Data Analytics and Business Intelligence Encantos

Encantos’ North Star Goal

Encantos collaborated with Analyze Agency to develop a data-driven learning analytics system designed to analyze usage patterns and enhance bilingual education for children. The goal was to create a scalable architecture that could provide deeper insights into how childcare and educational experiences could be improved.

From a product standpoint, the objective was to transition from fragmented, tool-specific reporting to a unified, real-time content intelligence system. This would empower decision-making across product development, content creation, and marketing teams.

‍

The production data was stored in Cloud SQL on Google Cloud Platform (GCP), while critical engagement metrics were scattered across different platforms: Braze for customer engagement, Segment for product event tracking, and RevenueCat for subscription and revenue management.

‍

The primary technical challenge wasn’t the availability of data but its fragmentation. This created gaps in understanding a child’s learning journey, as engagement signals were isolated in separate systems and analyzed independently. Analyze Agency worked closely with Encantos’ Data and Engineering teams to build a unified analytics platform. This platform integrated content interactions, enriched them with catalog metadata, and delivered actionable key performance indicators (KPIs) to business stakeholders in real time through Looker.

‍

Every interaction whether a video watched, a book read, a song played, or a game completed on Canticos became a measurable learning event. The key metric driving this initiative was the number of Weekly Active Learning Sessions per Subscribed Child on Canticos. By adopting a unified GCP data stack, Encantos gained the analytical power to refine and optimize content, reduce subscriber churn, and deliver a highly personalized bilingual learning experience at scale.  

‍

The Problem

Canticos had built a compelling bilingual learning product with a rapidly expanding catalog including videos, books, songs, and games. However, the data tracking how children engaged with this content was scattered across three separate systems: Braze for customer engagement, Segment for product event tracking, and RevenueCat for subscription and revenue management. Meanwhile, production catalog metadata resided separately in Cloud SQL.

‍

This fragmentation meant there was no unified view of a child’s learning journey. Content teams could only address basic questions: Which videos were children completing? Which books were left unfinished? Did high game completion rates correlate with subscriber retention?

‍

Without a centralized data strategy, decision-making relied more on intuition than intelligence. There was no single source of truth for campaign performance, content investment, or product prioritization.

‍

Most critically, Canticos lacked a robust analytics layer to connect content usage with business outcomes. As a result, they were operating without clear insights, struggling to keep pace with market demands.  

‍

Evolution of Client’s stack with Analyze Agency

Analyze Agency partnered with Encantos’ engineering and product teams to transform a fragmented data landscape into a unified, GCP-native analytics architecture.

‍

The implementation began with Segment, which was already integrated into the Canticos app for product event tracking. Rather than introducing additional ingestion tools, Analyze Agency positioned Segment as the central data pipeline. Braze was integrated via Braze Currents, streaming CRM and campaign engagement events into Segment. RevenueCat was connected using its native HTTP API Source connector, allowing in-app purchase events including trials, subscriptions, and cancellations to stream directly into Segment in real time.

‍

Content catalog metadata for Canticos including Videos, Books, Songs, and Games was stored in Cloud SQL. This data was synchronized directly into BigQuery using Change Data Capture (CDC), outside the Segment pipeline.

‍

All event streams ultimately flowed into Google BigQuery, which served as the single source of truth. Analyze Agency then implemented a dbt transformation layer within BigQuery. dbt models read the raw ingested tables, joined engagement events with content catalog metadata using top-level content identifiers, and produced clean, analytics-ready data models.

‍

Looker was layered on top of BigQuery, with a LookML semantic layer powering four dedicated dashboards. At the core is the Content Library KPI Dashboard, which tracks completion rates across each type of content catalog.

‍

The result was a fully decoupled, real-time analytics stack built entirely within GCP.

‍

‍Architecture

Encantos built a three-layer, GCP-native data architecture to bring together engagement data, subscription activity, and content metadata into a single analytics platform.

