Skip to main content

Save Budget: Best Practices to Remove Inefficient Products from Catalog

This article outlines the most effective ways to apply stop-loss logic and product removal

P
Written by Praveen Kumar
Updated over 2 months ago

If you're spending on products that aren't converting — you're not alone. BigAtom’s Save Budget feature helps brands automatically identify and exclude poor-performing products across Meta and Google Ads, using logic based on actual performance metrics like ROAS, conversions, revenue, and more.

This guide explains three proven approaches brands use to set up product-level stop-loss logic to avoid wasted spends — without hurting high performers.

🔎 What This Solves

  1. How to reduce ad spend on low-performing products

  2. How to apply ROAS-based stop loss rules

  3. How to use revenue and performance benchmarks to exclude products

  4. How to prevent removal of top sellers by mistake

  5. How to set channel-specific removal conditions

1. Product Removal Based on Website Revenue

Use this approach if:
You want to stop spending on products that aren't contributing backend revenue, regardless of the ad platform.

Recommended logic:
Let the product spend up to 3× your average Cost per Sale (CPS). If it doesn't deliver at least 40–50% of your website’s average blended ROAS, mark it for exclusion.

Example condition:

  • Total Spend > ₹1,000

  • Blended ROAS < 50% of website average
    ➡️ Exclude from both Meta and Google catalogs

Safety filters to avoid exclusion of top revenue drivers:

  • Do not exclude if revenue > ₹X or

  • Do not exclude if product is in top 30% of revenue contributors

2. Product Removal Based on Channel-Level Performance

Use this approach if:
You want to exclude products only from the platform where they are underperforming (Meta or Google).

Why this matters:
A product may perform poorly on Google but still do well on Meta. Removing it from both would be inefficient.

Recommended logic:
Create separate stop-loss rules for each platform using channel-specific metrics.

Example condition (Meta):

  • Meta Spend > 2× CPS

  • Meta ROAS < 30% of Meta account average
    ➡️ Remove only from Meta catalog

Important:
Ensure your logic is linked to the correct channel (Meta or Google) based on which platform’s metrics you are using.

3. Advanced Logic: Granular Stop Loss Based on Product Segments

Use this approach when your product catalog has varying ROAS expectations based on category, type, or margin.

a. By Category or Brand

Set different ROAS thresholds for each category or brand.
Example:

  • “Footwear” requires ROAS > 2.5

  • “Accessories” requires ROAS > 1.8

This avoids applying one blanket rule across all products.

b. New Arrivals

New products typically take time to perform.


Set a different (lower) ROAS expectation or temporarily exclude them from stop-loss logic.


New arrivals can be defined by category, launch date, or discount attributes.

c. Profitability-Based Rules

If profit margin data is passed into BigAtom, you can fine-tune ROAS expectations:

  • High-margin product: Allow ROAS > 3x

  • Low-margin product: Require ROAS > 5x

This ensures only unprofitable products are excluded.

Common Practices to Follow

To make your stop-loss setup more effective and stable, follow these additional best practices:

✅ Analyze Over a Longer Date Range

Always use a 14-day or 30-day lookback window when applying stop-loss logic.
Short-term data (like 1–3 days) may reflect temporary fluctuations and lead to unnecessary exclusions.

✅ Set the Rule Frequency to Run Daily

Enable stop-loss checks on a daily basis. This ensures products are reviewed and excluded regularly, maintaining catalog hygiene without manual effort.

Did this answer your question?