Zero Manual Data Entry: Full Sales Pipeline Automation for a Nutrition Consulting Firm

The Problem

Biorevolution, an Italian nutrition consulting firm with one setter and two to three active closers, processed every stage of their sales pipeline by hand. Instagram DM leads required manual CRM entries, Calendly bookings needed someone to update pipeline status and assign the closer, and after each call, a team member spent 20 to 30 minutes reviewing the transcript, updating the CRM, creating a service record, and logging payment installments. With three closers averaging several calls per day, the administrative burden grew alongside sales activity.

Results

• 100% manual data entry eliminated across the full lead-to-client pipeline
• 5 automated workflows covering lead intake, booking sync, transcript analysis, record creation, and KPI reporting
• 15-minute transcript processing: new call files in Google Drive are analyzed and CRM records updated within 15 minutes of upload
• Daily and weekly KPI reports delivered automatically at 08:00 CET every morning and 09:00 CET every Monday
• 263 CRM records, 166 service agreements, and 148 finance records managed without any manual input

Before vs After

Before: After every sales call, a closer manually reviewed the transcript, updated the lead's pipeline status in Notion, created a service record if the deal closed, and entered each payment installment individually. Lead data, booking history, and deal outcomes lived in disconnected manual processes.

After: A new transcript file in Google Drive triggers automatic AI extraction, outcome classification, and record creation across CRM, Services, and Finance databases within 15 minutes. No one on the team touches a data field unless the system flags an exception that requires human review.

Client Goal

Mattia Alemanno, founder of Biorevolution, wanted his sales team focused on conversations, not administration. With three closers running multiple calls per day, the post-call data entry was consuming hours each week that belonged to follow-up, preparation, and client delivery. The goal: zero manual data entry from the first Instagram DM through to the final payment record.

Challenges

• ManyChat webhook filtering: The platform fires webhooks on both automated sends and user replies. Processing automated sends as leads would flood the CRM with invalid entries. The system had to filter at the webhook level, passing only user-initiated replies through to Notion.
• Payment detection accuracy: AI extraction needed to distinguish between a payment being discussed and a payment being confirmed. Marking a payment as collected when the client had only expressed intent, without explicit confirmation, would corrupt the finance records.
• Partial failure handling: If Finance record creation failed after a Services record was already saved, rolling back the Services record would permanently lose that data. The system had to preserve completed writes and alert with the pending data instead.
• Timezone accuracy: All KPI calculations run for an Italian company, so daily and weekly report windows had to respect CET across both the 08:00 daily and 09:00 Monday triggers.

Solution Overview

We built five interconnected n8n workflows on a self-hosted VPS covering the entire pipeline from Instagram contact to finance record. Each workflow handles one layer of the process and passes structured data to the next stage. A global error handler sits across all five, catching failures by severity and routing alerts with full execution context so any exception can be resolved without manual log review.

How It Works

1. Instagram DM lead intake: ManyChat fires a webhook to n8n on user replies only. n8n extracts email via regex, queries the Notion CRM by email or Instagram handle, and either creates a new record with source and pipeline status populated, or updates the existing contact's last-contact timestamp. Conflicting matches, where both keys resolve to different records, route to the error handler.
2. Calendly booking sync: When a lead books a call, n8n catches the Calendly webhook, matches the invitee to their CRM record, updates pipeline status to Call Booked, assigns the closer from the round-robin rotation, and appends the booking details to the Notion page.
3. Google Drive transcript monitoring: A scheduled trigger runs every 15 minutes and polls three seller folders in Google Drive, one per closer. New transcript files are validated, metadata is extracted from the filename structure, and the content is passed to the AI extraction workflow.
4. AI transcript extraction and routing: OpenRouter processes each transcript with a structured extraction schema and returns a JSON object covering call type, outcome, lead details, service terms, and payment status. Payment is only marked collected if the payment link was explicitly sent and confirmed by the client in the transcript. The outcome routes the update: Closed Won creates CRM, Services, and Finance records; In Progress updates pipeline status; Closed Lost archives the contact; Delivery calls append notes only.
5. Partial failure protection: If Finance record creation fails after a Services record is already saved, the Services data is preserved. The error handler receives an alert with both record IDs and the pending payment information for manual creation. No rollback runs on completed writes.
6. Automated KPI reporting: At 08:00 CET daily and 09:00 CET every Monday, n8n queries CRM, Services, and Finance databases, calculates pipeline and revenue metrics in CET timezone, and delivers a formatted HTML email to the founder covering weekly pipeline activity, monthly revenue, cash collected, and overdue payments.

Key Features

• Deduplication engine: matches leads by email first, then Instagram handle, with conflict alerting when both keys resolve to different records
• Payment confirmation logic: marks collected only when the payment link was sent and explicitly confirmed in the transcript, preventing false positives
• Partial failure tolerance: preserves completed records and sends pending data with error context when downstream steps fail
• Severity-based alerting: the global error handler routes failures with workflow ID, last executed node, and execution URL
• Timezone-aware reporting: all CET calculations for both the daily and weekly report windows

Tools Used

n8n

Self-hosted workflow orchestration engine powering all 5 automations, webhook handling, scheduling, and error routing.

Hostinger VPS

Runs n8n 24/7 in the background at $6.75/month. Self-hosted VPS for full workflow control and data sovereignty.

Notion

CRM, Servizi, Finance, and Performance KPIs databases. Single source of truth for the entire client lifecycle.

OpenRouter

AI transcript extraction. Structured JSON output from Italian sales call recordings, with outcome classification and payment detection.

Google Drive

Transcript storage and monitoring. Three seller folders polled every 15 minutes to trigger AI processing on new call recordings.

Calendly

Round-robin call booking with webhook events for created, rescheduled, and cancelled appointments synced directly to the Notion CRM.

ManyChat

Instagram DM automation. Webhook fires on user reply only, filtering automated sends so only genuine lead interactions reach the CRM.

Gmail

KPI report delivery. Daily and weekly HTML emails sent automatically to the founder with pipeline, revenue, and finance metrics.