‍

Layer 1: Data Sources
The pipeline brings together data from four core systems: Braze (CRM and customer engagement), Segment (product event tracking), RevenueCat (subscription and billing data), and Cloud SQL, which stores metadata for the Canticos content catalog Videos, Books, Songs, and Games.

‍

Layer 2: Segment as the Central Hub
Segment acts as the primary event pipeline. Braze is connected through Braze Currents, while RevenueCat sends events through its HTTP API Source connector. Product events from the Canticos app already flow directly into Segment. From there, all events are routed to BigQuery using Segment’s native warehouse destination.

‍

Content catalog metadata stored in Cloud SQL is streamed directly into BigQuery using Change Data Capture (CDC), bypassing the Segment pipeline.

‍

Layer 3: GCP Analytics Layer
BigQuery serves as the central data warehouse where all raw data lands. A dbt transformation layer runs inside BigQuery, joining engagement events with content metadata and producing analytics-ready models.

‍

On top of this, Looker provides the analytics interface. A LookML semantic layer powers four dashboards that allow teams to explore content performance, engagement patterns, and key product metrics.

‍

Implementation  

Three challenges emerged during the implementation of the Content Library KPI Dashboard.  
‍

• The first challenge was aligning user identities between Segment and RevenueCat, since Canticos used different identifier types within its authentication layer and SDKs, and thus needed to be reconciled before subscription events could be appropriately joined to content engagement data in BigQuery.  

• The second challenge was to make sure that content_id could serve as a reliable join key between Cloud SQL’s catalogue metadata, and Segment’s consumption events which required validation directly against the instrumentation of the app to verify every track () event included the right identifier.  

• The third, and most devilishly nuanced challenge was determining what “completion” meant for each type of catalogue.  
  ‍

A completed video, a finished book, a fully Listened song and a concluded game session each needed its own threshold agreed with the Encantos product team and consistently instrumented in Segment before it would carry any kind of analytical meaning in KPI calculations powered through Looker.

The first two weeks were focused on discovery and alignment. Existing schemas across Braze, Segment, RevenueCat, and Cloud SQL were audited, and a unified data dictionary was created to map fields across all four systems. A shared content_id was established as the join key between content catalogues and engagement events. During this phase, COPPA compliance boundaries across the pipeline were also defined, and Encantos’ product team aligned on the North Star metric and KPI framework.

‍

The next phase was focused on establishing the data pipeline. Segment was positioned as the central hub, with Braze integrated through Braze Currents and RevenueCat connected via its HTTP API Source connector. User identities were aligned across the Braze, Segment, and RevenueCat SDKs to ensure consistent tracking. Product, identity, and purchase events were validated as they flowed through Segment before being routed to BigQuery through Segment’s native warehouse destination. In parallel, content catalogue metadata for Videos, Books, Songs, and Games was streamed from Cloud SQL into BigQuery using Change Data Capture (CDC).

‍

The final stage was focused on building the analytics layer on GCP. dbt models were created to transform raw data into analytics-ready datasets by combining user profiles, engagement events, subscription activity, and content catalogue metadata. A content KPI model joined engagement events with catalogue metadata through content_id, and data quality tests validated each transformation. Looker was then layered on top of BigQuery with a LookML semantic model powering four dashboards: Content Library KPI, Learning Engagement, Campaign Effectiveness, and Subscription + Engagement. The project concluded with pipeline monitoring, governance controls, documentation of full data lineage, and knowledge transfer to the Encantos data team. The total implementation time was 10 weeks.

‍

Evaluation of the Model  

The GCP-native analytics stack provided Canticos with an unprecedented, unified view of how children engaged with its content a first for the team.

‍

Previously, there was no aggregated visibility into content performance across videos, books, songs, and games. After implementation, the Encantos product and content teams gained their first real-time, consolidated dashboard in Looker. This dashboard displayed completion rates across all four content types (post-KPI definition), segmented by language and storyworld, offering actionable insights from the Content Library.

