Turning AI Output
into Usable Insight
Users didn't struggle with getting data — they struggled with knowing what the data meant for their decisions. The real problem wasn't accuracy. It was comprehension.
The Problem
Food Spy already had a working product before I joined — users could scan food and receive detailed nutritional output. The gap wasn't the product itself. It was that there was no clearly defined high-intent use case driving retention.
Engagement dropped quickly after initial use. The system was working as designed, but users weren't returning. The issue wasn't accuracy — it was that users didn't know how to use the information to make decisions.
The system was working as designed, but users weren't returning. The problem wasn't the model — it was that the output had nowhere to land.
The Key Insight
My focus was to identify which user segment actually had repeatable need and engagement. Through 20+ user interviews and behavioral analysis, one segment stood out immediately: pregnant users showed significantly higher intent and repeat usage compared to the general user base.
This wasn't a small signal. It was a clear pattern — a group of users with a specific, urgent, and ongoing need that the product was already partially serving, without being designed for them.
Clearer use case. Stronger functional and emotional need. Higher likelihood of retention. This was the segment worth building for.
The Product Direction Decision
The key product decision was not to add more features, but to define a primary user segment and focus the product around it.
Based on research and usage patterns, I recommended prioritizing the pregnancy segment as the core use case. The tradeoff was narrowing the product focus in exchange for stronger retention and clearer value.
This shifted the product from a general-purpose nutrition tool to a more targeted experience designed for high-intent users.
Without a defined segment, even a working product has no clear direction. Defining who the product is for was the most important call.
The Key Tradeoff
After defining the target segment, the next decision was how to present information in a way that supported real user decisions. We initially considered exposing full nutritional breakdowns to maximize accuracy and completeness. In testing, this increased cognitive load and made it harder for users to make quick decisions. We chose to simplify the output, accepting reduced detail in exchange for clarity and usability.
"The goal was not data accuracy alone — but whether users could understand and act on the output."
From a system perspective, the problem was not generating nutritional data. It was determining what information to surface, when, and in what format so users could make decisions without cognitive overload.
Narrowing to a single segment meant deprioritizing broader use cases, but without that focus, retention continued to decline and the product had no clear direction to build toward.
Execution
The work focused on translating user insight into a usable product direction — not just generating more data.
Results
We saw a measurable +8% increase in feature usage within the targeted segment. More importantly, this work clarified product direction — shifting from a general-purpose tool with unclear retention to a focused experience built around a high-intent user segment.
Reflection
What this taught me
In this case, the most important decision was not what to build, but who to build for.
Product success was not driven by adding more features — it came from identifying a high-intent user segment and focusing the product around their specific needs.
In data-heavy products, accuracy alone is not sufficient. The output must be interpretable and actionable for users to derive real value. A correct answer that a user cannot act on is not a useful answer.
This reinforced my approach to product development: define the right user first, make deliberate tradeoffs, and optimize for real-world usage rather than theoretical capability.