About Tea of Sylhet
Tea of Sylhet is a direct-to-consumer eCommerce brand selling premium regional tea blends. The business relies heavily on paid marketing and needed a measurement system that works reliably on every device.
Project overview
The previous setup suffered from missing purchase values, unreliable event delivery, low-quality remarketing and incomplete GA4 eCommerce data. This hurt decision-making and prevented accurate ROAS measurement.
I implemented a dual ecosystem: Meta Pixel + CAPI for conversion attribution, and GA4 Enhanced eCommerce with custom event tracking for a complete analytics and revenue measurement framework.
Goals
- Send accurate purchase conversions with revenue to Meta Ads
- Enable server-side Meta Purchase events with value
- Implement a full GA4 Enhanced eCommerce funnel
- Track custom micro events for behavior analysis
- Reduce signal loss from cookies, browser limitations and ad blockers
- Boost Event Match Quality for stronger attribution
- Create behavior-based retargeting segments
- Enable automated, ROAS-driven Meta Ads optimization
Challenges
- Purchase events lacked accurate value, product items and revenue in Meta Ads
- GA4 standard eCommerce events were missing or incomplete
- Add to Cart was tracked inconsistently
- Custom micro-interactions were untracked
- Pixel-only tracking frequently failed due to ITP and ad blockers
- Low Event Match Quality (EMQ) led to weak attribution
- No event deduplication framework for Meta
- Audiences were too generic for effective retargeting
Strategy
I combined hybrid server-side tracking for Meta with complete GA4 enhanced ecommerce measurement. A Meta Pixel and Conversion API architecture on a custom tracking domain delivers first-party events reliably, event_id matching prevents duplicates and advanced matching boosts EMQ, while GA4 receives full product-level data and a custom event schema for reporting, optimization and behavior-driven audiences.
What I implemented
GA4 Enhanced eCommerce
- Implemented view_item, add_to_cart, view_cart, begin_checkout, add_payment_info and purchase
- Configured dynamic values: item_id, price, quantity, category and currency
- Enabled full funnel visibility: product → cart → checkout → purchase
GA4 custom event tracking
- Scroll depth tracking (25–100%)
- WhatsApp click tracking
- Coupon apply tracking
- Button interaction events
- Form submission events
Meta Pixel (client-side)
- ViewContent, AddToCart, InitiateCheckout and Purchase
- Included contents array, value, currency and content_ids
- Configured browser event_id for deduplication
Conversion API (server-side)
- Configured a GTM server container via Stape
- Connected a custom tracking domain
- Sent server-side events with enriched customer data
- Enabled hashed email, phone, IP and user agent matching
Deduplication
- Synced event_id between Pixel and CAPI
- Prevented duplicate purchase conversions
Audience & retargeting setup
- ViewContent without AddToCart
- Cart abandoners
- Checkout abandoners
- Repeat purchase audiences
QA & validation
- Meta Events Manager diagnostics
- GA4 DebugView
- Testing with Pixel Helper
- Server log monitoring in Stape
Tools used
Results
- Purchase, Initiate checkout, Add to cart, View content and PageView received from browser and server in Meta Events Manager, with the server copies deduplicated (test environment)
- Server-side Facebook CAPI and GA4 Ecommerce tags fired 44 times each in the server GTM preview, through to purchase
- Meta Pixel AddToCart and InitiateCheckout sent with BDT currency, value, content IDs and a unique event ID for deduplication
- Meta purchase tracking: from missing or inaccurate to 100% accurate server-side revenue events
- GA4 eCommerce: from an incomplete funnel to full product → purchase funnel visibility
- Custom events: from untracked to scroll, WhatsApp, coupon and form events fully tracked
- Event delivery: from Pixel-only to a Pixel + CAPI hybrid
- Match quality: from low to 8–10/10 with advanced matching
- Optimization: from manual to automated, ROAS-driven bidding
