On-Site Campaign Structure

Campaign structure determines whether retail media learns. Sponsored search, display, brand, and the negatives and placement modifiers that make the math work.

Why campaign structure matters more than creative

In paid social, creative does most of the heavy lifting and structure is mostly about cleanliness. In retail media, structure determines what data the auction sees, which determines whether your bids learn anything useful. A poorly structured Amazon or Walmart account can spend $50k/month while learning nothing about which keywords convert and which keywords burn cash on irrelevant clicks. A well-structured account makes every dollar a data point that improves the next dollar.

The principles below apply across Amazon, Walmart Connect, Target Roundel, Instacart, and most other major retail media platforms. Specific feature names differ; the structural logic is consistent.

The universal retailer ad hierarchy

• Account. The brand or seller entity. Billing and access control happen here.
• Portfolio. Groupings of campaigns by product line, brand, business unit, or budget owner. Used for reporting and budget capping.
• Campaign. The strategic container with budget, dates, targeting model (manual vs automatic), bidding strategy.
• Ad group. Where keywords, products, or audiences live. The unit of bidding-strategy variation.
• Ad / Product / Creative. The actual SKU or asset.

The structural decision at every level: how much variance do you want the algorithm to see at this level vs above it? Tight structure (one SKU per ad group, narrow keyword themes) gives you granular control and clean data. Loose structure (many SKUs per ad group, broad keyword themes) gives you flexibility and exposes you to wasteful spend on the wrong combinations. The right answer depends on your catalog size and management capacity.

Sponsored search campaign structures that scale

The dominant pattern in mature Amazon and Walmart accounts:

1. Branded campaigns. Separate campaigns for your own brand name + branded variants. Defensive. Low bid, high ROAS, near-certain conversions.
2. Competitor campaigns. Campaigns targeting competitor brand names and their high-traffic SKUs (via ASIN/ITEM targeting). Lower conversion rate but new-to-brand-rich.
3. Category campaigns. Generic category keywords ("running shoes," "protein powder"). High volume, lower conversion rate, prime new-to-brand source.
4. Long-tail / specific feature campaigns. Narrow, modifier-rich keywords ("women's zero-drop trail running shoes size 8"). High intent, lower volume, high conversion rate.
5. Automatic / discovery campaigns. Let the platform find new converting search terms. Mine them weekly into manual campaigns. Cap budget — automatic campaigns can burn cash on irrelevant queries.

Match-type architecture (Amazon-specific, similar logic across retailers)

• Broad match: Discovery. Find new converting queries. Lowest bid, lots of negatives needed.
• Phrase match: Mid-tier. Higher control than broad, more reach than exact.
• Exact match: Confirmed converters. Highest bid because conversion rate is known.

A defensible pattern: same keyword theme runs in three parallel campaigns (one broad, one phrase, one exact). Broad explores, exact harvests. Negative keywords flow between them so each campaign sees only its match-type traffic. This is the "funneling" structure that gives clean data while still discovering new terms.

SKU grouping in ad groups

The dominant tradeoff: tightly themed ad groups (1–3 SKUs each) give clean per-SKU data but require ongoing management of dozens or hundreds of ad groups. Loosely themed ad groups (10+ SKUs) reduce management overhead but obscure which SKU is driving performance.

Rule of thumb: tightly themed for your hero SKUs (top 20% by revenue), loosely themed for long-tail catalog (bottom 80%). The hero SKUs deserve the management attention.

Sponsored display

Sponsored display has three primary targeting modes:

1. Contextual targeting. Place ads on product detail pages for a chosen category or set of ASINs/UPCs. This is where you compete directly with rivals — appearing on their PDPs to convert their browsers.
2. Audience targeting. Reach shoppers based on past behavior — viewed your products, viewed your competitors, in-market for category, lifestyle audiences.
3. ASIN / ITEM targeting. Specifically target your own PDPs (cross-sell) or competitor PDPs (conquest).

The structural recipe: separate campaigns for each targeting mode and for each strategic intent (defensive on your own PDPs, conquest on competitor PDPs, prospecting on category audiences, retargeting on recent viewers).

Sponsored brand and brand store strategy

Sponsored brand ads (the banner at the top of search results) work as both lower-funnel conversion drivers and upper-funnel brand-building moves. The structural choice:

• Headline + 3 SKUs format: Defensible default. Brand-driven creative with a curated SKU selection.
• Video format: Higher CTR and conversion rate than static. Required for mature programs.
• Brand store spotlight: Sends to dedicated brand store landing page within the retailer. Use when you have a true catalog (10+ SKUs) and want to showcase range.

Brand stores are not just landing pages — they're a meaningful conversion channel in their own right. Mature brand stores get organic search visits, repeat shoppers, and serve as the landing page for upper-funnel awareness campaigns.

