Part BOutcome & teaching note

Platform Economics · 1998–2024

Google search advertising

Google launched AdWords Select with CPC pricing and ranked sponsored links using both bid and click-through performance while retaining CPM-priced Premium ads.

Part A decision time: February 20, 2002, 9:00 a.m. PST
Part A knowledge cutoff: February 19, 2002, 11:59:59 p.m. PST
Outcome horizon: February 2002 through fiscal 2024
Outcome classification: Success, with material causal and governance qualifications

This note reveals evidence unavailable to the Part A decision maker. Structured ledgers, reported facts and hash-bound excerpts are authoritative over this narrative.

Outcome in one sentence

Google launched CPC AdWords Select and ranked sponsored links using both bid and click-through performance; advertising later became enormous, but the evidence supports only that the design was a consequential enabling mechanism—not that Google invented CPC or that Select alone caused later revenue, income, enterprise value or market capitalization. source · google.outcome.select-launch source · google.outcome.register-launch

What Google actually did

The February 20, 2002 announcement introduced CPC pricing while retaining CPM pricing for Premium ads. Rank combined click-through performance and the advertiser's bid rather than automatically awarding first place to the highest bid. Contemporary independent coverage described this as Google's twist on a market then dominated by Overture. source · google.outcome.select-launch source · google.outcome.register-launch

This was adaptation and differentiation, not invention. Overture's filing says its predecessor service launched in June 1998; the Part A GoTo filing documents bid-ordered, pay-per-click listings. The important design move was pairing CPC measurability with a response-based relevance input and Google's existing paid-organic separation. source · overture.outcome.2001-10k

Google's launch assertion that organic results were automated and unaffected by payment remains an attributed company claim. The 2004 registration statement repeated that organic results were not sold and advertisements were labeled. Those statements establish the declared product principle, not an independent audit of every later ranking and product decision. source · google.outcome.select-launch source · google.outcome.2004-s1

Early commercial outcome

Google later said revenue growth was driven in part by the early-2002 CPC model and advertiser growth, and to a lesser extent by AdSense for search. Its 2004 Form 10-K also said AdWords became exclusively CPC in first-quarter 2002 and moved to one CPC structure for advertisers in 2004. These are useful chronology and issuer attribution, not a controlled causal estimate. source · google.outcome.2004-s1 source · google.outcome.2004-10k

USD millionsFY 2002FY 2003FY 2004
Google websites advertising306.978792.0631,589.032
Google Network advertising103.937628.6001,554.256
Total advertising410.9151,420.6633,143.288
Consolidated revenue439.5081,465.9343,189.223
Traffic acquisition costs94.5526.51,228.7

Lineage: table.google.outcome.early-ad-economics. Advertising supplied most of the displayed revenue, while TAC grew rapidly with partner distribution. A top-line narrative that omits TAC would miss a core cost of the scale mechanism. source · google.outcome.2004-10k

The initial S-1 used an earlier net-revenue presentation in parts of its historical discussion. This note does not splice those net figures into the later gross-revenue series; the early table above comes from the 2004 Form 10-K's stated presentation.

Long-horizon reported scale

USD millionsFY 2002FY 2004FY 2010FY 2015FY 2020FY 2024
Consolidated revenue439.5083,189.22329,32174,989182,527350,018

Lineage: table.google.outcome.reported-scale.

USD millionsFY 2010FY 2015FY 2020FY 2024
Consolidated net income8,50516,34840,269100,118

Lineage: table.google.outcome.net-income. These values establish later reported scale. They are not a deterministic value-created bridge for Select and should not be converted into a causal percentage. source · google.outcome.2010-10k source · google.outcome.2015-10k source · google.outcome.2020-10k source · google.outcome.2024-10k

The advertising mix itself changed:

USD millionsFY 2010FY 2020FY 2024
Google websites or Search & other19,444104,062198,084
YouTube adsNot separately disclosed19,77236,147
Google Network8,79223,09030,359
Total Google advertising28,236146,924264,590

Lineage: table.google.outcome.ad-mix. The first row changes from Google websites to Search & other, and YouTube appears separately only later. Product-line values are retained under their reported labels rather than treated as one perfectly comparable series. [claim.google.outcome.advertising-scale; claim.google.outcome.metric-definition-changes]

Metric and segment definitions changed

Google reported aggregate paid-click growth of about 16 percent and average-CPC growth of 5 percent from 2009 to 2010. Alphabet reported 22 percent paid-click growth and an 11 percent CPC decline from 2014 to 2015, but its 2015 definition included viewed YouTube engagement ads. In 2020 Alphabet removed those ads and related revenue from both current and historical paid-click and CPC metrics presented. source · google.outcome.2010-10k source · google.outcome.2015-10k source · google.outcome.2020-10k

