Bundling Distortions - How Product Bundles Break Your Google Ads Bidding

The Bundle Problem

Bundles are everywhere in ecommerce. "Buy the set and save." "Starter kit - everything you need." "Gift bundle - three products, one price." They increase average order value, reduce per-item fulfilment costs, and help clear slow-moving stock by pairing it with bestsellers.

From a merchandising perspective, they make sense. From a Google Ads bidding perspective, they create a specific and predictable distortion: bundles inflate revenue signals while compressing margin signals.

The higher price point makes bundles look like premium products to Smart Bidding. The algorithm bids accordingly - aggressively - because it sees conversion value, not contribution margin. And unless your feed architecture explicitly accounts for bundle economics, every bid on a bundle is based on incomplete data.

How Bundles Break Margin Tracking

Individual products have relatively clean economics. Product A costs £12 to source, sells for £35, has a known fulfilment weight, and a predictable return rate. You can calculate contribution margin with confidence.

Bundles introduce layers of complexity:

• • Combined COGS ambiguity: The bundle price is typically discounted from the sum of individual prices. But the COGS is the full sum of component costs. The discount comes entirely from margin, not from reduced costs.
• • Fulfilment cost inflation: Bundles are heavier, larger, and often require custom packaging. A three-product bundle might cost £4.50 to fulfil versus £2.00 for a single item. This cost rarely gets attributed back to the bundle SKU.
• • Return complexity: Customers return individual items from bundles, not the whole bundle. If someone returns the least valuable item, your COGS doesn't change but your revenue drops disproportionately. The return rate calculation becomes product-level, not transaction-level.
• • Payment processing on inflated AOV: Payment processing fees are percentage-based. A £80 bundle incurs higher absolute processing fees than three separate £30 purchases - even though the margin might be similar or worse.

The net effect: bundles look more valuable than they are. If your COGS data doesn't account for these factors, your POAS calculation is systematically wrong for every bundle in your catalogue.

The ROAS Distortion

Consider a simple comparison. Product A sells for £35 with £20 contribution margin. A bundle containing Product A plus Product B sells for £55 with £15 contribution margin (because the bundle discount ate the margin on Product B).

At a CPC of £0.80, each click costs the same regardless of which product the customer buys. But the bundle generates £55 in conversion value versus £35 for the individual product. Smart Bidding sees the bundle as 57% more valuable and bids accordingly.

The reality: the bundle is 25% less profitable per sale. The algorithm is optimising in the wrong direction.

This distortion compounds when bundles compete with their own component products in the same Shopping feed. Google may choose to show the bundle instead of the individual product - capturing the click at lower margin. Or it might show both, splitting your impression share and diluting data signals across two listings that serve the same demand.

Smart Bidding Gets Confused

Smart Bidding optimises for conversion value (tROAS) or conversions (tCPA). Neither metric understands product bundles:

• • tROAS sees revenue, not margin. The bundle's £55 revenue looks better than the individual product's £35. Smart Bidding doesn't know the bundle has £5 less profit.
• • tCPA treats all conversions equally. One bundle sale equals one individual product sale. But the cost to acquire, fulfil, and support the bundle customer is higher.
• • Value-based bidding requires accurate values. If you're passing revenue as conversion value, bundles get inflated bids. If you're passing contribution margin, bundles get correct bids - but only if the margin calculation is correct. Which, as we've established, it usually isn't.

The algorithm isn't broken. It's doing exactly what you've told it to do. The problem is that the signal it receives is distorted by bundle economics that nobody has translated into conversion value.

Worked Example: The Invisible Loss

A skincare brand offers individual serums at £28 each and a "Complete Routine" bundle of three serums for £65 (saving £19). Let's trace the economics:

On the surface, the bundle has higher absolute margin (£35.71 vs £18.49). But selling three individual serums generates 3 × £18.49 = £55.47 in total contribution margin. The bundle costs £19.76 in lost margin.

If the bundle cannibalises individual sales - which it frequently does, because the same customer who would have bought one serum now buys the bundle instead - every conversion is less profitable. And Smart Bidding is paying premium CPCs for the privilege.

Feed Architecture for Bundles

The fix starts in your Shopping feed:

• • Separate bundle SKUs: Every bundle needs its own product ID, GTIN (or MPN), and product data. Don't inherit attributes from component products.
• • Accurate bundle COGS: The cost_of_goods_sold attribute must reflect the combined component costs plus any bundle-specific packaging. Not an average. Not the hero product's COGS.
• • Bundle-specific titles: Include "bundle," "set," or "kit" in the title. This affects search matching and ensures Google serves the bundle for bundle-intent queries, not product-intent queries.
• • Product group separation: Place bundles in their own product group within your Shopping campaign. This gives you independent bid control and prevents bundles from competing directly with their component products.
• • Custom labels: Use a custom label to flag bundles (e.g., custom_label_3 = "bundle"). This enables reporting segmentation so you can monitor bundle performance independently.

If your feed treats bundles as a product type within your main catalogue, they inherit bidding signals from individual products. The data bleeds in both directions, and neither the bundle nor the individual products get accurate optimization.

A Bidding Framework for Bundles

Once your feed is clean, adjust your bidding approach:

When Bundles Actually Work

Bundles aren't inherently bad for Google Ads. They work well in specific scenarios:

• • Stock clearance: Bundling a slow mover with a bestseller moves inventory that would otherwise sit. The SKU Job here is "Recovery" - the goal is cash, not margin.
• • New customer acquisition: A low-margin starter bundle acquires customers at a cost justified by LTV. The bundle's job is gateway - getting a first purchase that leads to repeat revenue.
• • Gift purchases: Gift buyers have different economics: lower return rates, no brand loyalty needed, higher urgency. Bundles serve this intent well and the margin trade-off is acceptable.
• • Genuine cost savings: When bundling actually reduces per-unit costs (shared packaging, reduced picks), the margin improvement is real, not fictional.

The common thread: bundles work when their commercial role is explicit and their economics are measured separately. They fail when they're thrown into the general Shopping feed and left to compete on revenue signals alone.

Next Steps

If you can't tell us the contribution margin on every bundle in your Shopping feed - separate from the component products - your bidding is based on inflated signals. Fix the data before you fix the campaigns.