AdTech Platform

What's the decision path to a world-beating advertising platform?

Are your tools working together, or are they expensive silos?

Tight Five Job

Platform is the operating base. It names the assets and standards that let advertising work run without guesswork.

Layer	Job	Good Sign	Bad Sign
Data assets	Collect usable audience and outcome signals.	First-party and verified data can activate campaigns.	Data sits in silos or cannot be trusted.
Execution stack	Buy, serve, and settle media.	DSP, SSP, ad server, and exchange paths are clear.	Tools overlap while handoffs leak budget.
Intelligence	Attribute, optimize, and report.	AI and analytics improve allocation decisions.	Dashboards explain the past but do not change action.
Protocols	Standardize exchange and proof.	OpenRTB, consent, attribution, and verification rules are explicit.	Every platform defines truth differently.

The ABCD Overlay

Layer	Function	Advertising Application
A - AI	Targeting, bidding, creative	Predictive bidding, DCO, audience modeling
B - Blockchain	Verification, settlement	On-chain RTB, transparent auctions, instant payout
C - Crypto	Payment, incentives	Micropayments for attention, token rewards, EAT protocol
D - DePIN	Physical intelligence	375ai sensors, GEODNET location, verified presence data

Core Architecture

DATA LAYER              → EXECUTION LAYER         → INTELLIGENCE LAYER
CDP + DMP + DePIN       → DSP + SSP + On-Chain    → Attribution + AI + Analytics

Ecosystem Diagram

The AdTech Ecosystem Diagram maps how money and data flow between all parties. The entire chain — user visits page, auction runs, ad renders — completes in under 200 milliseconds. Four horizontal layers with arrows between them.

Layer	Players	Role	200ms Phase
Demand	Advertisers, agencies, DSPs	Buy impressions	Bid request
Middle	Ad exchanges, DMPs, identity	Run auctions, data enrichment	Auction + match
Supply	SSPs, ad networks, publishers	Sell inventory	Win notification
Measurement	Analytics, verification, attribution	Close the feedback loop	Post-impression

The Programmatic Supply Chain is the transactional view of this same diagram: advertiser budget flows through DSP, exchange, SSP to publisher — losing 40-60% to intermediaries along the way. The Sui stack collapses the middle layers.

Tool Landscape by JTBD

$100B+ market, 10,000+ tools, increasingly consolidated by AI agents. Organized by job to be done — what progress is the buyer making?

#	JTBD	Incumbents	AI Disruptor	Entry Price
1	Competitor intelligence	AdSpy, BigSpy, SpyFu	Perplexity agents	$39-149/mo
2	Keyword and audience research	SEMrush, Ahrefs, Similarweb	Clay	$99-129/mo
3	Creative analysis and generation	Motion, Smartly, HeyGen	AdCreative.ai	$29-99/mo
4	Campaign management	Google/Meta native, Optmyzr	Perplexity Computer	Free-$249/mo
5	Data integration	Supermetrics, Funnel.io	Agent APIs	$99-299/mo
6	BI and reporting	Looker, Tableau, GA4	Supermetrics AI	Free-$1K+/mo
7	Attribution	AppsFlyer, Northbeam, Triple Whale	MMM models	Free-$1K+/mo

See Players for detailed provider profiles per category.

AI Consolidation

Perplexity Computer ($200/mo, launched Feb 2025) coordinates 19 AI models to replace 5-10 point solutions. One weekend test: replaced $225K/yr in tools (Looker + Smartly + Motion + AdSpy + Supermetrics). 224 micro-optimizations in a single run — hourly scans, auto-adjusting budgets, fatigue detection, cross-campaign coordination.

What It Replaced	Annual Cost	Job
Looker	~$12K+	Governed BI dashboards
Smartly	% of spend	Creative personalization
Motion	~$6K	Creative trend analysis
AdSpy	~$1.8K	Competitor intelligence
Supermetrics	~$1.2K	Data integration
Total replaced	$225K/yr	vs $2.4K/yr

Caveats: production scalability unproven, Meta automation bans risk, compliance gaps. The pattern matters more than the specific tool — general-purpose AI agents subsume point solutions when the JTBD is information processing, not infrastructure.