Segment definitions also changed. In 2015 Alphabet reported one Google segment and combined Other Bets, retrospectively adjusting prior periods. Beginning in fourth-quarter 2020, it reported Google Services, Google Cloud and Other Bets. The segment snapshot table preserves these structures and marks 2015 Cloud as unavailable rather than manufacturing comparability. source · google.outcome.2015-10k source · google.outcome.2020-10k

Causal assessment

Primary hypothesis: CPC plus relevance was an enabling mechanism

CPC linked advertiser spending to measurable visits. Click-response weighting gave useful ads a path to outrank low-response high bids. Google's 2004 filing attributed early growth in part to CPC and advertiser expansion. This chronology and mechanism support a moderate-confidence conclusion that Select mattered. [hypothesis.google.quality-cpc-enabler; judgment.google.outcome.enabling-mechanism]

They do not identify an incremental cash-flow series or control group. No percentage of the displayed financial outcome is assigned to Select. assumption.google.outcome.partial-attribution and assumption.google.outcome.no-value-attribution preserve that boundary.

Rival 1: organic quality and distribution created the demand

Google's organic-search usefulness plausibly attracted users, while Yahoo, EarthLink and later paid defaults expanded distribution. TAC was USD 1.2287 billion by 2004; Alphabet reported TAC rates of 22.3 percent in 2020 and 20.7 percent in 2024. In 2024 it said mobile search carried higher TAC because more searches flowed through paid access points. The auction monetized traffic whose creation and acquisition had separate causes and costs. [hypothesis.google.search-quality-distribution; claim.google.outcome.distribution-co-cause; claim.google.outcome.tac-material]

Rival 2: later products and secular markets dominate

AdSense extended advertisements to partner inventory. DoubleClick, display and YouTube broadened formats and properties. Mobile search, Android and Chrome changed access and usage. Alphabet later added separately reported Cloud economics and, by 2024, an AI-altered search experience. Its 2015 filing attributed website growth to mobile-search ad improvements and YouTube, partly offset by a stronger U.S. dollar. These are material co-causes over a 22-year horizon. source · google.outcome.2010-10k source · google.outcome.2015-10k source · google.outcome.2024-10k

Rival 3: advertiser liquidity and network effects

More advertisers can deepen keyword coverage and auction liquidity; better measurement can support repeat budgets; more partner inventory can create additional data and demand. Overture's approximately 53,000 paying advertisers and affiliate dependence show that this mechanism was broader than Google. Its exact contribution remains unmeasured and confidence is low. source · overture.outcome.2001-10k

Process quality

The launch preserved a Premium CPM fallback and introduced the new pricing and ranking logic as a distinct product. That is directionally consistent with the Part A recommendation to buy information without abandoning organic-paid separation. It does not prove the rollout was actually capped, that holdouts were stable or that fraud, pricing, partner and rollback gates existed. Public sources do not reveal the internal process. [judgment.google.outcome.process-quality]

Regulatory and legal boundaries

The later record must be typed and scoped precisely:

  • In 2017 the European Commission found that Google gave its own comparison-shopping service prominent placement while demoting rivals. The cited practice began in Europe in 2008. It is a later self-preferencing finding, not a ruling on the 2002 CPC auction. source · google.outcome.eu-shopping
  • The 2020 DOJ complaint alleged unlawful monopoly maintenance in general search, search advertising and general search text advertising. At filing, those were allegations. source · google.outcome.doj-complaint
  • In August 2024 the district court held Google liable for maintaining monopolies in general search services and general search text advertising through exclusive distribution agreements. The same opinion said Google lacked monopoly power in a broader search-advertising product market and was not liable for SA360 conduct. The holding concerns later distribution agreements; it does not declare AdWords Select's 2002 auction illegal. source · google.outcome.court-opinion

The teaching lesson is not that quality-aware monetization was mistaken. It is that product neutrality, user choice, partner concentration and contract governance must remain live controls as a platform becomes powerful.

Counterfactuals

Three cutoff-feasible alternatives remain unpriced:

  1. Keep CPM AdWords and Premium sponsorships while testing subscriptions or waiting for stronger economics.
  2. Copy Overture's pure highest-bid CPC model.
  3. Launch CPC on google.com first and add partner inventory only after quality and contribution gates mature.

