Imported from jeremylongshore/tons-of-skills-marketplace (skills/.curated/clari-cost-tuning/SKILL.md). Install upstream with npx skills add jeremylongshore/tons-of-skills-marketplace --skill clari-cost-tuning. Copyright stays with the author (MIT).

Clari Cost Tuning

Overview

Minimize Clari API overhead: reduce export frequency, cache aggressively, export only needed data types, and monitor usage.

Prerequisites

• An approved consumer inventory and freshness requirement for each report
• Usage telemetry for API calls, export jobs, warehouse loads, and cache age
• Data-owner approval for retention, aggregation, and delivery changes
• A non-production validation path for schedule or payload changes

Instructions

Export Only What You Need

# Full export (6 data types) -- more API load
full_types = ["forecast", "quota", "forecast_updated",
              "adjustment", "crm_total", "crm_closed"]

# Minimal export (2 data types) -- faster and lighter
minimal_types = ["forecast", "crm_closed"]

# Use minimal for dashboards, full for audit/compliance

Optimize Export Frequency

Cache to Avoid Redundant Exports

# Cache recent exports (see clari-performance-tuning)
cache = ExportCache(ttl_hours=8)

def smart_export(client, forecast_name, period):
    cached = cache.get(forecast_name, period)
    if cached:
        print(f"Cache hit for {period}")
        return cached

    data = client.export_and_download(forecast_name, period)
    entries = data.get("entries", [])
    cache.set(forecast_name, period, entries)
    return entries

Usage Tracking

class ClariUsageTracker:
    def __init__(self):
        self.api_calls = 0
        self.exports = 0

    def track_call(self):
        self.api_calls += 1

    def track_export(self):
        self.exports += 1

    def report(self) -> dict:
        return {
            "api_calls": self.api_calls,
            "exports": self.exports,
        }

Error Handling

Output

Publish a cost-and-usage decision with current and proposed cadence, payload scope, cache policy, observed API/warehouse load, forecast impact, owner, and rollback threshold. Keep business amounts and rep-level data out of the operational report unless the recipient is authorized.

Examples

Move a non-critical dashboard from hourly full exports to daily minimal exports in staging, verify freshness and audit coverage, and compare job count and warehouse cost for a week. Roll back the schedule if the certified data window or required field coverage no longer meets the report contract.

Resources

• Clari Pricing

Next Steps

For architecture patterns, see clari-reference-architecture.