Analytics

Full-Stack Analytics Dashboard & Attribution Engine - React & Node

Lead Full-Stack Engineer & Automation Architect

Business summary
High-Ticket E-Learning & Digital Products • Real-time analytics dashboard & dual-touch cohort attribution engine
  • Recovered ~$16,000/month in unattributed sales with sub-1% untracked rate
  • Replaced fragile 14-tab Google Sheet with sub-second real-time React analytics app
  • Engineered localStorage UTM pass-through for sandboxed SamCart checkout iframes

What I built

Replaced a fragile 14-tab Google Sheet reporting system with a real-time analytics web dashboard (React 18, TypeScript, Node.js, Express, Tailwind CSS, Recharts). Engineered a proprietary dual-touch cohort attribution engine resolving first-touch Meta ad acquisition against downstream email nurture conversion, recovering ~$16,000/mo in unattributed sales. Developed client-side localStorage pass-through scripts for sandboxed SamCart iframes, an in-app UTM Builder, and webhook provisioning relays connecting SamCart checkouts to instant Podia course delivery.

Tech stack

ReactTypeScriptNode.jsMeta Graph APIConvertKitReact 18ExpressTailwind CSSRechartsKitSamCartPodiaManyChatZapierMake.comAttribution EngineWebhooksCohort AnalysisUTM TrackingDual-Touch Attribution
Deep Dive Case Study

Full-Stack Analytics, Attribution Engine & E-Commerce Infrastructure

Role: Lead Full-Stack Engineer & Automation Architect
Client: Arica Angelo (High-Ticket E-Learning & Digital Product Brand)
Tech Stack: React 18, TypeScript, Node.js, Express, Tailwind CSS, Recharts, Meta Graph API, Kit (ConvertKit), SamCart, ManyChat, Podia, Zapier, Make.com
Timeline: August 2026 - Present
Key Outcomes: Replaced fragile 14-tab Google Sheet system with a real-time web dashboard; recovered ~$16,000/mo in unattributed sales; engineered dual-touch cohort tracking for Meta evergreen funnels; automated checkout-to-course provisioning for thousands of customers.


1. Executive Summary

Arica Angelo is a high-volume direct-to-consumer digital education brand driving millions in revenue through Meta Ads, YouTube, email marketing, and automated Instagram funnels.

The business suffered from severe attribution blindness: Facebook API tokens frequently failed, SamCart checkout pages stripped UTM parameters across multi-step funnels, and revenue was tracked manually via a fragile 14-tab Google Sheet.

I was brought in to lead the technical infrastructure overhaul. Over the course of the engagement, I:

  1. Built a custom, real-time analytics web dashboard (React, TypeScript, Express) replacing the entire Google Sheets reporting pipeline.
  2. Engineered a proprietary dual-touch attribution engine resolving first-touch acquisition cohorts (Meta Ads) against conversion-touch marketing (email nurture sequences).
  3. Solved a persistent multi-step funnel tracking issue by engineering client-side localStorage pass-through scripts that operate inside isolated SamCart iframes.
  4. Developed an in-dashboard UTM Builder & Link History system enforcing strict canonical source conventions across the entire marketing team.
  5. Maintained automated provisioning pipelines connecting SamCart orders to Podia course delivery via deduplicated webhook relays.

2. The Challenge & Technical Debt

When I inherited the project, the tracking and automation stack had reached a breaking point:

  • Single-Point-of-Failure Reporting: All reporting relied on a 14-tab Google Sheet joined via complex spreadsheet formulas. Token expirations in Make.com scenarios silently halted Meta ad spend syncs, distorting ROAS calculations for weeks before detection.
  • The "No Source" Attribution Leak: Over ~$16,000/month in course sales landed with blank UTM sources. SamCart only records UTMs present on the URL at the exact millisecond of checkout; visitors clicking an opt-in page and progressing through multi-page video sales letters (VSLs) lost all query parameters.
  • Iframe & Server-Side Redirect Barriers: SamCart landing pages render inside sandboxed about:srcdoc iframes (blinding normal location.search scripts). Furthermore, SamCart's native lead-capture forms execute server-side 302 redirects that completely strip query parameters.
  • First-Touch vs. Last-Touch Conflict: Paid Meta ads acquired leads through a free opt-in (5 Naughty Things). When leads purchased courses 5-14 days later via email nurture broadcasts, traditional last-click analytics attributed 100% of revenue to email, making profitable ad campaigns appear negative in ROAS.
  • Platform Counting Discrepancies: Course ownership counts in the delivery LMS (Podia) conflicted with checkout reports (SamCart) due to bundle purchases, upsell bumps, and unrecorded historical datasets.

3. End-to-End System Architecture

[ TRAFFIC SOURCES ]
Meta Ads (FB/IG) ───► Dynamic Ad Tokens: {{ad.id}}, {{campaign.name}}
YouTube Videos   ───► UTM-tagged description & pinned comment links
Email Broadcasts ───► ConvertKit Broadcast & Automation Flow links
ManyChat (IG DM) ───► Automated Keyword Trigger Links
         │
         ▼
[ FUNNEL & CLIENT-SIDE INGESTION ]
Landing Pages / SamCart VSLs
   ├── Script A (Opt-in): window.top search param capture ──► localStorage('aa_utm')
   ├── Form Injection: MutationObserver injects hidden inputs into lead forms
   └── Script B (Next Step): Rehydrates query params onto all outgoing CTA buttons
         │
         ▼
[ PLATFORM EVENTS & WEBHOOKS ]
   ├── Kit (ConvertKit)  ──► Lead Opt-in Webhook (Email, Ad ID, utm_source)
   ├── ManyChat          ──► DM Lead Webhook (Subscriber data)
   └── SamCart Checkout  ──► Order Completed Webhook (Transaction, SKU, Price)
         │
         ▼
[ CUSTOM DASHBOARD BACKEND (Node.js / Express) ]
   ├── Ingestion Endpoints (/webhooks/samcart, /webhooks/kit, /webhooks/manychat)
   ├── 90-Day Identity Resolution Graph (Matches orders to leads by email)
   ├── Dual-Touch Attribution Engine (Acquisition Cohort vs. Conversion Touch)
   ├── Deduplication & Feed Heartbeat Monitor (503 circuit-breaker for stale feeds)
   └── Webhook Relay Dispatcher ──► Zapier ──► Instant Podia Course Grant
         │
         ▼
[ EXECUTIVE REACT DASHBOARD ]
   ├── Executive KPI & ROAS Overview (Blended vs. Isolated Channel Metrics)
   ├── Role-Based Access Control (Admin, Media Buyer, Email Marketer)
   ├── Evergreen Funnel Waterfall (Meta Spend ➔ CPL ➔ Tripwire ➔ Upsells ➔ Cohort ROAS)
   ├── Course Market Penetration Intelligence
   └── Automated UTM Builder & Sanitizer Module

4. Key Solutions & Engineering Implementations

Solution 1: Real-Time Analytics Web Application (React + TypeScript)

  • High-Performance Architecture: Built a single-page analytics application featuring sub-second aggregation over thousands of orders, leads, and ad-spend data points using Recharts and Tailwind CSS.
  • Role-Based Access Control (RBAC): Configured strict role filtering (admin, facebook_media_buyer, email_marketer). For media buyers, the backend filters out non-Meta revenue, isolating true Meta ROAS, CPC, and CPL without data leakage from organic or email sales.
  • Resilient Backend & Sync Engine: Created a Node.js/Express service with automated reconciliation workers, feed heartbeat monitors (triggering health warnings if feeds go stale), and automated data recovery scripts.

