Business Ops

WhatsApp-to-Database Real-Time Integration

Achieved 100% message capture rate with real-time WhatsApp integration, eliminating data entry chaos.

100%Message capture rate (was ~70%)

The Challenge

A business operations firm managing 200+ active client conversations on WhatsApp was losing critical information in the gap between messaging and their FileMaker database. Staff spent 3-4 hours daily copy-pasting message content, downloading media files, renaming them with client IDs, and uploading to the correct database records. Despite this effort, roughly 30% of messages — especially images, voice notes, and PDFs — never made it into the system. When disputes arose, there was no reliable audit trail. The manual process also introduced a 2-4 hour delay between receiving a message and it being searchable in the CRM, meaning staff often couldn't find recent conversations when clients called in for follow-ups.

Our Solution

We implemented a real-time webhook integration between WhatsApp Business API and FileMaker that captures every message within seconds. The core is a 7-route message router built in Make.com that handles each content type differently: plain text gets stored directly, images and videos are downloaded and Base64-encoded for FileMaker container fields, PDFs are processed and linked, voice notes are transcribed using Whisper before storage, and location pins are converted to address strings via reverse geocoding. The system handles full Hebrew RTL text natively, properly processes emoji and special characters, and maintains conversation threading by tracking WhatsApp message IDs. We also built a Cloudflare Worker as a middleware layer that handles webhook verification, request queuing during high-volume periods, and automatic retry for failed database writes.

Detailed Approach

We started with a deep dive into the FileMaker database structure, mapping every relevant table and field. The client's schema had evolved organically over years, with inconsistent field naming and some container fields that only accepted specific formats. We documented all constraints and built a transformation layer that normalizes data before insertion.

The WhatsApp Business API sends webhook events for every message type, but each payload structure is different. We built the 7-route message router in Make.com with a primary dispatcher that examines the message type field and routes accordingly. Text messages go through a sanitization step that handles Hebrew RTL markers, strips invisible Unicode characters, and preserves line breaks. Media messages (images, video, documents) trigger a download from WhatsApp's CDN (URLs expire after 24 hours), followed by Base64 encoding sized appropriately for FileMaker's container field limits.

The Cloudflare Worker sits between WhatsApp and Make.com, solving three problems: it handles the webhook verification challenge that WhatsApp requires, it queues messages during brief Make.com downtime windows (preventing data loss), and it deduplicates webhook deliveries since WhatsApp occasionally sends the same event twice. The worker uses Cloudflare KV for the dedup cache with a 1-hour TTL. We also implemented a dead-letter queue in Google Sheets for any messages that fail processing after 3 retries.

Key Results

🕐 Requirements gathering and FileMaker schema analysis took 4 days

✓100% message capture rate — up from ~70%, with zero messages lost in 6 months
✓Real-time sync: messages appear in FileMaker within 3-8 seconds of receipt
✓3.5 hours of daily manual data entry eliminated completely
✓Voice notes auto-transcribed with 96% accuracy for Hebrew content
✓Full audit trail enabled — resolved 4 client disputes in first quarter using message history
✓System handles 500+ messages/day without queuing delays during peak hours

Requirements gathering and FileMaker schema analysis took 4 days. The Make.com integration with all 7 content-type routes was built in 2 weeks. The Cloudflare Worker middleware was added in week 3 after load testing revealed the need for request queuing.