Stack Archetypes

Stage	Stack Focus	Annual Cost
Startup	Free natives (Google/Meta Ads Manager + GA4)	$0 + ad spend
Mid-market	+ Intelligence layer (SEMrush + Motion + Supermetrics)	$5-50K
Enterprise	+ Unification (CDP + DSP + MMM + verification)	$100K-1M+
AI-native	Agent replaces mid-market stack	$2.4-5K

The Platforms

Data Layer

Platform	Data Type	Best For	Status
CDP	First-party, unified	Retention, personalization	Growing — the moat
DMP	Third-party, aggregate	Prospecting	Declining — cookies dying
DePIN data	Verified, real-world	Ground truth attribution	Emerging — 375ai, GEODNET

Leading CDPs	Strength
Segment	Developer-friendly, events
mParticle	Mobile-first, real-time
Tealium	Enterprise, governance
Bloomreach	E-commerce, search
ActionIQ	Enterprise, analytics

Execution Layer

Platform	Function	Key Players
DSP	Media buying automation	The Trade Desk, Google DV360, Amazon
SSP	Inventory management	Google Ad Manager, Magnite
Ad Server	Creative delivery	Google CM360, Flashtalking
On-Chain Exchange	Transparent RTB + settlement	Alkimi (Sui)

Leading DSPs	Strength
The Trade Desk	Enterprise, CTV, identity
Google DV360	Google ecosystem, scale
Amazon DSP	Retail media, commerce
StackAdapt	Native, programmatic
Simpli.fi	Geofencing, local

Intelligence Layer

Function	Tools	Purpose
Attribution	Rockerbox, Northbeam	Touchpoint credit
MMM	Measured, Recast	Budget allocation
Analytics	GA4, Amplitude	Behavior understanding
Verification	DoubleVerify, IAS	Fraud and brand safety
On-chain audit	Blockchain explorer	Immutable delivery proof

Platform Comparison

Platform	Primary Function	Data Source	Target	Main Goal
DSP	Automates ad buying	Mixed	All audiences	Media buying efficiency
DMP	Aggregates third-party data	Third-party	New prospects	Audience prospecting
CDP	Unifies first-party data	First-party	Known customers	Retention and LTV
SSP	Sells publisher inventory	Publisher	Advertisers	Yield optimization
Ad Server	Delivers and tracks creatives	Campaign data	All	Creative management

AI Capabilities

86% of companies use or plan to implement AI in advertising.

Function	AI Application	Impact
Bidding	Predictive optimization	25-30% CPA reduction
Targeting	Audience modeling, lookalikes	Reach expansion
Creative	Dynamic optimization (DCO)	10-20% CTR lift
Content	Generative ad creation	60% time savings
Attribution	Multi-touch modeling	Better allocation
Fraud	Invalid traffic detection	Budget protection

Identity Infrastructure

Third-party cookies are dying. Identity strategy determines targeting ceiling.

Solution	Type	Scale	Privacy
First-party cookies	Deterministic	Own traffic	Low risk
Hashed emails	Deterministic	Logged-in users	Low risk
Unified ID 2.0	Industry standard	Growing	Medium
Privacy Sandbox	Google-controlled	Massive	Medium
Contextual targeting	No identity	Unlimited	Zero risk
Data clean rooms	Privacy-safe	Partner data	Low risk
zkLogin (Sui)	ZK-proof identity	Web2 onboarding	Zero risk

The Web3 shift: zkLogin allows familiar Web2 logins (Google, Apple) while keeping identity pseudonymous on-chain. ZK proofs verify audience segment membership without exposing personal data.

Channel Tech

Connected TV (CTV)

$34.49B spend projected in 2025 (+23.2% YoY). Transitioning from awareness to performance through shoppable ads.

• CTV-specific DSP integration (The Trade Desk, Amazon)
• ACR (Automatic Content Recognition) for measurement
• Cross-device identity resolution
• Shoppable ad formats
• Incrementality testing

Digital Out-of-Home (DOOH)

Programmatic DOOH projected to exceed $1B in 2025. DePIN sensors transform measurement.

Traditional DOOH	DePIN-Enhanced DOOH
Modelled audience estimates	Verified foot traffic (375ai sensors)
Coarse geofencing	Centimeter precision (GEODNET RTK)
Delayed reporting	Real-time data streams
Self-reported metrics	Cryptographic proof of presence

Social and Commerce

Social commerce projected to reach $8.5T by 2030.

• Platform API integrations (Meta, TikTok, Pinterest)
• Shoppable content formats
• Influencer management (58% use AI for selection)
• In-app purchase tracking

Web3 Disruption Layer

How Sui + DePIN + EAT protocol restructure the advertising stack.

