolds. The task was deconstructing the ‘Golden Ticket’ feature from the popular game Candy Crush. Lacking direct access to the game’s design documents, I turned to an alternative source of information - the ‘Golden Ticket’ Candy Crush Wiki on fandom.com. This platform offered a detailed, albeit not exhaustive, description of the feature, akin to what one might expect from a game designer’s initial feature brief but not as comprehensive as a full Game Design Document (GDD). Let’s explore this and discern how AI manages this particular task.
The experiment:
Using Chat GPT, I input the following prompt:
You are an expert Scrum product owner. Your job is to help write users’ stories. These user stories will follow this format: I, as a (ROLE), want to (ACTION), so that (BENEFIT). I will provide you with a game feature and other related design material. You will create user stories from the perspective of the player. You will break that feature down to the smallest user story size possible. If you do not have enough information, you can ask questions.
Based on that, here are the results:
Based on the description of the Golden Ticket feature in Candy Crush Saga, here are some user stories that capture various aspects of this feature:
As a player, I want to spot golden tickets as I pass levels, so that I feel a sense of luck and excitement.
As a player, I want to see special levels on the map that show gold tickets, so I can aim to collect them for rewards.
As a player, I want to receive a golden ticket after passing a Golden Ticket level, so that I can accumulate tickets for rewards.
As a frequent in-game purchaser, I want to collect the right number of golden tickets to unlock a Sodalicious deal, so that I get more value for my money.
As a player, I want the Golden Ticket feature to appear intermittently in the game, so that there is an element of surprise and variety in my gaming experience.
As a player, I want to see a Golden Ticket icon at the bottom of the map screen when the offer is active, so that I am aware of the opportunity to participate.
As a player, I want to see how many tickets are required and how many I have collected, so I can track my progress towards unlocking a deal.
As a player, I want to be able to pass levels when tickets are visible, so that I can effectively collect these tickets.
These user stories aim to cover different aspects of the feature, from the discovery and collection of tickets to the rewards and the user interface elements that support the feature.
I then prompted:
Can the above user stories be broken down any more?
Based on that, I got:
The user stories for the Golden Ticket feature in Candy Crush Saga are already quite granular, but they can be broken down further for more specific aspects of the feature:
As a player, I want to occasionally spot a golden ticket after completing a level, to experience a random element of reward.
As a player, I want to know which special levels contain golden tickets, to target these levels for rewards.
As a player, I want immediate notification upon earning a golden ticket, to feel a sense of achievement.
As a player, I want a clear count of golden tickets required for a Sodalicious deal, so I can set my collection goals.
As a player, I want notifications when the Golden Ticket feature is active, to be aware of limited-time opportunities.
As a player, I want an easy-to-identify Golden Ticket icon on the map screen, for quick recognition of the feature's availability.
As a player, I want a visible counter for collected tickets versus required tickets, to monitor my progress.
As a player, I want to understand the specific requirements to pass levels with visible tickets, to strategize effectively for collection.
These further breakdowns focus on more specific interactions and goals within the overall feature, providing a more detailed view of player needs and expectations.
I then prompted:
Create the user stories for the backend operations of the Golden Ticket feature. These would be for the work done behind the scenes to make this feature available to players and configure functionality. The user would be “Product Managers”.
And I got this:
For the backend operations of the Golden Ticket feature in Candy Crush Saga, here are user stories from the perspective of Product Managers:
As a Product Manager, I want to configure the random appearance of golden tickets, so that they sporadically appear to players post-level completion.
As a Product Manager, I want to designate specific levels that contain golden tickets, to strategically distribute rewards.
As a Product Manager, I want to implement a notification system for players when they earn a golden ticket, to enhance user engagement.
As a Product Manager, I want to set up a tracking system for the number of tickets collected by players, to manage the rewards system.
As a Product Manager, I want to determine the frequency and duration of the Golden Ticket feature's availability, to maintain game balance and player interest.
As a Product Manager, I want to establish criteria for levels where golden tickets can be earned, to ensure a challenging yet achievable game experience.
As a Product Manager, I want to configure backend support for Sodalicious deals that can be unlocked with golden tickets, to incentivize player participation.
