Should I track API costs by endpoint and consumer?

Yes — in most cases you should track API costs by both endpoint and consumer.

Why track by endpoint?

It helps you see:

• which endpoints are expensive to serve
• which parts of your API drive infrastructure or third-party usage costs
• where to optimize performance, caching, rate limits, or payload size
• which features are most costly relative to their value

Why track by consumer?

It helps you see:

• which customers, apps, teams, or API keys generate the most cost
• whether usage aligns with pricing tiers
• how to bill fairly for usage-based plans
• how to detect abuse, anomalies, or inefficient integrations

Best practice

Track costs at multiple levels:

• endpoint
• consumer
• method/version
• tenant/org
• environment (prod, staging)
• cost category (compute, storage, third-party calls, bandwidth)

If you only choose one

If your API is customer-facing and monetized, prioritize consumer.
If your main goal is optimization/internal operations, prioritize endpoint.

Practical recommendation

Use both, and make sure each request is tagged with:

• consumer ID
• endpoint route/template, not raw URL
• request count
• estimated cost components
• latency and response size

That gives you enough data for:

• billing
• profitability analysis
• engineering optimization
• anomaly detection

If you want, I can also suggest a simple schema for tracking API cost data.