The plumbing nobody wants to think about — until it's broken. Tracking is the foundation every later decision rests on, and most companies' tracking is held together with duct tape. I rebuild it once, properly, on first-party infrastructure you own.
Per implementation. Often paired with PPC or Engine Build.
First-party tracking on your own subdomain
Event taxonomy + naming conventions
BigQuery / Snowflake / DuckDB landing
Stitching anon → known user across devices
Same rhythm every engagement — clear phases, weekly demos, written deliverables. Nothing is "in review" forever.
Map every event, every tool, every dashboard. Find the duplicates, the misses, the broken stuff. You leave with a written assessment.
Design a clean event taxonomy. Naming conventions. Who fires what, when, with which properties. This is the document your team will reference for years.
Server-side first. Client-side as backup. Test events fire correctly across browsers, ad-blockers, and platforms. Validate against the spec.
Your analytics team learns the taxonomy. Your engineers learn to add events without breaking it. Dashboards rebuilt against the new foundation.
Eventually, yes. Even if you're not using one today, I'll set up the pipeline so it lands somewhere queryable. BigQuery is free for the first terabyte — there's no reason not to.
The CMP layer is part of the implementation, not an afterthought. Consent-mode handled correctly so you stay compliant without losing the signal.
Sometimes — but in 80% of audits, the existing setup has accumulated enough drift that a clean rebuild is faster than patching. I'll tell you honestly.
Production-grade websites built to convert and ship fast. Marketing sites, app shells, landing pages.
Google, Meta, LinkedIn, and TikTok campaigns instrumented end-to-end. Built on offline conversions, not vanity metrics.
The operational software a growing company needs but never has time to build — invoicing, admin, ops dashboards.