As a Product Manager, I want to integrate an icon on the map screen that indicates the active status of the Golden Ticket feature, for clear communication with players.
These stories focus on the necessary backend functionalities and configurations that support the Golden Ticket feature, ensuring a seamless and engaging experience for the players.
As a bonus exercise, I prompted:
Create a Use Story Map backbone using all the stories above.
And was pleasantly surprised with:
To create a user story map for the Golden Ticket feature in Candy Crush Saga, envision the following layout:
Epic: Golden Ticket Feature
Discovery of Tickets
Spot golden tickets after level completion.
Identify special levels with golden tickets.
Receive notification upon earning a ticket.
Collecting Tickets
Accumulate tickets after passing levels.
Track number of tickets collected vs. required.
Understand level requirements for ticket collection.
Rewards and Incentives
Unlock Sodalicious deals with collected tickets.
Get notified when Golden Ticket feature is active.
Visible Golden Ticket icon on the map screen.
User Interface and Feedback
Clear display of ticket count needed for deals.
Immediate achievement feedback after earning a ticket.
Notification and visibility of feature's availability.
Epic: Backend Operations for Golden Ticket Feature
Backbone: Configuration and Setup
Configure the random appearance of golden tickets.
Designate specific levels with golden tickets.
Implement a notification system for ticket earnings.
Set up a tracking system for collected tickets.
Determine the feature's frequency and duration.
Backbone: Game Experience Management
Establish criteria for golden ticket levels.
Configure backend for Sodalicious deal redemptions.
Integrate Golden Ticket active status icon on the map screen.
Each of these categories represents a stage or aspect of the Golden Ticket feature, with the user stories detailing specific functionalities and experiences within each category. This map helps in understanding the overall flow and organization of the feature from a user and Product Manager perspective.
Conclusion:
Upon delving into the outcome of this experiment, my reaction is one of unexpected delight. My background as a producer, immersed in translating design documentation into user stories, has ingrained in me a deep understanding of the intellectual depth and the iterative nature inherent in this process. Typically, undertaking such a task is not trivial; it demands intense focus and time. A small group, or even an individual, could easily invest a good hour or two to dissect and articulate a single aspect of a feature. Yet, in this instance, leveraging AI transformed this typically time-intensive process into an exercise that spanned less than ten minutes. This efficiency is nothing short of remarkable.
Diving into the specifics, the structure, format, and content of the stories crafted by AI were surprisingly superior to what I had anticipated. A standout feature was integrating the benefit operation within the stories – a critical yet frequently neglected aspect in User Story creation. This inclusion is not just a matter of thoroughness; it reflects a deep understanding of what makes a User Story genuinely functional and impactful.
The level of decomposition was another area where the AI excelled. One of the more common challenges I’ve noticed, particularly with those new to User Story writing, is a hesitancy or inability to break down stories to a sufficiently granular level. This often leads to unclear and actionable items in the stories. However, the AI demonstrated a capacity for breaking down complex features into digestible, understandable parts, addressing this issue adeptly.
Furthermore, the stories resonated well with the INVEST criteria – a testament to their quality. This mnemonic, vital in agile development, lays out the foundational attributes of a well-constructed User Story: Independent, Negotiable, Valuable, Estimable, Small, and Testable. Each of these attributes was discernibly incorporated into the stories generated by AI. This is no small feat, as it ensures that each story can stand alone, invites collaboration and iteration, delivers tangible value, can be realistically estimated in terms of resources and time, is concise enough for practical execution, and possesses clear criteria for successful implementation and testing.
I did not compare this AI output to what was put into production for accuracy. First of all, the feature is no longer active. Secondly, I did not want to rely on the accuracy of the Wiki or the expertise of the author of the Wiki post. That is an experiment for another time.
While these AI-generated stories may not be ready for immediate integration into a project backlog in their current state, their value as a starting point for further exploration is undeniable. They provide detail and clarity that can effectively guide the validation or reassessment of designers’ and product managers’ intentions. In essence, these stories lay a solid groundwork as a ‘rough sketch’ that can be built upon, refined, and tailored to meet the specific needs and nuances of a project, demonstrating the potential of AI as a tool for enhancing efficiency and depth in the User Story development process.
If you like this, then you may like:
