bamboohr-cost-tuning

What this skill does

# BambooHR Cost Tuning

## Overview

BambooHR pricing is per-employee-per-month (not per-API-call), but excessive API usage triggers rate limiting (503 errors) which causes sync failures and operational issues. This skill covers reducing API call volume, monitoring usage, and building efficient sync patterns.

## Prerequisites

- BambooHR integration in production
- Understanding of current API usage patterns
- Application logging capturing API calls

## Instructions

### Step 1: Understand BambooHR Pricing

BambooHR charges by **employee count**, not API calls:

| Plan | Pricing Model | API Access |
|------|--------------|------------|
| Essentials | Per employee/month | Full REST API |
| Advantage | Per employee/month | Full REST API + advanced reports |
| Custom/Enterprise | Negotiated | Full API + dedicated support |

**Key insight:** API call volume does not directly affect your bill, but hitting rate limits causes operational failures. Optimize for reliability, not cost.

### Step 2: Audit Current API Usage

```typescript
// Instrument your client to log all API calls
class InstrumentedBambooHRClient {
  private callLog: { endpoint: string; method: string; timestamp: number; durationMs: number }[] = [];

  async request<T>(method: string, path: string, body?: unknown): Promise<T> {
    const start = Date.now();
    const result = await this.innerClient.request<T>(method, path, body);
    this.callLog.push({
      endpoint: path.split('?')[0], // Strip query params
      method,
      timestamp: start,
      durationMs: Date.now() - start,
    });
    return result;
  }

  generateReport(): void {
    // Group by endpoint
    const byEndpoint = new Map<string, number>();
    for (const call of this.callLog) {
      const key = `${call.method} ${call.endpoint}`;
      byEndpoint.set(key, (byEndpoint.get(key) || 0) + 1);
    }

    console.log('\n=== BambooHR API Usage Report ===');
    console.log(`Total calls: ${this.callLog.length}`);
    console.log(`Time window: ${((Date.now() - this.callLog[0]?.timestamp || 0) / 1000 / 60).toFixed(1)} minutes`);
    console.log('\nBy endpoint:');
    for (const [endpoint, count] of [...byEndpoint.entries()].sort((a, b) => b[1] - a[1])) {
      const pct = ((count / this.callLog.length) * 100).toFixed(1);
      console.log(`  ${count.toString().padStart(5)} (${pct}%)  ${endpoint}`);
    }
  }
}
```

### Step 3: Eliminate Wasteful Patterns

**Pattern 1: Replace polling with webhooks**

```typescript
// BAD: Polling every 5 minutes (288 calls/day minimum)
setInterval(async () => {
  const dir = await client.getDirectory();
  checkForChanges(dir);
}, 5 * 60 * 1000);

// GOOD: Use webhooks for real-time changes (0 polling calls)
// See bamboohr-webhooks-events skill
// Only poll as a fallback safety net (once per hour)
setInterval(async () => {
  const changed = await client.request('GET',
    `/employees/changed/?since=${lastSync}`);
  // Only process if webhook missed something
}, 60 * 60 * 1000);
```

**Pattern 2: Request only needed fields**

```typescript
// BAD: Requesting all fields when you only need 3
const emp = await client.getEmployee(id, [
  'firstName', 'lastName', 'displayName', 'jobTitle', 'department',
  'division', 'location', 'workEmail', 'homeEmail', 'mobilePhone',
  'hireDate', 'payRate', 'payType', 'ssn', 'dateOfBirth', // ...etc
]);

// GOOD: Only request what you use
const emp = await client.getEmployee(id, ['firstName', 'lastName', 'workEmail']);
```