No cutoff-valid cohort panel, partner contract, implementation budget or causal model quantifies their effects. Delaying might have preserved capital and trust or sacrificed learning and advertiser adoption. Pure bid ordering might have simplified the auction or degraded relevance. A slower partner rollout might have reduced concentration or forfeited inventory. assumption.google.outcome.counterfactual-uncertainty is deliberately unquantified.

Decision rules

rule.quality-aware-auction-rollout is a candidate rule: when an outcome-linked auction can reward low quality or gaming, combine the economic bid with a verified quality signal and scale only after user, customer, abuse, fairness and contribution gates clear.

rule.distribution-economics-and-governance is also candidate: reconcile partner contribution and concentration, preserve user choice and neutrality, and require independent legal review before material exclusive or default-expanding commitments.

Both remain candidates because this module is one historical success case, not a validated cross-case rule.

Abstentions and hindsight limits

This module does not establish:

  • the internal 2002 relevance, fraud, latency, budget or rollback gates;
  • Select-specific revenue, cost, capital or cash flow;
  • a stable paid-click, CPC, product or segment series across all horizons;
  • the causal share of any displayed revenue, income, enterprise value or market capitalization;
  • the financial outcome of CPM, highest-bid or slower-rollout alternatives;
  • that issuer trust claims audit all later conduct; or
  • that later legal findings were knowable at the cutoff or adjudicated the 2002 auction.

No valuation, target price, material accounting adjustment or externally approved recommendation is presented. Human publication approval remains absent.

Source map

Every displayed number resolves to a reported fact and exact excerpt. Interpretations remain judgments, assumptions, conflicts, causal hypotheses and candidate rules.

Observed after the cutoff

Outcome financials

7 tables

Later values do not backfill Part A. Definition changes, unknowns, and derived endpoints remain labeled.

Selected reported consolidated revenue scaleAs Reported At Horizon · USDm
MeasureFY 2002FY 2004FY 2010FY 2015FY 2020FY 2024
Consolidated revenue439.50813,189.223129,321174,9891182,5271350,0181
USD · USDmReported values remain strings; no browser-side recalculation.
Early reported advertising revenue and traffic acquisition costAs Reported At Horizon · USDm
MeasureFY 2002FY 2003FY 2004
Google websites advertising revenue306.9781792.06311,589.0321
Google Network advertising revenue103.9371628.611,554.2561
Total advertising revenue410.91511,420.66313,143.2881
Consolidated revenue439.50811,465.93413,189.2231
Traffic acquisition costs94.51526.511,228.71
USD · USDmReported values remain strings; no browser-side recalculation.
Selected advertising revenue mix under period-specific definitionsAs Reported At Horizon · USDm
MeasureFY 2010FY 2020FY 2024
Google websites or Search and other19,4441104,0621198,0841
YouTube adsNot established19,772136,1471
Google Network8,792123,090130,3591
Total Google advertising28,2361146,9241264,5901
USD · USDmReported values remain strings; no browser-side recalculation.
Selected consolidated net incomeAs Reported At Horizon · USDm
MeasureFY 2010FY 2015FY 2020FY 2024
Net income8,505116,348140,2691100,1181
USD · USDmReported values remain strings; no browser-side recalculation.
Issuer-defined paid-click and average-CPC changesAs Reported At Horizon · percent
Measure2009 to 20102014 to 2015
Aggregate paid clicks change161221
Average cost-per-click change51-111
percentReported values remain strings; no browser-side recalculation.
Segment revenue snapshots under changing structuresAs Reported At Horizon · USDm
MeasureFY 2015 structureFY 2020 structureFY 2024 structure
Google segment or Google Services74,5411168,6351304,9301
Google CloudNot established13,059143,2291
Other Bets448165711,6481
USD · USDmReported values remain strings; no browser-side recalculation.
Selected issuer-reported traffic acquisition cost rateAs Reported At Horizon · percent
MeasureFY 2020FY 2024
TAC rate22.3120.71
percentReported values remain strings; no browser-side recalculation.

Transferable—but not universal

Candidate decision rules

2 hypotheses

These rules are case-derived hypotheses. Each retains “unless” conditions, kill criteria, counterexamples, and promotion gaps.

Candidatemoderate confidence

rule.quality-aware-auction-rollout

Roll out the auction in bounded cohorts; combine economic bids with a verified quality signal and expand only after user, customer, abuse, pricing and normalized-contribution gates clear.

Outcome-linked pricing improves measurability, while quality weighting makes low-response high bids less able to dominate; staged traffic and holdouts reveal adverse selection, gaming and trust effects before broad commitment.

Use when

  • A marketplace can move from impression pricing to outcome-linked pricing, but the outcome signal can be gamed or can reward low-quality supply.
  • User trust depends on relevance and clear separation between commercial and non-commercial results.
  • A capped treatment, stable holdout and reversible legacy product can expose real behavior before full migration.

