Ads Pipeline Implementation
End-to-end ads data pipeline and mobile attribution system that unified Facebook Ads, Google Ads, and in-app event tracking into a single BI layer for campaign optimization.
Ads Data Infrastructure
An automated ETL pipeline for paid campaign data and a Mobile Measurement Partner integration that gave Shikho’s marketing team unified insight into ad spend and conversion for the first time.
The Problem
Shikho was spending across Facebook Ads and Google Ads but had no centralized data. Reports were exported manually, cross-platform comparison was impossible, and the Android app lacked a Mobile Measurement Partner — meaning installs and in-app events could not be attributed back to the campaigns that drove them. The marketing team was spending on ads without the infrastructure to measure return on ad spend.
Two parallel tracks were launched to fix this:
- Track 1: ETL pipeline via Hevo Data to pull campaign data from Facebook Ads and Google Ads into a centralized Analytics Database.
- Track 2: Adjust MMP integration to track installs, in-app events, and attribution with server-side callbacks into the KPI Database.
The Architecture
Ads Data Architecture
Two parallel data tracks feeding into a unified BI layer
Campaign data flows left-to-right; attribution data flows bottom-to-top — scroll right if needed
- Track 1 pulls campaign performance data from Facebook and Google through Hevo into a PostgreSQL Analytics Database.
- Track 2 uses the Adjust SDK to attribute every install and in-app event to the campaign that drove it, streaming that data into the KPI Database via server callbacks — while also sending optimization signals back to the ad platforms.
Both tracks converge in Tableau.
Hevo Data Pipeline
Hevo connects directly to the Facebook Ads and Google Ads APIs and loads campaign data into the Analytics Database on a daily schedule. Four pipelines were active by end of March, with a 6-month historical backfill for immediate trend analysis.
Pipeline Throughput — End of Q1 2024
Daily event volume across all four active Hevo pipelines
Adjust MMP Integration
Adjust is a Mobile Measurement Partner that attributes every app install and in-app event back to the ad campaign that drove it — answering the fundamental question: which Facebook or Google campaign led to a specific conversion? The Android SDK was integrated across two sprints, enabling:
- Attribution tracking — every install attributed to campaign, ad set, and ad
- Signature v3 — security layer preventing fake install reporting
- In-app events — login, registration, and key user actions mapped to event tokens
- Server-side callbacks — attribution and event data streamed directly into the KPI Database
- Return path — install and conversion signals sent back to Facebook and Google for campaign optimization
Q1 2024 Implementation Timeline
Two parallel tracks delivered across five milestones
What’s Next
The original project plan included several additional components that were not completed in Q1 2024. These items remain on the roadmap for future quarters:
Implementation Roadmap
Q1 2024 delivery status — completed items and planned next steps
Impact
The project replaced manual spreadsheet exports with an automated, unified BI layer. Campaign performance and attribution data now live side by side in Tableau, enabling the team to measure ROAS, identify top-performing campaigns, and feed conversion signals back to the ad platforms for better targeting. The four Hevo pipelines loaded over 3 million events in March 2024 alone, and the Adjust integration established the foundation for all future mobile attribution at Shikho.
More Projects
In-House CRM Transformation
Led the shift from a third-party CRM to a custom in-house sales platform for Shikho's 200-agent telesales team, improving data access, customization, and cost control.
Installment Plan System
Flexible installment-based course purchasing that expanded access to education for students who could not pay upfront.
Quick Start Activation Flow
Redesigned Shikho's post-signup home experience to surface lessons and quizzes immediately, reducing early drop-off and improving new-user activation.