1. Background

Most PPC accounts allocate budget using intuition, last‑click performance, or platform recommendations. This case study demonstrates a different approach:

Use Linear Programming (LP) to mathematically determine the optimal budget allocation across all campaigns to maximize total conversions.

The dataset includes hundreds of campaigns across Search, Shopping, Display, and Video, spanning industries such as Fintech, SaaS, Healthcare, EdTech, and E‑commerce.

Examples from the dataset:

  • “59886 impressions, 2113 clicks, CPC 1.26, conversions 159” (Fintech Search UAE)

  • “116472 impressions, 4006 clicks, CPC 2.68, conversions 291” (SaaS Shopping USA)

  • “152979 impressions, 4650 clicks, CPC 0.79, conversions 330” (E‑commerce Search UAE)

These rows give us enough structure to build a real optimization model.

2. Objective

Maximize total conversions across all campaigns given a fixed budget.

This is the simplest and purest LP model — no business constraints, no ROAS floors, no geo balancing. Just pure mathematical efficiency.

3. Data Preparation

For each campaign, we compute Conversion Efficiency (CE):

CEi=conversionsiad_spendi

 

Examples from your dataset

Fintech Search UAE:

CE=1592662.38=0.0597

 

E‑commerce Search UAE:

CE=3303673.5=0.0898

 

SaaS Shopping USA:

CE=29110736.08=0.0271

 

These efficiencies become the coefficients in the LP objective function.

4. LP Model Formulation

Decision Variable

xi=budget allocated to campaign i

 

Objective Function

maxiCEixi

 

Constraints

Total budget constraint:

ixiB

 

Non‑negativity:

xi0

 

This is a classic linear optimization problem.

5. LP Solution Behavior

Because LP is linear and unconstrained (beyond budget), the model will:

  • Allocate all budget to the highest CE campaigns

  • Ignore ROAS, CPA, CPC, or industry

  • Ignore risk, volatility, or diminishing returns