Skip to content

I set up tracking, clean the data and build reporting that is easy, automated, accurate.

I'm Joe. For ten years I have tagged websites for analytics and ad platforms, cleaned and stored the data in a warehouse, and built the reports on top. I work in Google Tag Manager, GA4, dbt, Fivetran, Airbyte, Supermetrics, BigQuery, Snowflake, Databricks, Power BI and Looker Studio. Let's get you out of copy-pasting Excel files and QA'ing data every day.

Tracking
GA4 / Tag Manager / Google Ads / Meta
Warehouses
BigQuery / Snowflake / Databricks
Reporting
Power BI / Looker Studio
Engagements
From a one-week audit to an eight-week build

What I do, in the order the data moves.

It's almost always a data management problem.

Nobody actively owns the data. Definitions drift between teams. Mapping tables go stale. The models that matter were built by someone who left, and the documentation is a Notion page from 2023 half the team forgot exists.

You feel it two ways. First, errors. Two reports for the same metric that don't match, dashboards nobody trusts, weekly QA calls to reconcile numbers by hand before anyone can share them.

Second, no automation. Every source change breaks a pipeline. Every new question becomes a ticket. The stack that was supposed to run itself needs a human in the loop every day to keep the numbers honest.

Good stacks assume things will break.

Pipelines break. Sources change schemas without warning. Business logic evolves. That's normal. Errors happen in every stack. What sets a healthy one apart is how fast you find out about them.

That means QA checks built into the models themselves, running every time the pipeline runs: row counts, uniqueness, freshness, expected ranges, error flagging. And an alerting system that pings the right person the moment a check fails, not the next time someone opens a dashboard.

The goal is a stack that tells you the second something breaks, so you can fix it before anyone reports on a bad number.

Two other ways to work with me.

You don't need a full-time hire. You need experience.

I rewrite the models that matter, build what's missing, and fix what's silently wrong. Then I stick around long enough to hand it over properly, so your team can keep it running without me.