Portfolio Project · Marketing Measurement · Search Incrementality

Search Incrementality Control Tower

Paid search can look highly profitable in platform reports while quietly cannibalizing traffic that organic SEO, branded demand, or AI-search visibility would have captured anyway. This prototype evaluates search investment by query cluster and recommends whether to scale, reduce, defend, or test paid search spend.

Paid Search Organic SEO GEO Visibility Competitor Pressure Incrementality Recommendation

Business problem

Attribution can over-credit paid search

Paid search attribution can overstate value when organic SEO and AI-search visibility already capture demand.

My role

Measurement framework and dashboard UX

I designed the scoring, built the dashboard, and wrote the test plan a marketing lead could hand to an analyst.

Output

Budget decisions by query cluster

A decision system that classifies search investments into scale, reduce, defend, test, or SEO/GEO-led actions.

Decision readout

Executive Recommendation

Synthetic prototype

Reallocate budget away from low-risk branded paid search where organic SEO and GEO visibility are already strong. Maintain defensive coverage where competitor pressure is high. Scale high-intent non-brand clusters where organic rankings and AI visibility are weak. Use holdout testing before making large changes in mixed-signal clusters.

Investment Area	Current	Recommended	Action
Brand paid search	35%	20%	Reduce
Non-brand paid search	30%	38%	Scale
Competitor defense	10%	12%	Defend
SEO content	15%	20%	Increase
GEO / AI-search optimization	10%	10%	Measure

Brand exact queries show high organic recapture risk and low competitor pressure.
Non-brand commercial clusters show stronger paid incrementality where organic rank is weak.
Competitor comparison terms justify defensive paid coverage even with moderate organic strength.

Interactive dashboard

Evaluate query clusters by incrementality signal

Static synthetic data. Controls update the prototype readout.

Selected cluster

Brand exact

22 / 100 Reduce

Organic rank and GEO visibility are already strong, competitor pressure is low, and the query has high navigational brand intent. Paid search is likely receiving credit for demand that would be captured organically.

Next action

Run a branded paid search holdout or budget step-down test before making a permanent reduction.

Query clusters and recommended paid search actions
Query Cluster	Query type						Recommendation

Overlap matrix

Where paid search is likely helping

Budget simulator

Prototype assumptions

Organic recapture assumption 60% Competitor pressure sensitivity GEO visibility weight 15% Risk tolerance

Under a balanced scenario, the model recommends moving budget out of brand paid search and into non-brand paid search plus SEO content.

Method

How the measurement logic works

I segmented search demand into branded, product, non-brand, competitor, and informational query clusters. For each cluster, I compared paid performance against organic rank strength, organic recapture risk, competitor ad pressure, and GEO / AI-search visibility. The dashboard assigns an incrementality score and recommends whether to scale, reduce, defend, test, or let SEO/GEO lead.

Strong organic + strong GEO + low competitor pressure
-> Reduce paid search or run a holdout test

Weak organic + weak GEO + high commercial intent
-> Scale paid search while building SEO/GEO coverage

Strong competitor pressure on brand/product terms
-> Defend, but test minimum efficient budget

Mixed paid/organic/GEO signals
-> Run incrementality test before changing budget

Incrementality Score =
  commercial intent
  + competitor pressure
  + organic weakness
  + GEO weakness
  + test evidence
  - brand navigational risk
  - organic recapture risk

Validation plan

Recommended incrementality tests

Test 1

Branded paid search holdout

Purpose: Estimate how much organic search recaptures when branded paid search is reduced or paused.

Measure: Paid clicks, organic clicks, total conversions, competitor impression share, branded volume, and revenue or pipeline impact.

Decision rule: If organic search recaptures most lost paid clicks and total conversions remain stable, reduce brand paid search.

Test 2

Non-brand geo or market-level test

Purpose: Measure whether non-brand paid search increases total conversions, not just paid clicks.

Measure: Spend change, paid traffic lift, organic movement, total search conversions, revenue lift, and incremental cost.

Decision rule: Scale non-brand paid search only where total conversions rise enough to justify incremental cost.

Test 3

SEO/GEO content investment test

Purpose: Determine whether improving organic and AI-search visibility can reduce paid dependence over time.

Measure: Rank improvement, organic clicks, AI answer mention rate, site citation rate, competitor share, and assisted conversions.

Decision rule: Shift long-term investment toward SEO/GEO where visibility gains reduce paid dependence without hurting total conversions.

Program artifacts

What makes this a measurement program

Measurement brief

States the question, who owns the budget decision, what success looks like, and how confident the answer has to be.

Business question
Stakeholders
Decision needed
Metrics and data sources
Risks and timeline

Data requirements

Specifies the minimum dataset needed to evaluate paid, organic, GEO, and conversion outcomes by query cluster.

Google Ads or equivalent paid search platform
Google Search Console
GA4 or web analytics
CRM or ecommerce revenue data
Rank tracking, GEO checks, and auction insights

Executive readout

Converts measurement findings into a budget recommendation with confidence level, risk, and next test.

Recommended budget movement
Expected business impact
Confidence level
Key evidence
Next test and decision required

90-day rollout

Audit search overlap, baseline incrementality risk, run tests, and deliver budget recommendations.

Phase	Timeline	Deliverable
Audit	Weeks 1-2	Query taxonomy, overlap, GEO baseline
Data readiness	Weeks 2-4	Paid, organic, analytics, CRM, and GEO datasets
Test design	Weeks 4-5	Branded holdout and non-brand market test plan
Execution	Weeks 6-9	Controlled spend changes and monitoring
Decision	Weeks 11-12	Budget recommendation and executive readout

Implementation notes

Prototype, synthetic numbers

This is a portfolio prototype using synthetic data. The numbers are invented to show the logic. In production it would run on real spend, Search Console, rank tracking and revenue data.

GEO / AI-search visibility is treated as a leading indicator, not direct proof of revenue impact.

v0.1.0 · Updated 2026-06-02