ecem.fyi
Back to projects
2026·Designer & Developer

Attribution Tycoon — The Tracking Wars

A 30-day budget simulation where making the sale is not enough — you have to be able to measure it. Four agents model traffic, signal loss, EMQ and ad-platform bidding.

TypeScriptReactServer-Side TaggingMeta CAPIConsent Mode v2Canvas

A measurement-infrastructure simulation built as four independent agents that hand off to each other every simulated day: a Customer Traffic agent generates a cohort with a real device/browser/ad-blocker/consent composition, a Tracking Engine decides per segment which of those events survive, an Ad Platform agent prices delivery from the surviving signal, and a Market Event agent fires the real-world crises — Apple's Link Tracking Protection, Google Cookielock, TikTok's in-app browser lockdown. Retention is S_retention = (1 − AdBlocker_eff) × (1 − SignalLoss) × PipelineIntegrity, and that retention plus your installed modules produce an EMQ score the bidding engine turns into cost: CPA_effective = CPA_base ÷ (EMQ / 10). The point of the game is the gap this produces. A client-side-only stack lands at EMQ ≈ 3.4, which turns Meta's $20 base CPA into ~$59 and shows 0.4x ROAS on the dashboard while reality is 1.3x — so the player is tempted to cut budget on a channel that is actually working. Install CAPI without event deduplication and the trap inverts: the dashboard now reports more revenue than happened.

Try it

Highlights