Part 2 of 6

Google Ads Campaign Architecture: Branded Search + Tiered Shopping

This is part 2 of a 6-part Google Ads framework — see Part 1 for the foundations this architecture depends on. A full scaling account runs 5 campaign layers total; this article covers the two that fund everything else. Part 3 covers the other three (top-of-funnel, mid-funnel, and conquest).

Layer 1: branded search — protect it, don't optimize it

Keywords: [brand name], [brand name] + product types, [brand name] + reviews/discount/vs/alternative].

Bidding: manual CPC. Always. Smart Bidding overpays on branded traffic that was going to convert anyway: the algorithm doesn't know the click was inevitable, so it bids like it's competing for a cold customer. Manual CPC holds 95%+ impression share at the lowest bid that clears the auction.

Landing page: product page or homepage. This traffic already knows you. It doesn't need education, it needs a fast path to checkout.

Campaign setup: one dedicated campaign, two ad groups maximum — exact match for known brand terms, phrase match to catch variants and misspellings. Negative your own brand terms out of every other campaign in the account, without exception, so spend never gets misattributed.

Target: 95%+ impression share. Below 90% means competitors are actively bidding on and stealing your branded queries: check your auction insights before anything else in this campaign.

Layer 2: tiered shopping — where most ecommerce revenue actually lives

80% of ecommerce Google Ads spend typically flows through shopping placements. This is the layer that carries the account.

• Tier 1: hero SKUs. Your top 5-10 products by revenue and margin. 40-50% of total shopping budget. ROAS target set slightly below account average: deliberately, to let the algorithm spend into winning opportunities instead of throttling your best sellers.
• Tier 2: growth products. Items showing commercial potential but not yet proven at scale. 25-30% of budget. ROAS target 15-20% higher than Tier 1, to keep the campaign efficient while it invests in unproven upside.
• Tier 3: long-tail catalog. 15-20% of budget. Highest ROAS target of the three: efficiency over volume, because these products don't have the margin or demand to justify aggressive spend.

The algorithm can't prioritize what it can't see. Custom labels (built in Part 1) are what give it that structure: margin tier, bestseller status, use case, price tier — mapped onto listing groups or PMax asset groups.

The 2026 reality: Performance Max changes how you build tiers

The original version of this playbook assumed pure Standard Shopping campaigns with manual listing-group bidding per tier. That still works, and if you want granular manual control over negatives, audiences and bids by product group, Standard Shopping remains the right tool: it isn't dead.

But Performance Max now sits at a structural priority above Standard Shopping in Google's auction, and it's where most new ecommerce budget is going. PMax doesn't expose the same manual listing-group bid controls. Instead:

• Build custom labels exactly as described in Part 1 (margin, bestseller, use case, price tier).
• Structure asset groups around those labels, and let PMax's conversion-value optimization do the tier-weighting: it will naturally push more impressions toward high-value-signal products if your feed and conversion values are accurate.
• If you want a hard tier split (not just algorithmic weighting), run both campaign types with careful product exclusions so Standard Shopping and PMax never compete for the same SKU: for example, hero SKUs in a tightly controlled Standard Shopping campaign, and everything else in PMax where broader automation is less risky.
• Whichever structure you choose, the tiering logic — protect margin, fund growth, contain long-tail — doesn't change. Only the mechanism does.