**Pattern 3: Cache the directory**

```typescript
// BAD: Fetching directory on every page load
app.get('/employees', async (req, res) => {
  const dir = await client.getDirectory(); // Called 1000x/day
  res.json(dir.employees);
});

// GOOD: Cache with webhook-based invalidation
let cachedDirectory: any = null;
let cacheTimestamp = 0;

async function getDirectory() {
  if (cachedDirectory && Date.now() - cacheTimestamp < 5 * 60 * 1000) {
    return cachedDirectory;
  }
  cachedDirectory = await client.getDirectory();
  cacheTimestamp = Date.now();
  return cachedDirectory;
}

// Invalidate on webhook
function onWebhookReceived() {
  cachedDirectory = null;
}
```

**Pattern 4: Use custom reports for bulk data**

```typescript
// BAD: 500 individual employee GETs
for (const id of employeeIds) {
  await client.getEmployee(id, ['firstName', 'department']);
}

// GOOD: 1 custom report
const all = await client.customReport(['firstName', 'lastName', 'department']);
```

### Step 4: Implement Request Budget

```typescript
class RequestBudget {
  private count = 0;
  private windowStart = Date.now();
  private readonly maxPerHour: number;

  constructor(maxPerHour = 500) {
    this.maxPerHour = maxPerHour;
  }

  async acquire(): Promise<void> {
    // Reset counter every hour
    if (Date.now() - this.windowStart > 3600_000) {
      this.count = 0;
      this.windowStart = Date.now();
    }

    if (this.count >= this.maxPerHour) {
      const waitMs = 3600_000 - (Date.now() - this.windowStart);
      console.warn(`Request budget exhausted. Waiting ${(waitMs / 1000).toFixed(0)}s`);
      await new Promise(r => setTimeout(r, waitMs));
      this.count = 0;
      this.windowStart = Date.now();
    }

    this.count++;
  }

  stats() {
    return {
      used: this.count,
      budget: this.maxPerHour,
      remaining: this.maxPerHour - this.count,
      windowResetIn: Math.max(0, 3600_000 - (Date.now() - this.windowStart)),
    };
  }
}

const budget = new RequestBudget(500);

// Wrap all BambooHR calls
async function budgetedRequest<T>(operation: () => Promise<T>): Promise<T> {
  await budget.acquire();
  return operation();
}
```

### Step 5: Usage Dashboard Query

```sql
-- If logging API calls to a database
SELECT
  DATE_TRUNC('hour', timestamp) AS hour,
  endpoint,
  COUNT(*) AS calls,
  AVG(duration_ms) AS avg_latency,
  COUNT(*) FILTER (WHERE status >= 400) AS errors,
  COUNT(*) FILTER (WHERE status = 503) AS rate_limits
FROM bamboohr_api_log
WHERE timestamp >= NOW() - INTERVAL '7 days'
GROUP BY 1, 2
ORDER BY 1 DESC, calls DESC;
```

## Output

- API usage audit identifying wasteful patterns
- Polling replaced with webhooks where possible
- Request budget preventing rate limit hits
- Field-level optimization (request only needed data)
- Caching with webhook-based invalidation

## Optimization Impact Summary

| Optimization | Calls Before | Calls After | Reduction |
|-------------|-------------|-------------|-----------|
| Webhooks vs polling | 288/day | 24/day (safety net) | 92% |
| Custom reports vs N+1 | 501/sync | 1/sync | 99.8% |
| Directory caching | 1000/day | 12/day | 98.8% |
| Incremental sync | Full pull | Delta only | 90-99% |

## Error Handling

| Issue | Cause | Solution |
|-------|-------|----------|
| Budget exhausted | High-traffic feature | Increase budget or add caching |
| Stale cached data | Cache TTL too long | Reduce TTL or invalidate on webhook |
| Webhook delivery gaps | BambooHR delivery failure | Keep hourly polling as fallback |
| Rate limit during sync | Too many parallel requests | Use queue with concurrency limit |

## Resources

- [BambooHR Pricing](https://www.bamboohr.com/pricing)
- [BambooHR API Technical Overview](https://documentation.bamboohr.com/docs/api-details)

## Next Steps

For architecture patterns, see `bamboohr-reference-architecture`.