Why has Rokt's audience segmentation tool experienced a 20% increase in processing time for large datasets in the last two weeks?

Introduction

Rokt's audience segmentation tool has experienced a 20% increase in processing time for large datasets over the past two weeks. This performance degradation is concerning and requires immediate attention. I'll approach this issue systematically, focusing on identifying the root cause, validating hypotheses, and developing both short-term and long-term solutions.

Step 1

Clarifying Questions (3 minutes)

• Looking at the timing, I'm thinking there might have been a recent update. Has there been any software deployment or configuration change in the last 2-3 weeks?

Why it matters: Recent changes often correlate with performance issues. Expected answer: Yes, there was a minor update. Impact on approach: If yes, we'd focus on the update's impact; if no, we'd look at gradual degradation factors.

• Considering the specificity of "large datasets," I'm curious about the data volume threshold. What's the definition of a "large dataset" in this context?

Why it matters: Helps identify if the issue is isolated to specific data sizes. Expected answer: Datasets over 1 million records. Impact on approach: Narrow focus to optimizations for larger datasets if confirmed.

• Given the precise 20% figure, I'm wondering about our measurement accuracy. How is processing time being measured, and has this method changed recently?

Why it matters: Ensures we're addressing a real issue, not a measurement anomaly. Expected answer: Consistent measurement through system logs. Impact on approach: If changed, investigate measurement methods; if consistent, focus on performance factors.

• Thinking about user impact, are all users experiencing this slowdown, or is it limited to specific segments or use cases?

Why it matters: Helps determine if it's a systemic issue or related to specific user behaviors. Expected answer: Affects enterprise clients more severely. Impact on approach: If segmented, investigate those specific use cases; if universal, look at core system components.