Do not transfer when

  • The measured outcome is not meaningfully connected to customer value.
  • The quality signal can be manipulated more cheaply than the harm it creates.
  • Regulation or safety requirements prohibit live experimentation.
  • The legacy fallback cannot be retained and the migration is economically irreversible.

Reverse or kill if

  • User quality, trust or latency breaches the control threshold.
  • Mature customer cohorts fail to retain or achieve verified outcome value.
  • Invalid events, refunds or support burden exceed the approved envelope.
  • Quality weighting is manipulable or produces systematic low-quality winners.
  • Normalized contribution remains negative without an approved and evidenced path to improvement.
Limitations and promotion gaps
  • Candidate status reflects one historical episode and no completed controlled counterexample.
  • Public evidence does not reveal Google's internal launch thresholds or validate an optimal quality-weight formula.
  • The rule may not transfer to safety-critical or regulated allocations where experimental harm is unacceptable.
  • No expected monetary range is supported.
Candidatemoderate confidence

rule.distribution-economics-and-governance

Reconcile partner-level contribution and concentration, cap dependency, preserve user choice and product neutrality, and require legal and governance review before exclusive or default-expanding commitments.

Distribution can compound demand but creates recurring acquisition cost and control over access; economic and governance gates prevent apparent top-line scale from hiding weak contribution or exclusionary dependence.

Use when

  • A platform depends on defaults, affiliates, channel partners or paid access points for a material share of demand.
  • Partner payments can accelerate scale while obscuring contribution and increasing bargaining or foreclosure risk.
  • The platform also controls ranking, access or monetization rules affecting downstream participants.

Do not transfer when

  • Distribution is immaterial and independently verifiable as non-exclusive and user-selected.
  • Regulation dictates the access arrangement and leaves no discretionary ranking or exclusivity decision.
  • The contract is a short, capped experiment with no renewal, default or foreclosure effect.

Reverse or kill if

  • Partner contribution stays negative after normalized serving and support cost.
  • Query or revenue dependency breaches the approved concentration limit.
  • Default or exclusive terms materially impair user choice or rival access.
  • The arrangement cannot pass independent competition and product-neutrality review.
  • Contract economics or traffic cannot be reconciled to source records.
Limitations and promotion gaps
  • Candidate status reflects one case and should not be treated as corpus-validated competition policy.
  • The public episode lacks the actual Yahoo, EarthLink and later default contracts.
  • Economic concentration thresholds must be specified for each market and cannot be inferred from this case.
  • Legal findings vary by jurisdiction, market definition, conduct and procedural posture.

Lineage

Complete case source ledger

18 records

This list combines decision-cutoff and outcome evidence. Each report citation resolves to a source ID below. Third-party documents remain with their original publishers.

T3

src.google.cutoff.search-paper

The anatomy of a large-scale hypertextual Web search engine

Stanford University · Apr 18, 1998

Academic ResearchSecondaryContemporaneous

Used for: Contemporaneous description of Google's search-quality architecture · Evidence that organic relevance was foundational before paid-search scaling

T2

src.google.cutoff.original-adwords

Google Launches Self-Service Advertising Program

Google Inc. · Oct 24, 2000

Issuer DisclosurePrimaryContemporaneous

Used for: Original AdWords pricing and placement design · Cutoff advertiser-adoption and search-volume claims

T1

src.overture.cutoff.2000-10k

GoTo.com, Inc. Form 10-K for year ended December 31, 2000

U.S. Securities and Exchange Commission · Mar 16, 2001

Regulatory FilingPrimaryContemporaneous

Used for: Prior-art evidence for cost-per-click paid search · Competitor financial and partner-concentration reference class

T3

src.google.cutoff.register-premium

Google mulls premium subscription services

The Register · Oct 26, 2001

Reputable NewsSecondaryContemporaneous

Used for: Independent contemporaneous monetization context · Evidence of subscription and paid-placement alternatives

T3

src.google.outcome.register-launch

Google gives its twist on pay-per-click

ComputerWire, published by The Register · Feb 21, 2002

Reputable NewsSecondaryContemporaneous

Used for: Independent launch interpretation · Competitive context

T1

src.google.outcome.court-opinion

United States et al. v. Google LLC memorandum opinion

U.S. District Court for the District of Columbia, mirrored by CourtListener RECAP · Aug 6, 2024

Court Or Government RecordPrimaryContemporaneous

Used for: Adjudicated antitrust outcome · Distribution-economics boundary condition