Portfolio ROAS Optimization Using Real Google Ads Data

Overview

Case 3 introduces one of the most practical and revenue‑aligned optimization models in the SkyForge system: maximize total revenue while enforcing a minimum ROAS threshold and staying within a fixed budget.

Unlike Case 1 (unconstrained) and Case 2 (minimum spend stability), Case 3 adds a profitability guardrail. This prevents Solver from shifting budget into low‑value campaigns simply because they are cheap.

Using your real Google Ads dataset (multi‑industry, multi‑country, multi‑channel), Case 3 demonstrates how a portfolio‑level ROAS floor transforms optimization quality.

Why Case 3 Matters

Most accounts don’t fail because of a lack of spend. They fail because spend is misallocated.

Case 3 solves this by enforcing:

A budget cap
A minimum ROAS floor
A revenue‑maximizing objective

This produces a model that is:

Scalable
CFO‑friendly
Industry‑agnostic
Resistant to noise
Perfect for forecasting and scenario planning

The Optimization Model

Objective

Maximize total revenue across all campaign groups.

Mathematically:

Maximize \sum (S p e n d_{i} \cdot R O A S_{i})

Where:

Spendᵢ = optimized spend for group i
ROASᵢ = historical ROAS for that group

Constraints

1. Budget Cap

\sum S p e n d_{i} \leq 1,682,410.45

2. ROAS Floor

Portfolio ROAS must stay above a minimum threshold (e.g., 3.0):

\frac{\sum (S p e n d_{i} \cdot R O A S_{i})}{\sum S p e n d_{i}} \geq R O A S_{f l o o r}

Rearranged for Solver:

\sum (S p e n d_{i} \cdot R O A S_{i}) \geq R O A S_{f l o o r} \cdot \sum S p e n d_{i}

3. Non‑Negativity

S p e n d_{i} \geq 0

4. Optional: Max Spend per Group

Useful for preventing Solver from over‑allocating into a single high‑ROAS cluster:

S p e n d_{i} \leq H i s t S p e n d_{i} \cdot M u l t i p l i e r

Data Engineering: Reducing 700+ Rows to <200 Variables

Your dataset contains hundreds of rows across:

Search
Shopping
Display
Video
10+ industries
10+ countries

Solver cannot handle 700+ variables directly.

So Case 3 uses grouped optimization.

Grouping Keys

campaign_type
industry
country

This reduces the model to 40–60 variables, depending on unique combinations.

For each group, we compute:

Total historical spend
Weighted ROAS
Weighted revenue
Weighted conversions

These become the optimization units.

Insights From Your Real Data

Your dataset shows massive ROAS variance, which is exactly why Case 3 is powerful:

Some Shopping + SaaS + USA groups deliver 7–11 ROAS
Some Display + Healthcare + India groups deliver 0.6–1.2 ROAS
Search + SaaS + UAE hits 8.48 ROAS
Video + EdTech + Australia hits 11.27 ROAS
Many Display clusters underperform (<1.5 ROAS)

Case 3 forces Solver to prioritize high‑ROAS clusters while still respecting the budget and ROAS floor.

This is where the model becomes transformative.

What Case 3 Produces

1. A Revenue‑Maximizing Spend Plan

Budget flows into the highest‑ROAS groups across all channels and countries.

2. A Profitability‑Protected Portfolio

Low‑ROAS clusters are automatically suppressed unless needed to meet the budget.

3. A CFO‑Ready Scenario Model

You can instantly answer:

“What if we raise the ROAS floor to 4.0?”
“What if we cut the budget by 20%?”
“What if we double the budget?”
“What if we isolate only Search + Shopping?”

4. A Scalable Optimization Engine

Case 3 becomes the backbone for:

Monthly budget planning
Multi‑market forecasting
Channel mix modeling
Portfolio‑level performance guarantees

Why Case 3 Outperforms Traditional Optimization

Traditional Approach

Optimize each campaign individually
No portfolio view
No ROAS guardrails
No cross‑channel tradeoffs

Case 3 Approach

Portfolio‑level optimization
ROAS‑protected
Revenue‑maximizing
Cross‑channel, cross‑industry, cross‑country
Mathematically defensible

This is the model executives trust.

Conclusion

Case 3 is the first real‑world‑ready optimization case in the SkyForge 40‑Case System.

It balances:

Revenue growth
Profitability
Budget discipline
Portfolio efficiency

Using your real dataset, Case 3 reveals exactly where spend should flow to maximize revenue while maintaining a healthy ROAS.

This is the model you can deploy monthly, quarterly, or across all markets to drive predictable, scalable performance.