Why Sui for Advertising

Sui enables real-time ad transactions by moving both data and value on-chain with sub-second finality.

Capability	What It Enables
Parallel execution	100k+ TPS — handles high-volume ad auctions
Object-oriented model	Ad entities (campaigns, creatives, budgets) as native objects
Sub-second finality	Real-time settlement replaces 30-90 day payment cycles
Programmable Transaction Blocks	Atomic: verify impression + update budget + split payment in one tx
zkLogin	Web2 onboarding — Google/Apple login, pseudonymous on-chain
ZK Compression	NFT minting costs reduced 5,200x ($0.005/MB vs $1,000/MB)

On-Chain Ad Objects

Traditional Entity	Sui Object Model	Why
User Cookie/ID	Sui Address / Kiosk	Persistent, user-controlled asset container
Ad Campaign	Shared Object	Multiple publishers interact simultaneously (parallel)
Ad Creative	Immutable NFT	Tamper-proof — prevents malvertising
Budget	Coin Object (SUI/USDC)	Actual liquidity, not a database number
Impression Proof	ImpressionProof Object	Cryptographic delivery receipt
DePIN Node	DePINNode Object	Registered capabilities, location, uptime

Alkimi / EAT Protocol

On-chain Real-Time Bidding on Sui. The Exchange Advertising Terminal eliminates middlemen.

Traditional RTB	Alkimi on Sui
Hidden auction mechanics	Public, immutable bid record
30-90 day payment cycles	Sub-second settlement
50-60% of spend reaches publisher	90%+ reaches publisher
Self-reported metrics	Cryptographic audit trail
Reconciliation required	Payment IS the settlement

Physical Intelligence Layer

375ai edge sensors + GEODNET precision location = verified physical context.

375ai SENSORS → Foot traffic, dwell time, presence    ─┐
                                                        ├→ ImpressionProof on Sui
GEODNET RTK   → Centimeter-level location attestation  ─┘
                         ↓
              On-chain verification
                         ↓
              Atomic payment split (PTB)

Component	Data Provided	Fraud Eliminated
375ai	Verified human presence, dwell time, flow patterns	Ghost impressions, bot traffic
GEODNET	Centimeter-level position attestation	Location spoofing
ZK proofs	Audience segment match without identity exposure	Privacy violations
Sui settlement	Instant, atomic, auditable payment	Payment fraud, hidden fees

Data Velocity Model

Not everything goes on-chain. Separate by speed and trust requirement.

Data Type	Location	Rationale
Raw RTB bid/response logs	Off-chain (data warehouse)	High volume, low individual value
Sensor video frames	On-device (375ai edge processing)	Privacy — never leaves device
GEODNET correction streams	Off-chain (real-time service)	High frequency, low value per datum
Winning bids + settlement	On-chain (Sui)	Trust point — drives payment
Impression proofs	On-chain (Sui)	Trust point — verified delivery
Location attestations	On-chain (Sui)	Trust point — anti-spoofing
Revenue splits	On-chain (Sui)	Trust point — instant payout

Privacy-First Targeting

Zero-party data: users own and share voluntarily for compensation.

Traditional	Web3 (Sui)
Covert cookie tracking	User grants read-only capability to advertiser
Data stored on advertiser servers	Data stays in user's wallet (Sui object)
All-or-nothing consent	Granular: allowed verticals, formats, frequency caps
User gets nothing	User earns tokens for verified attention

DePIN Integration

How decentralized data infrastructure plugs into the stack:

DePIN Sensors → Verified Data → CDP Integration → Audience Enhancement → DSP Targeting
                      ↓
              On-chain attestation
              (provenance proof)

Integration Point	DePIN Data	Value Add
CDP enrichment	Location, weather, environment	Hyperlocal audience segments
DSP targeting	Real-time context signals	Moment-based targeting
Creative triggers	Weather, foot traffic, events	Dynamic creative optimization
Measurement	Physical world verification	Offline attribution
Settlement	On-chain delivery proof	Instant, fraud-resistant payment

Measurement Stack

The Measurement Triad

Best-in-class measurement combines three approaches:

1. MMM — Strategic allocation, offline + online, no tracking required
2. MTA — Tactical optimization, customer journey mapping
3. Incrementality — Causal proof, ground truth calibration

Function	Tools
Web Analytics	GA4, Adobe Analytics, Amplitude
Mobile Attribution	AppsFlyer, Adjust, Branch
Multi-touch Attribution	Rockerbox, Northbeam, Triple Whale
Marketing Mix Modeling	Measured, Recast, internal builds
Incrementality Testing	Geo experiments, holdout tests
Ad Verification	DoubleVerify, IAS, MOAT
On-chain Verification	Sui block explorer, DePIN attestations

Marketplace Infrastructure

Three types of marketplace platforms enable programmatic buying and selling.

