Spark's Knowledge Graph: Connecting code, experiments, and metrics
Why we're building the knowledge graph
Most teams can ship quickly now. What actually limits velocity is learning.
When a metric moves in an alert fire, teams need to understand what changed, why it mattered, and what to do next. As products get more complex, that learning step takes longer, even though access to data has never been easier.
Where learning breaks down
Spark already connects features, experiments, and metrics. But those objects don't show how the product actually predicts those outcomes. That understanding lives in code and in the mental models engineers build over time.
The knowledge graph makes that understanding explicit so teams and the AI systems they use can learn what matters and act faster.
How it works
We built a system that automatically maps relationships between:
- Code changes: Every PR, every feature flag toggle
- Experiments: A/B tests and their configurations
- Metrics: The outcomes teams care about
What this enables
With the knowledge graph, teams can:
- Trace any metric movement back to the code change that caused it
- Understand which experiments affect which metrics
- Get AI-powered suggestions for what to test next
- Reduce debugging time from hours to minutes
Looking ahead
The knowledge graph is just the beginning. We're building toward a future where your product development platform understands your product as well as your best engineers do.