Solution 2: Client-Side UTM Pass-Through Engine for SamCart Iframes

To eliminate the ~$16,000/month "no source" tracking gap across multi-step sales funnels, I developed a lightweight JavaScript pass-through engine:

  • Overcoming Iframe Sandboxes: Scripts detect if they are embedded within SamCart's about:srcdoc iframe container and safely reach the parent window via window.top.location.search || window.parent.location.search.
  • State Handoff via localStorage: Because SamCart's lead capture executes server-side 302 redirects that wipe URL parameters, Script A captures incoming UTMs on first landing and stores them in localStorage.
  • Dynamic MutationObserver Injection: When SamCart renders dynamic checkout forms or buttons, a MutationObserver automatically attaches hidden input fields and appends UTM parameters to all checkout CTAs, guaranteeing 100% parameter preservation.

Solution 3: Dual-Touch Cohort Attribution Engine

Solved the multi-channel attribution gap between paid acquisition and email nurture:

  • Email as the Immutable Identifier: Leveraged the customer's email address (contact_email) as a permanent cross-device identity anchor, bridging initial Meta mobile ad opt-ins to later desktop checkouts.
  • Dual Attribution Model:
    • Acquisition Cohort (First Touch): Permanently associates the customer with the original Meta ad campaign (e.g. 5 Naughty Things). This cohort attribution is locked and never overwritten.
    • Conversion Touch (Last Touch): Captures the exact email broadcast, automation flow, or YouTube video that triggered the transaction.
    • Cohort Waterfall: Aggregates total downstream customer value ($27 Tripwire → $97 Core Offer → $197 Upsell → $147 Downsell), allowing the business to determine True Cohort ROAS and Revenue Per Lead (RPL).

Solution 4: Intelligent In-App UTM Builder & Team Governance

  • Sanitization Engine: Created a custom regex slugifier that prevents query-string corruption (e.g., converting raw & to and, removing # and % escape characters).
  • Automated Parameter Mapping: Enforces the platform's convention where paid checkouts receive utm_content=[slug] and lead magnets receive utm_term=[slug].
  • Team Link History: Logs every generated link into an internal database with use counts, duplicate warnings, and copy-to-clipboard functionality.
  • Executive SOPs: Authored standardized SOP documentation (Markdown and formatted Word .docx) governing YouTube, ConvertKit, Website, ManyChat, and Meta Ads.

Solution 5: Webhook Processing & Automated Course Delivery

  • Engineered an automated webhook receiver for SamCart order completions.
  • Implemented payload validation, duplicate transaction suppression, and immediate asynchronous forwarding to Zapier to enroll students into Podia courses.
  • Built an internal webhook logging and simulation view with visual payload inspectors for technical troubleshooting.

5. Measurable Business Impact

Metric Before Optimization After Implementation Impact
Unattributed Sales ("No Source") ~$16,000 / month < 1% of total orders Recovered complete visibility on ~$190k/year in revenue
Reporting Latency Manual Google Sheet updates (Hours/Days) Sub-second real-time dashboard Instant visibility into daily ROAS and ad performance
Media Buyer Visibility Blended store revenue confused ad metrics Strict Meta-only KPI view Accurate ad scaling decisions by media buyer
Funnel LTV Tracking Zero downstream tracking from opt-in ads Full evergreen cohort waterfall Direct calculation of True ROAS and Revenue Per Lead
Link Integrity 10% of spreadsheet links broken by unencoded & 100% sanitized via UTM Builder Zero attribution loss from malformed URLs

6. Key Competencies Demonstrated

  • Full-Stack Web Development: React, TypeScript, Node.js, Express, Tailwind CSS, REST APIs, WebSockets, Vite.
  • Data Engineering & Analytics: Multi-touch attribution modeling, cohort analysis, identity resolution graphs, data normalization, ingestion pipelines.
  • E-Commerce & MarTech Ecosystems: SamCart, ConvertKit (Kit), Meta Ads Manager & Graph API, Podia LMS, ManyChat, Zapier webhooks.
  • Technical Product Management: Technical writing, standard operating procedure (SOP) authoring, stakeholder communication, role-based workflows.