Salesforce Cost Tuning

Overview

Optimize Salesforce costs by reducing API call consumption, choosing the right edition, and monitoring API usage budgets. Salesforce charges per-user licenses (not per-API-call), but API limits are tied to edition + license count.

Prerequisites

• Access to Salesforce Setup > Company Information
• Understanding of current API usage patterns
• Access to contract/license details

Instructions

Step 1: Understand Salesforce Pricing Model

Key insight: API calls are per-org, not per-user. A 50-user Enterprise org gets 100,000 + (50 * 1,000) = 150,000 daily API calls. All integrations share this pool.

Step 2: Monitor Current Usage

const conn = await getConnection();
const limits = await conn.request('/services/data/v59.0/limits/');

const apiUsage = {
  daily: {
    used: limits.DailyApiRequests.Max - limits.DailyApiRequests.Remaining,
    remaining: limits.DailyApiRequests.Remaining,
    max: limits.DailyApiRequests.Max,
    percentUsed: ((limits.DailyApiRequests.Max - limits.DailyApiRequests.Remaining) / limits.DailyApiRequests.Max * 100).toFixed(1),
  },
  bulk: {
    ingestJobs: limits.DailyBulkV2QueryJobs,
    queryJobs: limits.DailyBulkV2QueryJobs,
  },
  storage: {
    dataMB: `${limits.DataStorageMB.Max - limits.DataStorageMB.Remaining}/${limits.DataStorageMB.Max} MB`,
    fileMB: `${limits.FileStorageMB.Max - limits.FileStorageMB.Remaining}/${limits.FileStorageMB.Max} MB`,
  },
};

console.log('API Usage:', JSON.stringify(apiUsage, null, 2));

Step 3: Reduce API Call Count (Biggest Cost Lever)

// BEFORE: 1 API call per record = expensive
for (const contact of contacts) {
  await conn.sobject('Contact').create(contact); // 1000 calls for 1000 records
}

// AFTER: Batch with sObject Collections = 5 calls for 1000 records
for (let i = 0; i < contacts.length; i += 200) {
  const batch = contacts.slice(i, i + 200);
  await conn.sobject('Contact').create(batch); // Max 200 per call
}

// AFTER: Use Bulk API for 10K+ records = 1 job regardless of count
await conn.bulk2.loadAndWaitForResults({
  object: 'Contact',
  operation: 'insert',
  input: csvData, // Can be millions of rows
});
// Bulk API has its own separate daily limit (15,000 jobs)

// Cache describe calls — saves 50+ calls/day if you describe objects frequently
const describeCache = new Map();
async function cachedDescribe(objectName: string) {
  if (!describeCache.has(objectName)) {
    describeCache.set(objectName, await conn.sobject(objectName).describe());
  }
  return describeCache.get(objectName);
}

Step 4: API Call Budget Tracking

class ApiCallBudget {
  private dailyBudget: number;
  private callsToday = 0;

  constructor(dailyBudget: number) {
    this.dailyBudget = dailyBudget;
  }

  async refreshFromOrg(conn: jsforce.Connection): Promise<void> {
    const limits = await conn.request('/services/data/v59.0/limits/');
    this.callsToday = limits.DailyApiRequests.Max - limits.DailyApiRequests.Remaining;
    // Note: this call itself costs 1 API call — don't check too frequently
  }

  canSpend(estimatedCalls: number): { allowed: boolean; reason?: string } {
    const projected = this.callsToday + estimatedCalls;

    if (projected > this.dailyBudget * 0.95) {
      return { allowed: false, reason: `Would exceed 95% of ${this.dailyBudget} daily budget` };
    }

    if (projected > this.dailyBudget * 0.80) {
      console.warn(`API budget warning: ${this.callsToday}/${this.dailyBudget} used`);
    }

    return { allowed: true };
  }
}

Step 5: Edition Right-Sizing

Decision tree for Salesforce edition:

If API calls/day < 15,000:
  → Developer Edition (free) or Professional ($80/user/month)

If API calls/day 15,000-150,000:
  → Enterprise Edition ($165/user/month)

If API calls/day > 150,000:
  → Unlimited ($330/user/month) or API Add-on Pack
  → OR reduce calls with batching/caching (usually cheaper)

If you need just data sync:
  → Consider Heroku Connect ($$$) for automatic bi-directional sync
  → Eliminates most API calls — data syncs via Change Data Capture

Output

• Current API usage analyzed
• Cost reduction strategies applied (batching, caching, Bulk API)
• API call budget tracking implemented
• Edition recommendation based on usage

Error Handling

Examples

Reduce API consumption with a measured batch migration

Measure current API calls for one integration in a sandbox-equivalent window, replace its record-at-a-time loop with a bounded Bulk API or collection operation, and compare call count, error rate, and processing time. Configure an alert below the organization limit and preserve a feature flag or prior integration schedule for rollback. Do not reduce polling frequency until downstream freshness requirements have been agreed with the system owner.

Resources

• Salesforce Editions & Pricing
• API Request Limits by Edition
• Limits REST Resource

Next Steps

For architecture patterns, see salesforce-reference-architecture.