How StickEarn Built a Shared Business Intelligence Foundation

Free image: “Analytics graphs on a MacBook screen” by Mikael Kristiansen, CC0, via Wikimedia Commons.

StickEarn needed more than another dashboard. It needed a business intelligence foundation that could pull together data from different teams, create one shared source of truth, and make performance visible across the company.

Before that foundation existed, reporting lived in silos. Each department tracked its own numbers, definitions drifted, and nobody could easily compare performance across functions. Even when the numbers were available, they were not always talking about the same thing.

The result was familiar:

• different teams reported different versions of the same metric
• multi-source analysis took too long
• leaders spent time reconciling reports instead of making decisions
• no one had a single operating view of the business

StickEarn solved that by treating BI as an operating layer, not just a reporting tool.

What the stack needed to do

The implementation had to do four things well:

1. collect data from several operational tools
2. load it reliably into a central warehouse
3. standardize definitions so metrics meant the same thing everywhere
4. serve that data in a dashboard layer the whole company could actually use

The stack that supported this was:

• Airflow for orchestration
• Airbyte for ingestion
• BigQuery as the warehouse
• Zoho, MySQL, Sheets, Jira, and Trello as source systems
• Metabase as the serving and exploration layer

That combination sounds straightforward on paper. The hard part was not the tools. The hard part was making the business agree on what the data meant.

Before: performance was trapped in silos

In the earlier setup, each department had its own way of tracking work.

Sales looked at one set of numbers. Operations looked at another. Project delivery had its own reporting rhythm. Leadership had to ask multiple people before getting a full picture.

That creates a hidden tax:

• meetings become reconciliation sessions
• dashboards lose trust
• teams optimize local metrics instead of company outcomes
• questions take longer to answer than they should

The real issue was not just reporting. It was the lack of a shared metric foundation.

The foundation: one pipeline, one warehouse, one logic layer

The first step was to move data into a place where it could be shaped consistently.

1. Ingest the sources

Airbyte handled the movement from source systems into BigQuery.

That mattered because the company was not working with a single clean source. It had operational data scattered across:

• CRM and workflow tools
• project trackers
• spreadsheets
• transactional databases

Each source had its own quirks, refresh patterns, and field conventions. Airbyte gave the team a repeatable way to bring all of that into the same warehouse.

2. Orchestrate the pipeline

Airflow controlled the flow end to end.

Instead of running ad hoc jobs or manual exports, the pipeline could be scheduled, monitored, and retried in a predictable way. That improved reliability and made failures visible quickly.

3. Centralize the model in BigQuery

BigQuery became the place where the company could normalize and join the data.

Once the raw inputs landed there, the team could:

• align field names and business logic
• create shared dimensions and fact tables
• join data from multiple departments
• build reporting models once, not separately for every team

That is the point where BI stops being a collection of charts and starts becoming a foundation.

4. Serve the business through Metabase

Metabase became the place where people actually used the data.

It worked well because it made the warehouse accessible without forcing everyone into a technical workflow. Leaders and operators could consume one dashboard, explore data, and trust that they were looking at the same definitions.

After: one company view instead of many disconnected ones

The biggest outcome was not just cleaner infrastructure. It was a change in how the company operated.

StickEarn moved from siloed reporting to one shared dashboard for the whole company.

That meant:

• performance could be reviewed in one place
• cross-department analysis became much easier
• teams could join data from multiple sources without rebuilding everything each time
• decision-making got faster because the numbers stopped arguing with each other

When one dashboard becomes the company default, alignment improves quickly. People stop debating which spreadsheet is correct and start discussing what the numbers mean.

The numbers that matter

The implementation produced measurable stability and usage:

• 99% pipeline uptime
• 70-80% Metabase retention measured through WAU/MAU

Those are strong signals for a BI system.

High uptime means the warehouse and orchestration layer are dependable enough to support daily work. Healthy retention means people are coming back to the dashboard because it is useful, not because someone told them to check it once.

In practice, retention is often the better metric. A dashboard only matters if teams keep using it.

Why this worked

The stack worked because it respected a few simple principles.

One source of truth, not one source system

The company did not pretend all data lived in one tool. It accepted that the truth would come from several operational systems, then built a single place to unify them.

Metric definitions came first

The business needed a shared language before it needed prettier charts.

That is why the foundation mattered so much. If teams disagree on what a metric means, a dashboard only amplifies the confusion.

The dashboard was designed for the company, not the analyst

Metabase was useful because it made the data visible to decision-makers.

The point was not to create a sandbox for one power user. The point was to give the whole organization a common operating surface.

Reliability was treated as a feature

Pipeline uptime was not a background detail.

If data fails silently, trust drops fast. The 99% uptime target helped turn the BI layer into something teams could depend on.

What this changes in daily work

Once the BI foundation was in place, a few things become easier:

• leaders can see company performance without waiting for manual rollups
• teams can compare metrics across departments on the same basis
• source systems can change without breaking the entire reporting process
• new dashboards can be built on top of the same warehouse logic

That is the difference between reporting and infrastructure.

Reporting answers a question. Infrastructure changes the speed and quality of the whole organization.

Lessons from the implementation

For teams planning a similar BI foundation, a few lessons are worth keeping:

1. Start with the decisions that need support, not with the tool list.
2. Make metric definitions explicit before building dashboards.
3. Treat source systems as inputs, not as final truth.
4. Centralize transformations in the warehouse where possible.
5. Design the serving layer for broad access, not just technical users.
6. Measure adoption as seriously as uptime.

That combination is usually what turns a data project into an operating system for the company.

The broader lesson

StickEarn’s shift shows that business intelligence is most valuable when it connects sources, definitions, and people.

Airflow and Airbyte handled the movement. BigQuery handled the model. Metabase handled the access. The business got a shared language and a shared view.

That is what a BI foundation is supposed to do.

It does not just show performance. It makes performance visible, comparable, and usable across the whole company.