Your Performance Max campaign is doing exactly what it’s designed to do: optimise for conversions at scale. So when it starts pushing spend toward the “wrong” products, it can feel confusing, even frustrating. Why is Google doubling down on low-margin items, out-of-stock products, or SKUs you’d never prioritise?
PMax aggressively follows the data you feed it, and if those signals are incomplete, misaligned, or too broad, the algorithm will optimise in ways that don’t match your business goals.
In this article, we’ll break down why PMax campaigns often scale the wrong products, what’s actually happening under the hood, and most importantly, how to regain control without fighting the automation.
The Root Cause: Conversion Bias
At its core, PMax is a machine learning system trained to maximise outcomes based on signals you feed it. By default, those signals are:
- Conversion count
- Conversion value (revenue)
What is Conversion Bias?
Conversion bias is the tendency of PMax to favour products that:
- Convert more frequently
- Have shorter purchase cycles
- Appeal to broader audiences
These products generate more data, faster. And machine learning systems love data density.
Why This Becomes a Problem
Not all conversions are equal.
Products that convert easily are often:
- Lower-priced
- Discount-driven
- High-demand but low-margin
Meanwhile, high-margin or premium products:
- Have longer decision cycles
- Convert less frequently
- Provide weaker signals to the algorithm
So PMax starts doing what it’s designed to do:
Scale what converts fastest, not what makes you the most money.
The Blind Spot: No Margin Awareness
Performance Max does not inherently know:
- Your cost of goods sold (COGS)
- Your profit margins
- Your operational costs (shipping, returns, etc.)
What It Sees vs What You See
Success is defined purely by the numbers it can measure, primarily revenue and conversion volume. If a campaign generates ₹10,000 in revenue, the system treats that as a clear win. However, from a business standpoint, that same ₹10,000 could be far less impressive if it comes with razor-thin margins. For example, revenue generated at a 5% margin may actually contribute very little to overall profitability.
Example Scenario:
- Product A: ₹1,000 price, ₹700 cost → ₹300 margin
- Product B: ₹5,000 price, ₹2,000 cost → ₹3,000 margin
If Product A converts 10x more frequently, PMax will likely prioritise it, even though Product B is far more profitable per sale.
Why Scaling Goes Wrong
a. Feedback Loop Reinforcement
PMax learns from past performance. If a product starts performing well:
- Gets more impressions
- Gains more conversions
- Strengthens its signal
This creates a self-reinforcing loop, pushing the budget further into suboptimal products.
b. Data Imbalance Across SKUs
Catalogs are rarely uniform. You typically have:
- Hero products (high visibility, low margin)
- Niche products (low volume, high margin)
PMax naturally leans toward the former because:
- They produce more data
- They reduce algorithmic uncertainty
c. Short-Term Optimisation
PMax optimises for immediate signals, not long-term value. It doesn’t inherently consider:
- Customer lifetime value (LTV)
- Repeat purchase probability
- Brand-building effects
The Illusion Of Growth
When campaigns scale incorrectly, metrics can look great:
- Revenue ↑
- Conversion volume ↑
- ROAS stable or slightly improving
But underneath:
- Profit margins shrink
- CAC-to-margin ratio worsens
- Inventory pressure increases on low-value SKUs
This creates what can be called a “vanity scale”. Growth that looks impressive but isn’t sustainable.
The Fix: Data Overlays
To correct PMax behaviour, you need to inject business intelligence into the system. This is where data overlays come in.
What Are Data Overlays?
Data overlays are additional data layers you apply to your product feed or campaign structure to guide optimisation beyond basic conversion signals.
Practical Ways to Apply Data Overlays
a. Profit-Based Conversion Values
Instead of passing revenue as conversion value:
Pass profit (or proxy profit)
How:
- Adjust conversion values using margin multipliers
- Use offline conversion imports
- Feed profit data via APIs or data pipelines
Result:
Performance Max starts optimising toward profit, not just revenue
b. Custom Labels in Product Feeds
Segment your catalog using custom labels like:
- High Margin
- Low Margin
- Clearance
- Seasonal
- Strategic Products
Then:
- Create separate asset groups or campaigns
- Allocate budgets intentionally
c. Margin Bucketing Strategy
Group products into tiers:
- Tier 1: High margin, low volume
- Tier 2: Balanced
- Tier 3: Low margin, high volume
Apply different:
- Bidding strategies
- Budget allocations
- ROAS targets
d. Exclude or Downweight Low-Value Products
Not every product deserves paid media exposure.
Options:
- Exclude low-margin SKUs entirely
- Reduce visibility via feed rules
- Separate them into controlled campaigns
e. Use New Customer Value Signals
If available:
- Assign a higher value to new customers
- Focus on acquisition over repeat low-margin purchases
Structural Fixes Beyond Data
a. Break the “All-in-One” Trap
PMax works best when given structured inputs, not chaotic catalogs.
Instead of one campaign:
Split by category, margin, or business objective
b. Control Budget Distribution
Don’t let PMax decide everything:
- Allocate budgets based on profitability tiers
- Force exposure for strategic products
c. Monitor Incrementality, Not Just ROAS
Ask:
- Would these sales have happened anyway?
- Are we just capturing existing demand?
A Smarter Optimisation Mindset
To make PMax work for your business, shift your thinking:
From:
“How many conversions did we get?”
To:
“Which conversions actually matter?”
Key Takeaways
Conversion bias pushes PMax toward fast-selling, often low-margin products
Lack of margin awareness creates a disconnect between platform success and business success. Data overlays are essential to align algorithmic optimisation with profitability.
Final Thoughts
PMax is not broken. It’s doing exactly what it was designed to do. The problem is that most advertisers are feeding it incomplete signals.
If you want better outcomes, don’t fight the algorithm. Educate it.
Need a fresh perspective? Let’s talk.
At 360 OM, we specialise in helping businesses take their marketing efforts to the next level. Our team stays on top of industry trends, uses data-informed decisions to maximise your ROI, and provides full transparency through comprehensive reports.
.png)
