META
Building a Product Learning System That Turned Guesswork Intro Confident Decisions
Creator collaboration improvements
Leadership request
Monetization Concepts
Leadership request
Performance Insights Expansion
Leadership request
Attribution Enhancements
High priority
Discovery Experiments
High priority
Creator onboarding updates
High priority
Partner management features
High priority
Collaboration analytics
High priority
Sharing Tools Refresh
Under discussion
Workflow Optimization
TBD
Content Co-creation Tools
TBD
Trust & Safety Controls
Cross-posting Integrations
Notification system Ravamp
Partner Management Features
Notification system revamp
Creator crediting flows
Tipping creator explorations
Discovery onboarding experiements
Next Bets
Embedded Performance Feedback
Lightweight metrics post-publish
Strong Signal
Collaboration Templates
Faster set-up for repeat partnerships
Current Priorities
Guided Collaboration Creation Flow
Validated end-to-end experience for co-creation
Validated
Structured Credit as Core Data Layer
First-class attribution embedded into product
Iteration 2
Smart Collaborator Suggestions
Default recommendations based on behavior
Validated
Company
Meta
My role
Product Design Lead & System Architect
Overview
Built a Product Learning System That Replaced Opinion With Evidence
Real-world feedback pipeline embedded directly into product decision-making
Teams were shipping quickly but lacked real user signal to make confident decisions.
I designed and operationalized a repeatable product learning system that surfaced real-world behavior early — before engineering investment — turning intuition-driven roadmaps into evidence-based decisions.
Outcome
Turned Weeks of Guesswork Into Days of Confident Execution
Fast shipping, slow feedback, constant pivots
Teams were making roadmap decisions in isolation, driven by assumptions instead of real behavior.
Without fast validation loops, urgency consistently beat evidence — leading to late redesigns, churn, and eroding confidence.
Challenge
High Velocity Without Learning Was Creating Hidden Risk
Fast shipping, slow feedback, constant pivots
Teams were making roadmap decisions in isolation, driven by assumptions instead of real behavior.
Without fast validation loops, urgency consistently beat evidence — leading to late redesigns, churn, and eroding confidence.
Strategy
Shifted Product Development From Guesswork to Continuous Learning
Validation as a core operating system, not a research phase
Rather than treating research as a separate step, I embedded real-world feedback directly into the design and prioritization flow, making learning a default input for every major decision.
Assumption-Driven Roadmap - Q1
Creator collaboration improvements
Leadership request
Monetization Concepts
Leadership request
Performance Insights Expansion
Leadership request
Attribution Enhancements
High priority
Discovery Experiments
High priority
Creator onboarding updates
High priority
Partner management features
High priority
Collaboration analytics
High priority
Sharing Tools Refresh
Under discussion
Workflow Optimization
TBD
Content Co-creation Tools
TBD
Trust & Safety Controls
Cross-posting Integrations
Notification system Ravamp
Partner Management Features
Notification system revamp
Creator crediting flows
Tipping creator explorations
Discovery onboarding experiements
Learning pipeline validation loops
Evidence Driven Roadmap
Next Bets
Embedded Performance Feedback
Lightweight metrics post-publish
Strong Signal
Collaboration Templates
Faster set-up for repeat partnerships
Future Opportunities
Advanced collaboration controls
Custom roles and permissions
Discovery Boost for Credited Creators
Visibility powered by structured data
Strong Signal
Current Priorities
Guided Collaboration Creation Flow
Validated end-to-end experience for co-creation
Validated
Structured Credit as Core Data Layer
First-class attribution embedded into product
Iteration 2
Smart Collaborator Suggestions
Default recommendations based on behavior
Validated
Representative roadmap illustrating the shift from assumption-based planning to learning-driven prioritization
Execution
Build a Repeatable Pipeline for Real-World Signal at Scale
From prototype to insight in days
I created a standardized learning pipeline that moved teams from design mockups to representative user feedback with measurable outcomes, producing actionable themes that fed directly into prioritization and iteration.
To scale adoption, I paired the system with a simple decision framework that aligned teams around evidence instead of opinion.
Tokyo, Japan
joshua_l
1/3
September 19
joshua_l The game in Japan was amazing and I want to share some photos
Liked by craig_love and 44,686 others
Your Story
zackjohn
kieron_d
craig_love
karennne
LIVE
9:41
Feedback is specific providing the who, what, when, where and why to inform design decisions
Submit design mock-ups
Actionable themes and insights
Design session
ROUND 1
With measurements
Feedback from representative external group
ROUND 2
Feedback from representative external group
With measurements
High level framework stages
FEEDBACK FLOW
Product teams could submit designs and receive structured, decision-ready feedback within days - not weeks
Decision outputs
Make collaboration a guided creation flow
Guided Creation Flow
Establish default credit behavior in every collaboration
Simplified Templates
Theme Clusters
Synthesized patters
Friction in Creation Flow
Missed credit moments
Slow set-up
Lack of default Credit Behavior
Reliance on manual captions
Unclear where credit belongs
Raw Signals
Intersectional real-world users
“Forgot to tag collaborators before posting”
“Didn’t understand structured credit at first”
“Manual tagging felt slow for repeat partners”
“Where do I actually credit someone?”
“Where do I actually credit someone?”
Signals were synthesized into patterns that directly informed product decisions
My Role
Architected and Drove a New Product Learning Operating System
Cross-functional leadership across design, research, product, and engineering
I led the design of the pipeline, facilitated alignment across teams, operationalized validation workflows, and embedded learning into everyday decision-making without increasing headcount.
If key questions can’t be answered instantly, teams aren’t deciding, they’re guessing.
And guesswork scales poorly.
Impact
Shifted Instagram Toward a Creator Ecosystem
Credit redistributed opportunity, not just recognition
By embedding structured credit into everyday creation, the system increased visibility for historically under-credited creators, reduced attribution friction outside the product, and established a foundation for future discovery and monetization systems built on attribution.
Impact of change increases earlier in the development process
UNDERSTAND
Discover
IDENTIFY
Define & prioritize
Design & iterate
EXECUTE
Develop
Evalutate
EDUCATE
Training
GOVERN
Reviews
Learning earlier in the lifecycle compounds impact and reduces downstream risk
Decision Log
PRODUCT QUESTION
REAL-WORLD SIGNAL
RESULTING DECISION
Should creators see creator crediting upfront? Where?
Low engagement in round 1
Moved to secondary flow
Is the onboarding flow solving the core pain? What is it?
Strong qualitative and behavioral signal
Accelerated development
Does allowing for payment in chat reduce tipping friction?
Distrust remained high in US-based, indian women and urban in credit card entering
Redesigned before build
Examples of how real-world signals directly informed product decisions
Reflection
Learning as a Product Capbility
This work reshaped how I think about research, not as a phase, but as infrastructure embedded directly into decision-making.
When teams have real-world signal early, speed stops competing with confidence.
The right learning systems don’t slow product development, they eliminate wasted bets.
Product Manager @ Meta
This fundamentally changed how we made decisions in a way I haven't seen before.
Instead of debating opinions, we could point to real user signal, build with them and move forward with confidence, faster and with fewer reversals.
