Growth Engineering Playbook · Wave 4

Your analytics sees barely half of what happens.

Client-side tags lose events to ad-block, ITP and cookie expiry, and beacons that never fire — and the loss is biased by browser and device, so it quietly distorts every channel number. Server-side tracking wins most of that back, but it is not a consent workaround, and running client and server together double-counts unless you dedup on a shared event id. This models the fictional Northstar Outfitters traffic through each architecture, draws the consent ceiling no stack can beat, and separates the loss you can recover from the loss that's gone regardless. Deterministic, client-side.

Loading traffic model…

Events captured vs what actually happened

Blue = captured under the selected architecture. Amber line = the consent ceiling (the most any stack can legally see).

Captured Consent ceiling Double-counted (no dedup)

Where the client loses events

One segment's client-side loss, step by step.

Capture by segment

Client-side is lossy and the loss is uneven; server-side recovers the technical loss but never the consent loss. The last column is what a naive hybrid would over-report.

The Decision