Stella Case Study: Scaling Google Ads Profitably by Reframing ROAS Targets

How a $17M ecommerce brand uncovered the underreported impact of its most important channel through incrementality testing

‍

Overview

‍
A national athletic apparel brand partnered with Stella in July 2024 to evaluate the true business impact of its Google Ads investment. Despite consistently strong platform-reported performance (6.0 ROAS), leadership suspected the actual contribution to revenue was higher. Stella’s incrementality testing confirmed this hypothesis, revealing that Google Ads was generating 60% more incremental revenue than reported. This allowed the brand to adjust ROAS targets and scale spend confidently.

‍

Objective

‍
Quantify the incremental return on ad spend (iROAS) from Google Ads using causal inference and apply that insight to guide budget and bidding strategy.

‍

Ad Strategy Breakdown (Before Testing)

The brand’s Google Ads budget was thoughtfully diversified across multiple campaign types and conversion stages:

• Performance Max (~60% of budget)
  • Optimized specifically for new customer acquisition (a setting available within Google Ads)
  • Majority of delivery occurred through Product Listing Ads (PLAs)
  • Designed to efficiently capture high-intent shoppers while feeding the funnel with new-to-file customers
• YouTube (~30%)
  • Focused on brand awareness and retargeting, using strong creative in immersive, high-attention environments
  • These campaigns were not optimized for clicks, but rather for broader influence on consideration and intent
• Non-Branded Search (~8%)
  • Targeted mid-funnel discovery behaviors—users searching category-level or competitor terms
• Branded Search (~2%)
  • Captured users searching directly for the brand (e.g., “[Brand Name] leggings”)
  • Spend was minimal due to strong organic rankings and high brand familiarity

‍

‍

The Incrementality Study

Implementing Stella's Incrementality Tool

Experiment Design

• Methodology: Geo-based causal impact modeling
• Pre-period: 120 days of historical sales by region
• Test period: August to July 2024 (45 days)
• Primary metric: Total revenue by region
• Structure: Test locations were selected from Stella's location selection tool and control regions were built by synthetic weighted controls

Stella’s model constructed a synthetic control group that estimated what would have happened in test regions without Google Ads. This baseline was then compared to actual results to calculate lift.

‍

*The chart below is from Stella's old UI. Our current 2025 UI has evolved a lot since this screenshot was taken making it much easier to interpret the key findings.

‍

Model Results

• Platform ROAS: 6.0
• Incremental ROAS (iROAS): 10.11
• Incrementality Factor: 10 ÷ 6 = 1.67
• Statistical Significance: p < 0.01
• Model Accuracy:
  • MAPE = 8.9%
  • R² = 0.82

This meant that for every $1 Google Ads claimed in reported revenue, the true incremental lift was $1.67, a 67% increase. Model diagnostics confirmed a strong fit and high reliability.
‍

Strategic Application

With a validated incrementality factor, the brand adjusted its approach to budgeting and target-setting:

• New ROAS Target
  6.0 (iROAS goal) ÷ 1.67 = ~3.6 platform ROAS
  This gave the brand confidence to scale spend without sacrificing profitability
• Execution Plan
  • Increased Google Ads budget by approximately 20% weekly
  • Gradually lowered ROAS targets by approximately 10% each week
  • Transitioned primary success metric from platform ROAS to MER (Marketing Efficiency Ratio)

‍

Why Incrementality Was Likely Underreported

Two key factors in the brand’s setup suggested Google’s actual revenue contribution was being under-credited:

1. YouTube’s High Incrementality
   YouTube campaigns, especially those not optimized for conversions, often influence purchase behavior without being directly credited in platform attribution. These placements contribute to incremental lift that is invisible to last-click or platform models.
2. New Customer Optimization
   By using Google’s new customer acquisition goal within Performance Max, the brand trained the algorithm to find buyers who were unlikely to convert otherwise. This naturally raised the channel’s true incremental value beyond what platform-reported metrics could reflect.

‍

‍

Strategy Implementation

With a validated incrementality factor, the brand adjusted its approach to budgeting and target-setting:

• New ROAS Target
  6.0 (iROAS goal) ÷ 1.67 = ~3.6 platform ROAS
  This gave the brand confidence to scale spend without sacrificing profitability
• Execution Plan
  • Increased Google Ads budget by approximately 20% weekly
  • Gradually lowered ROAS targets by approximately 10% each week
  • Transitioned primary success metric from platform ROAS to MER (Marketing Efficiency Ratio)

‍

‍

Results

An analysis of the data from June to October 2024 demonstrates the impact of the implemented strategy (Study finished August 17th):

Observed Results (June to October 2024)

• Weekly Google Spend:
  Increased from $18,702 to $29,553
• Weekly Total Revenue:
  Grew from $341,129 to $497,114
• MER (Total Revenue ÷ Total Ad Spend):
  Improved from 3.00 to 4.45, a 48% increase
• Google-reported conversion value:
  Remained mostly flat, confirming attribution blind spots

Even as in-platform ROAS declined (as expected when scaling), MER increased steadily, signaling real business growth driven by expanded spend.

‍

Key Takeaways

• ROAS is not the ceiling
  Platform-reported ROAS significantly underrepresented actual value
• YouTube and new customer tactics drive high iROAS
  The channel mix helped explain the 1.67x lift
• MER is a better north star
  As the brand scaled spend, MER tracked overall efficiency more reliably than ROAS

‍

Implications for Other Brands

This study illustrates how incrementality testing can unlock profitable scale. Many brands unintentionally constrain growth by holding channels to platform-reported ROAS targets that ignore invisible impact. Stella helps reveal the full picture.

With Stella, brands can:

• Calculate the true iROAS of each channel
• Adjust spend targets based on real business outcomes
• Use MER to guide holistic marketing decisions at scale

‍

Learn more about Stella's incrementality tool here.

‍

‍
Raw Data:

‍