DSP Providers

Provider	Key Strength	Market Position
The Trade Desk	Omnichannel RTB, Unified ID 2.0, CTV leader	Largest independent
Google DV360	Google ecosystem integration, scale	Duopoly
Amazon DSP	Retail media, purchase intent data	Third force
Adform	Full ad server + DSP, transparency	EU-focused
StackAdapt	Native, programmatic	Mid-market
Simpli.fi	Geofencing, local targeting	SME/local

SSP Providers

Provider	Key Strength	Market Position
Google Ad Manager	Header bidding, direct deals	Dominant
PubMatic	OpenRTB, private marketplaces	Independent
Magnite	Premium video/display, global	Post-merger scale
OpenX	High-volume auctions, transparency	Independent

Ad Exchanges

Neutral marketplaces connecting DSPs and SSPs for RTB auctions.

Exchange	Key Strength
Google AdX	Largest inventory, premium access
Xandr	Data-driven programmatic
Alkimi (Sui)	On-chain transparent RTB, instant settlement

Build vs Buy

Build proprietary intelligence, buy commodity infrastructure.

Component	Recommendation	Web3 Option
DSP	Buy	Alkimi (on-chain exchange)
CDP	Buy or build	—
Attribution	Build + buy	On-chain proofs
Creative tools	Buy	NFT-based creative management
Optimization AI	Build	—
Reporting/BI	Build	On-chain analytics
Identity	Buy + partner	zkLogin (Sui)
Physical data	DePIN	375ai, GEODNET

Stack Evaluation

Metric	What It Shows
Time to activation	Speed from data to campaign
Data freshness	Lag between event and use
Match rates	% users identifiable cross-platform
Integration reliability	Sync uptime and success
Total cost of ownership	Platform + engineering + ops
Settlement speed	Time from impression to publisher payment
Fraud rate	% of spend on verified vs unverified impressions

Stack Priorities

1. Consolidate — Reduce tool count, increase integration depth
2. First-party foundation — CDP and identity infrastructure
3. AI-native — Platforms with embedded ML, not bolted-on
4. Privacy-ready — zkLogin, consent management, cookieless
5. CTV-capable — Streaming inventory access and measurement
6. On-chain settlement — Transparent, instant, auditable
7. Physical intelligence — DePIN sensors for ground-truth measurement

Context

• Tight Five of Platform — Machines, Tools, Software, Property Rights, Laws
• Advertising Overview — The transformation thesis and buyer problem
• Players — Who operates at each layer, tool providers by JTBD
• Performance — KPI decision map and benchmarks
• Protocols — How data flows through the stack
• JTBD Framework — The lens for organizing tools by buyer progress
• AI Data Industry — Data that powers targeting
• DePIN Devices — Sensors that generate ground-truth data
• 375ai — Verified physical presence for advertising
• GEODNET — Precision location attestation
• DePIN Investment Thesis — Evaluation framework
• Advertising SaaS — Product specs
• ABCD Stack — Broader technology framework

Links

• Alkimi Exchange on Sui
• 375ai
• GEODNET
• a16z DePIN Use Cases
• DePIN Scan
• Display Lumascape — The canonical AdTech ecosystem diagram
• Luma Partners Lumascapes — Full ecosystem maps by category
• The Trade Desk — Largest independent DSP
• Smartly.io — AI creative automation at scale
• Supermetrics — Marketing data integration
• Motion — Creative analytics for paid social
• Ghostery — Privacy browser extension exposing tracker landscape
• Wired — Is Facebook Listening? — The surveillance advertising question

Questions

When a $200/mo AI agent replaces $225K/yr in point solutions, what does that reveal about where value lives in the AdTech stack?

• Which layer of the ecosystem diagram captures the most margin — and is that the same layer that creates the most value?
• If the 200ms auction lifecycle is fully automated, what human decisions still matter — and at what timescale do they operate?
• What breaks first when you consolidate 5 tools into 1 agent: the tool's job, or the organizational workflow built around the tool?
• Does on-chain settlement kill the intermediary margin or make it transparent — and which outcome is worse for incumbents?