Bidding strategies and placement modifiers

Strategy	When to use
Dynamic bids — down only	Conservative default. Platform lowers bid for lower-converting impressions but never raises. Good for new campaigns.
Dynamic bids — up and down	Platform raises and lowers based on conversion likelihood. Use once campaigns have learning-period data.
Fixed bids	You control entirely. Use for specific tactical needs (defensive bidding on branded terms where you want exact bid certainty).

Placement modifiers (Amazon-specific but with parallels elsewhere): top-of-search and product-pages can be boosted by percentage. Top-of-search converts 2–4× better than rest-of-search for most categories. A +50% to +200% top-of-search modifier is common.

Negative keyword and ASIN architecture

The most important and least-glamorous part of retail media campaign management. Without disciplined negatives, broad and automatic campaigns burn budget on irrelevant queries indefinitely.

• Account-level negative libraries. Maintain a master list of terms that will never convert for your brand (competitor brand names you don't want to conquest, irrelevant categories, profanity, off-pattern queries).
• Campaign-level negatives. Keep broad/phrase/exact campaigns from cannibalizing each other (negative exact in broad campaigns).
• ASIN/Item negatives. For sponsored display, exclude competitor PDPs where you reliably lose, and exclude your own PDPs that should not get conquested by other campaigns.
• Refresh cadence. Weekly review of search term reports for new negative candidates. Monthly review of cumulative negative library for terms that should be added or removed.

Advanced playbook

• Day-parting for budget pacing. Some retailers (Amazon Sponsored Display, Walmart Connect) support dayparting. Lower bids during low-conversion hours (typically late-night) on most categories.
• Brand defense bidding ceiling. Don't pay top-of-page on your own branded terms unless competitors are bidding. Monitor for competitor entry and respond.
• SKU-level ROAS analysis. Some SKUs convert at 8× ROAS, some at 1×. Concentrate budget on high-converters; remove or move chronically underperforming SKUs.
• Seasonal campaign duplication. Duplicate your evergreen campaign structure with seasonal modifiers ("Black Friday," "back to school") so you don't pollute the always-on campaigns' bid learning.
• New product launch sequencing. For new SKUs: start with broad/automatic campaigns to discover queries. Within 2–3 weeks, mine converting queries into exact-match campaigns. Pause automatic once exact-match has scaled.
• Inventory-driven pacing. Sync ad spend to inventory levels. Low stock = pause campaigns (no point bringing shoppers to OOS PDPs). High stock = boost spend to clear.
• Margin-tiered bidding. Different SKUs have different margins. Bid higher (relative to ROAS target) on high-margin SKUs even if their conversion rate is slightly lower.
• Conquest campaign discipline. Conquest is expensive. Cap budget. Measure new-to-brand %, not just ROAS. If new-to-brand stays below 30%, you're paying to talk to your own customers.
• Brand store traffic source diversification. Drive paid traffic to brand stores from sponsored brand ads, but also from off-site display, CTV, and external social ads where supported.
• Bid harvest cadence. Monthly bulk operations: identify keywords above ROAS target with low impression share and raise bids; identify keywords below ROAS target and lower or pause.

Common mistakes

• One catch-all campaign with hundreds of SKUs and broad-match keywords — impossible to optimize.
• No negative keyword discipline — broad campaigns slowly drift into irrelevance.
• Bid setting that ignores margin — all SKUs treated equally regardless of contribution.
• Running automatic campaigns indefinitely without mining for manual.
• Top-of-search modifier defaulted at +0% — you're leaving the highest-converting placement to chance.
• Conquest campaigns running without new-to-brand measurement.
• Branded defense spend unbounded — paying premium on your own brand name regardless of competitive pressure.
• Sponsored display run as "set and forget" without distinct campaigns for contextual, audience, and ASIN.
• Brand store ignored as a landing page choice when it would outperform PDP for catalog-rich brands.
• Campaign structure unchanged for 18+ months while catalog has tripled in size.

Operating checklist

• Account organized into portfolios by brand, business unit, or budget owner
• Branded, competitor, category, long-tail, and automatic/discovery campaigns separated
• Match-type funneling structure (broad / phrase / exact in parallel) for top categories
• Hero SKUs in tightly themed ad groups; long-tail SKUs grouped more loosely
• Sponsored display split by contextual / audience / ASIN with distinct strategic intent
• Sponsored brand running headline+SKU, video, and brand store variants
• Brand store live, optimized, and serving as upper-funnel landing page
• Top-of-search and PDP placement modifiers set per category
• Account-level and campaign-level negative keyword libraries maintained weekly
• SKU-level ROAS analysis monthly; budget shifts based on contribution and margin
• Inventory-pause automation in place
• Quarterly structural review — campaigns split/merged as catalog evolves

---

Part of the Retail Media series. Continue to the next module or take the series exam.