‍

The dbt transformation layer resolved the critical challenge of joining content_id between Cloud SQL’s catalog metadata and Segment’s consumption events. This created a reliable content_kpi_model, establishing a single source of truth for all content performance metrics.

‍

Architecturally, Segment served as the central pipeline hub, seamlessly integrating Braze Currents, RevenueCat’s HTTP API, and native product events into a unified BigQuery destination. This eliminated the data silos that had previously hindered cross-system analysis.

‍

Most importantly, this architecture directly supported Canticos’ North Star metric: Weekly Active Learning Sessions per Subscribed Child. For the first time, Encantos had the instrumentation to observe, analyze, and act on this metric effectively. Content investment decisions, Braze re-engagement campaigns, and subscription retention strategies could now be driven by real child engagement data not intuition.  

‍

Future Priorities

The foundation of Encantos' data evolution on GCP centered around the Content Library KPI Dashboard. The next generation was focused on fostering intelligence in three dimensions
‍

• First, personalization at scale using BigQuery ML to generate child level content recommendation models based on completion history, language preference and storyworld affinity, surfaced back through the Canticos app.

• Second, extending use case coverage, we have instrumented our current architecture to deliver the other 4 use cases for subscription churn prediction, campaign effectiveness, LTV modelling, and COPPA compliant data governance simply ready in place to be enabled on existing stack without any additional ingestion effort.

• Third, realtime alerting, a brand new functional aspect of Looker where alerts are triggered for the Encantos team when key KPIs (e.g. video completion rates and weekly active sessions) fall below defined thresholds to help them react based on content and engagement signals instead of retrospectively.

‍

Why Choose Us?

Analyze Agency operates at the intersection of Data Engineering, Analytics, and Business Intelligence, designing architectures that are not only technically robust but also closely aligned with measurable business outcomes. Our approach prioritizes clarity and purpose with every architectural decision guided by the organization’s North Star metric.

‍

During the Encantos engagement, we unified four previously disconnected source systems into a single GCP-native analytics stack. This implementation provided the Canticos team with its first real-time, consolidated view of content engagement across Videos, Books, Songs, and Games. The outcome was a reliable analytics foundation that transformed raw operational data into decision-ready insights accessible through structured dashboards.

‍

Our methodology is grounded in a balanced understanding of data, product, and business context. By aligning these three dimensions, we build data platforms that provide a direct, dependable path from raw data to actionable intelligence enabling teams to make informed decisions with confidence.

‍

Our Success Framework

The Analyze Agency success framework begins with a clearly defined problem statement. In the case of Encantos, the challenge was straightforward with four separate source systems and no unified visibility into how children engaged with Canticos content. This clarity allowed us to focus on architecture design before implementation. The resulting design was intentional from positioning Segment as the central event pipeline, to establishing Cloud SQL CDC as the metadata backbone, and defining content_id as the primary join key before a single dbt model was written.

‍

Our design philosophy is based on a layered architecture where each component has a clear purpose and ownership. From the data sources through to the Looker dashboards, every layer of the system has a defined role, an accountable owner, and an explicit data contract. Implementation followed the design in structured phases, with each step validated and consistently tied back to the North Star metric.

‍

The outcome was more than just a technical architecture. It established a repeatable analytics foundation that Encantos can extend, scale, and govern with confidence.

‍

Get In Touch

If you are looking to modernize your data infrastructure, unify fragmented data systems, or turn raw data into actionable insights, we would be glad to help.

‍

Analyze Agency specializes in designing and implementing scalable, real-time data architectures that connect engineering, analytics, and business intelligence. Our focus is on building systems that not only work reliably, but also support better and faster decision-making across your organization.

‍

Contact us at Discovery@analyze.agency or visit Analyze.Agency to start the conversation. We look forward to learning about your challenges and exploring how we can support your data and analytics goals.