Advertising Data Flow

What data is critical in driving decision making in advertising?

Terminology

Advertising has its own nomenclature. The naming conventions reveal where power concentrates and where it shifts.

Term	What It Means	Who Named It	What It Hides
Impression	An ad was served to a screen	Publishers	Whether a human saw it, or a bot loaded it
Click	Someone interacted with the ad	Platforms	Whether intent was real or accidental
Conversion	A desired action completed	Advertisers	Which touchpoint actually caused it
Attribution	Credit assigned to a touchpoint	Measurement	The model's assumptions about causation
Reach	Number of unique users exposed	Platforms	Self-reported, grading their own homework
ROAS	Revenue per dollar of ad spend	Performance	Whether the revenue would have happened anyway
CPM	Cost per thousand impressions	Media buyers	That "impression" and "attention" are different things
Reconciliation	Matching delivery to payment	Finance	30-90 days of uncertainty treated as normal

Web3 renaming "impression" to "verified attestation" and "reconciliation" to "settlement" to reflect shifts ownership.

The Data Model

Understanding the industry starts with understanding its data footprint — the entities, relationships, and state transitions that drive every decision. This IS the knowledge schema for advertising.

Entity	Relationships	State Transitions
Identity	Joins impressions to conversions, owns audience membership	Anonymous → Cookied → Opted-in → Wallet-linked
Impression	Belongs to campaign, targets identity, serves creative	Served → Viewed → Clicked → Converted → Settled
Campaign	Has budget, targets audience, deploys creatives	Briefed → Targeting → Live → Optimizing → Reconciled
Creative	Belongs to campaign, served via impression	Created → Tested → Winner/Loser → Fatigued
Conversion	Attributed to impression(s), generates revenue	Detected → Attributed → Verified → Settled
Settlement	Joins impression + conversion + payment	Invoiced → Disputed → Reconciled (30-90d) or Instant

Every principle, metric, protocol, tool, and player operates on these six entities. The entire $1 trillion industry exists to make the Identity ↔ Conversion join work. Apple broke that join. Wallets restore it — with ownership inverted.

Data Flow

How data moves through the advertising system — from intent to settlement.

ADVERTISER                    PLATFORM                      PUBLISHER
    │                            │                              │
    ├─ Campaign briefed ────────►│                              │
    │  (budget, audience, goal)  │                              │
    │                            ├─ Audience matched ──────────►│
    │                            │  (identity join)             │
    │                            │                              ├─ Impression served
    │                            │                              │  (creative rendered)
    │                            │◄─ Signal returned ───────────┤
    │                            │  (view, click, conversion)   │
    │◄─ Attribution reported ────┤                              │
    │   (which touchpoints)      │                              │
    │                            │                              │
    └─ Settlement (30-90 days) ──┼──────────────────────────────┘

The critical join: Step 2 (Audience matched) requires the identity entity. When Apple's ATT removed the cross-site join key, the entire flow from "matched" to "attributed" broke for ~30% of users.

Web3 alternative: The wallet IS the identity. The blockchain IS the attribution database. Settlement is atomic — impression + verification + payment in one PTB.

Data Footprint

Map against the commissioning instrument — what maturity exists across five layers?

Data Domain	Schema	Data	API	UI	Feedback Loop	Moat Signal
Identity	Proven	Proven	Proven	Proven	Proven	Platform lock-in (Google, Meta)
Impression	Proven	Proven	Proven	Proven	Exists	Self-reported, unverified
Audience	Proven	Proven	Proven	Proven	Exists	First-party data = competitive edge
Attribution	Exists	Exists	Exists	Exists	None	Broken — last-click default
Creative	Proven	Proven	Proven	Proven	Exists	AI-generated, fast commoditizing
Settlement	Exists	Exists	None	None	None	30-90 day delay = hidden fee
Verification	None	None	None	None	None	The gap Web3 fills

Where the moat lives: Identity and Audience — whoever owns the join between these two owns the margin. Google and Meta's moat is not their ad serving technology. It's their identity graph.

Where the gap lives: Attribution, Settlement, and Verification — three domains where data is either broken, delayed, or nonexistent. This is the Web3 opportunity.

Decisions Data Drives

Decision	Data Required	Entity	Current Quality	Impact of Bad Data
Where to spend	Channel ROAS, incrementality	Campaign	Medium	Budget allocated to wrong channels
Who to target	Audience segments, lookalikes	Identity	Declining	Wasted impressions on wrong people
What creative to run	A/B test results, fatigue signals	Creative	High	Low CTR, creative burnout
When to stop	Frequency caps, diminishing returns	Impression	Medium	Ad fatigue, negative brand impact
What actually worked	Multi-touch attribution, lift testing	Conversion	Low	Wrong channel gets credit
How much to pay	Verified delivery, fraud detection	Settlement	Low	Paying for bot traffic

The bottom two rows — attribution and settlement — have the lowest data quality and the highest financial impact. That's not a coincidence. It's a business model.

Context

• Advertising Industry — The FACT hub
• Advertising Principles — The five immutable truths
• Naming Standards — Taxonomy, nomenclature, ontology
• Data Footprint — The commissioning instrument
• Knowledge Schema — Schemas influence attention
• First Principles — Nomenclature is the first principle

Links

• IAB Tech Lab — OpenRTB — The industry's data standard for real-time bidding
• ads.txt — Authorized digital sellers standard
• VAST Standard — Video ad serving template
• Antonio García Martínez — Chaos Monkeys — Inside Facebook's ad machine

Questions

If whoever names the entities owns the ontology, and whoever owns the ontology controls value flow — what happens when users rename "tracking" to "my data"?

• Which row in the data footprint table would you fix first — and does your answer reveal whether you're an advertiser, publisher, or platform?
• When Attribution has "None" for feedback loop, what compounds instead of learning?
• If Settlement moved from 30-90 days to sub-second, which intermediaries exist only because of the delay?
• What would the naming system look like if users designed it instead of platforms?