Fathom Cost Tuning

Fathom Cost Tuning

Prerequisites

• An aggregate usage/cost baseline, budget owner, recording/consent policy, quality/delivery SLO, and synthetic evaluation fixture.

Instructions

1. Measure aggregate meeting volume, processing, sync/follow-up behavior, errors, and cost by approved scope.
2. Change one reversible setting and compare against the baseline and consent/delivery safeguards.
3. Retain the change only after owner approval; restore the prior setting on regression.

Output

• A measured cost decision with owner, data/consent/delivery guardrails, and rollback threshold.

Examples

Evaluate a development workflow with synthetic meeting metadata, reduce duplicate processing or unnecessary integration calls, and compare aggregate cost/latency/error metrics. Revert on quality, consent, or CRM-sync regression; do not disable audit, retention, or access controls to save cost.

Overview

Fathom pricing scales with per-seat licensing for team features, with primary cost drivers being transcript storage volume and recording hours consumed. Every meeting generates a transcript and AI summary that persist in storage. For organizations running dozens of meetings daily, unchecked transcript accumulation and redundant API polling for meeting data create unnecessary spend. Optimizing retrieval patterns and storage lifecycle directly reduces both API costs and plan overhead.

Cost Breakdown

API Call Reduction

class FathomTranscriptCache {
  private cache = new Map<string, { transcript: string; summary: string }>();

  async getTranscript(meetingId: string, apiFn: () => Promise<any>): Promise<any> {
    // Transcripts are immutable after generation — cache permanently
    if (this.cache.has(meetingId)) return this.cache.get(meetingId);
    const result = await apiFn();
    this.cache.set(meetingId, result);
    return result;
  }

  async listMeetings(params: { include_summary: boolean }): Promise<any[]> {
    // Always use include_summary=true to avoid N+1 calls
    // Fetches summaries inline with the list response
    const response = await fetch('/api/meetings?include_summary=true');
    return response.json();
  }
}

Usage Monitoring

class FathomUsageTracker {
  private apiCalls = 0;
  private readonly rateLimit = 60; // 60 req/min
  private windowStart = Date.now();

  async throttledCall<T>(fn: () => Promise<T>): Promise<T> {
    if (Date.now() - this.windowStart > 60_000) {
      this.apiCalls = 0;
      this.windowStart = Date.now();
    }
    if (this.apiCalls >= this.rateLimit) {
      const waitMs = 60_000 - (Date.now() - this.windowStart);
      await new Promise(r => setTimeout(r, waitMs));
      this.apiCalls = 0;
      this.windowStart = Date.now();
    }
    this.apiCalls++;
    return fn();
  }

  getUsageReport(): { callsThisMinute: number; remainingCapacity: number } {
    return { callsThisMinute: this.apiCalls, remainingCapacity: this.rateLimit - this.apiCalls };
  }
}

Cost Optimization Checklist

• Use webhooks for meeting completion events instead of polling
• Always pass include_summary=true in list requests to avoid extra calls
• Cache transcripts permanently — they never change after generation
• Batch API processing within the 60 req/min rate limit
• Audit team seats quarterly and remove inactive users
• Archive transcripts older than 90 days to reduce storage costs
• Disable auto-recording for informal or standup meetings
• Skip CRM sync for internal-only meetings

Error Handling

Resources

• Fathom Pricing
• Fathom API Documentation

Next Steps

See fathom-performance-tuning.