AI-Enhanced Skincare Recommendation Workflow Automation for Estheticians

ChatGPT integration + Google Sheets + Zapier + Airtable

Problem

A growing beauty brand needed a scalable, standardised way for estheticians to give personalised skincare product recommendations. At the time:

  • All recommendations were being created manually

  • Estheticians used inconsistent logic

  • Product-to-skin-concern matching varied by person

  • No central CRM existed for client history

  • No automated messaging or follow-up

  • No system connected skincare intake → product catalogue → recommendations

The business wanted a system that could:

  • Collect structured client data

  • Automatically generate personalised recommendations

  • Support estheticians with consistent logic

  • Deliver recommendations immediately via email

  • Store everything for future client sales and retention

But they had no automation, no AI integration, and no scalable backend.

⭐ Solution

I designed and implemented a multi-platform, AI-powered recommendation system using:

  • Airtable (data architecture)

  • Google Sheets (AI prompt processing layer)

  • Zapier (automation + email delivery)

  • ChatGPT API (AI-generated recommendation text)

This combination became a full operational engine for personalised skincare recommendations.

1. Departmental Workflow Discovery

I mapped workflows across four departments to ensure alignment:

Product Development

  • Product tags

  • Ingredients

  • Skin concern mapping

  • Benefits

Marketing

  • Brand voice

  • Messaging style

  • Tone of recommendations

Sales / Client Services

  • How estheticians deliver recommendations

  • Internal guidance & use cases

Tech / Automation

  • Airtable schema

  • Zapier automation flow

  • ChatGPT prompt engineering

  • Google Sheets processing logic

⭐ 2. Client Intake Form (Airtable Form)

I built an Airtable form that captured:

Personal details

  • Name, email, phone, age, gender

Skin profile

  • Skin type (single select)

  • Skin concerns (multi-select)

  • Desired outcomes (multi-select)

Product context

  • Current products

  • Frequency of use

  • Allergies

  • Additional notes

These fields were chosen intentionally to support clean data, segmentation, and AI reasoning.

⭐ 3. Product Database Architecture (Airtable)

I built relational product tables that did the heavy lifting:

Products Table

  • Product name

  • Description

  • Skin concern tags

  • Benefits

  • Usage instructions

  • Brand tone data

Skin Concerns Table

  • Descriptions

  • Symptoms

  • Recommended products

  • Linked references to product database

Recommendations Table

  • Stores final personalised recommendations

  • Links to client record

  • Links to relevant products

Everything was designed so intake data → product tags → AI prompt → recommendation.

⭐ 4. AI Recommendation Logic (ChatGPT + Google Sheets)

I created a custom AI prompt system running through Google Sheets.

Why Google Sheets?
Because it allowed:

  • dynamic prompt construction

  • flexible data references

  • easy Zapier integration

  • repeatable logic that estheticians could understand

The workflow:

  1. Airtable form submission triggers Zapier

  2. Zapier sends skin type, concerns, and desired outcomes to Google Sheets

  3. Google Sheets builds a custom prompt using mapped Airtable product data

  4. Zapier sends that prompt to ChatGPT

  5. ChatGPT returns a personalised, branded skincare recommendation

  6. Zapier stores the recommendation in Airtable + emails the client

This eliminated manual writing and ensured brand-consistent messaging.

⭐ 5. Esthetician Internal Support Table

I built a backend resource used by all estheticians, including:

  • how to apply each product

  • concern-specific usage guidance

  • treatment timelines

  • frequency of use

  • pairing recommendations (e.g., acne + pigmentation)

This created standardisation across all practitioners.

⭐ 6. Automation (Zapier)

I built a multi-step Zapier workflow to handle:

  • Triggering on Airtable form submission

  • Sending client data to Google Sheets

  • Sending prompt to ChatGPT

  • Saving output back into Airtable

  • Emailing the client a branded recommendation

  • Linking all tables and records

The entire workflow ran automatically within seconds.

⭐ 7. System Architecture Overview

Airtable

  • Client database

  • Product catalog

  • Skin concerns database

  • Recommendations table

Google Sheets

  • Prompt builder

  • AI processing intermediary

  • Brand voice logic

ChatGPT API

  • Personalized recommendation generation

Zapier

  • Data movement

  • Automation

  • Email sending

  • System linking

Together, these tools formed a robust, scalable recommendation engine.

⭐ Result

🚀 A fully automated, AI-enhanced skincare recommendation system

The brand now has a standardised, intelligent workflow where:

  • Clients receive personalised skincare plans instantly

  • Estheticians follow consistent product logic

  • Recommendations align with brand tone

  • All data is stored for future sales

  • No manual writing or guesswork is required

Impact Summary

  • Recommendations generated in under 30 seconds

  • Standardised product logic across entire team

  • Eliminated inconsistent esthetician recommendations

  • Created CRM-style client history for upsells & future treatments

  • Set the foundation for future AI-driven follow-up workflows

  • Dramatically reduced time-to-recommendation from 10–20 minutes to near-instant

Business Value Delivered

  • Faster service delivery

  • Higher sales consistency

  • Improved brand professionalism

  • Data-backed product and client insights

  • Scalable backend ready for growth, franchise expansion, or product line increases

This project demonstrates my ability to design cross-platform, AI-powered operational systems that bridge real-world service delivery with modern automation and CRM thinking.

Previous

HubSpot Case Study: Operational transformation