Salesforce Decode
Salesforcedecode
Back to questions
Data CloudArchitect (Hardest)performancebenchmarkscalesegmentation

Benchmark segment query performance at 100 million unified profiles

Real World Scenario

Architecture review requires proof that priority segments refresh within four hours at projected 100M profile scale before board approves Data Cloud expansion budget.

Expected Answer

• Generate representative synthetic profiles matching production attribute cardinality and null rates • Benchmark segment types separately: rule-based, lookalike, insight-dependent, activation-optimized • Measure refresh duration, compute credits, and failure rates under concurrent job load • Identify expensive predicates—unindexed fields, wide OR clauses, non-selective filters • Optimize via pre-computed insights, segment splitting, and exclusion segment patterns • Document headroom percentage before SLA breach for capacity planning • Repeat benchmarks after major releases and significant data model changes

Follow-Up Questions & Answers

Click to expand — each follow-up includes a direct, interview-ready answer

Direct answer: Generate representative synthetic profiles matching production attribute cardinality and null rates Also consider: Benchmark segment types separately: rule-based, lookalike, insight-dependent, activation-optimized In practice: Measure refresh duration, compute credits, and failure rates under concurrent job load Document the decision in an ADR and align with enterprise standards.

Architect Perspective

Scale proof is empirical—architects benchmark with realistic data shapes, not vendor benchmark slides alone.