How to Structure Amazon PPC Campaigns

Amazon PPC is organized in a strict hierarchy: Portfolios → Campaigns → Ad Groups → Targets → Ads. Understanding which setting lives at which level matters because it determines where you can pull a lever. Budget and the auto/manual decision both live at the campaign level. Default bids and keyword themes live inside ad groups. Portfolios are folders ; they do not change how ads run, but they change how you organize and report across a multi-product account. If you sell multiple products, portfolios are among the highest-leverage organizational upgrades you can make.

Amazon offers three Sponsored ad formats: Sponsored Products, Sponsored Brands, and Sponsored Display. Sponsored Products promote individual listings and are the only format that supports automatic targeting ; making them the structural foundation for most accounts. Sponsored Brands promote a brand with a custom banner including a logo, headline, and up to three products, and require Brand Registry enrollment. Sponsored Display enables retargeting both on and off Amazon, reaching shoppers who viewed your listings or similar items. Each format serves a different intent point in the funnel and needs to live in its own campaign layer so the data stays readable and optimization decisions stay separated.

The starting point for any product is two Sponsored Products campaigns running in parallel: one automatic and one manual. Automatic targeting lets Amazon's algorithm decide which search terms and product pages trigger your ads, based on your product listing content and previous shopping queries. Manual campaigns give you direct control over every keyword or ASIN you target. Together they form a discovery-to-performance pipeline: auto finds what converts, manual scales it. Without this paired structure, downstream optimizations ; match-type segmentation, search term harvesting, bid control ; cannot work properly.

The failure mode of running only auto campaigns is that Amazon controls spend allocation at every level. Every dollar goes wherever the algorithm decides, at a flat bid per targeting group. You cannot raise the bid on a term that converts at 12% while holding back on one you are still testing. The failure mode of launching manual campaigns too early ; or skipping auto entirely and guessing at keywords ; is that you burn budget without any signal to bid against. The correct sequence is auto first, data second, manual third. Switching from discovery mode to targeted spending should happen once you have enough conversion data to justify a confident keyword list.

Inside your manual Sponsored Products campaigns, Amazon offers three keyword match types. Broad match has the widest reach: it triggers on your keyword's terms in any order, plus plurals, variations, synonyms, and semantically related queries. Phrase match is tighter ; your phrase triggers in order, with extra words allowed before or after. Exact match is the most controlled: the query must match the keyword closely. These match types trade reach for control, and combining them inside a single campaign creates a structural problem: aggressive broad terms consume budget that you intended for proven exact-match terms.

The solution is to keep broad, phrase, and exact match types in separate campaigns. This lets you set different daily budgets for research versus performance, apply different placement modifiers per match type, and optimize bids without one match type's behavior inflating or diluting another's ACoS. A single campaign housing multiple match types feels organized, but it hides performance data and creates budget imbalance at the ad group level. The best structure separates campaigns by targeting type ; auto versus manual ; and then by match type: broad, phrase, and exact each in their own campaign. Branded, non-branded, and competitor keyword intents add another axis of separation once a catalog matures.

Search term isolation is the mechanism that converts auto-campaign discovery into manual-campaign profitability. The workflow has four steps. First, run the automatic or broad-match campaign as a keyword discovery engine and let Amazon surface real purchase-intent queries. Second, review the Search Term Report ; available in Seller Central or the Amazon Ads console ; and identify terms that have generated two or more orders at or below your target ACoS. Third, promote those terms as Exact Match keywords in your manual campaign with a controlled bid based on actual performance data. Fourth, add the same terms as Negative Exact Match in the auto or broad campaign so the auto campaign stops competing for terms your manual campaign now owns.

Without the negation step, both campaigns bid on the same query simultaneously. Your auto campaign continues spending on proven winners at discovery-level bids ; typically lower than the optimized bids in your manual campaign ; while your manual campaign receives less volume than expected because the auto campaign captures much of the traffic. This is keyword cannibalization: you are bidding against yourself, which inflates CPCs, corrupts performance data, and drains budget. Adding negatives at the ad group level is the correct default for most terms. Campaign-level negatives are reserved for terms you know with certainty should never match anywhere inside that campaign, such as gender-specific exclusions on unisex products. Phrase negatives should be used carefully because they block entire root queries and can accidentally remove profitable long-tail traffic.

For new product launches or high-spend campaigns, reviewing the Search Term Report weekly is the minimum cadence. A weekly review catches irrelevant spend before it permanently damages ACoS and launch momentum. For mature, stable campaigns with consistent performance history, a biweekly check is reasonable. The practical cadence is: search term harvesting and negative keyword updates weekly, bid adjustments weekly or biweekly, budget allocation review monthly.

Before adjusting any bid, decide what PPC is supposed to accomplish for the product right now. A campaign that looks like it is failing at one goal can look healthy at another. There are three distinct modes. In launch mode, you accept a higher ACoS to gain sales velocity, reviews, and organic ranking. In advertising profit mode, you want low ACoS and profitable ad-driven sales even if growth is slower. In total profit mode, you optimize for overall business profit and accept a rising ACoS if total profit increases. Mixing these goals inside the same campaign produces unreadable data and the wrong bid decisions.

Set total budget based on your target TACoS ; Total Advertising Cost of Sale ; which measures ad spend as a percentage of total revenue, not just ad-attributed revenue. A TACoS between 8% and 15% is typical for most established Amazon sellers, though this varies by margin and growth stage. New product launches may run higher temporarily to build organic ranking and review velocity. A reasonable starting budget split for a mature Sponsored Products account is roughly 50 to 60 percent toward manual exact campaigns, 15 to 20 percent toward manual phrase, 10 percent toward broad and auto for ongoing discovery, and 10 to 15 percent toward product targeting campaigns. These ratios shift based on category, competitive intensity, and how mature a product is in its launch curve.

High-margin products warrant a larger budget allocation to maximize exposure. Products with smaller margins require a more conservative approach to avoid spending above break-even ACoS. One practical failure mode to watch: campaigns hitting their daily budget cap stop serving ads before the day ends, which means your best-converting time windows may be unserved. When a campaign consistently hits its cap, either increase the budget or reduce bids on lower-priority ad groups ; do not let the cap become a silent performance ceiling.

Campaigns named 'Campaign 1,' 'Test,' or 'New Campaign (3)' make reporting, filtering, and optimization nearly impossible at scale. A structured naming convention should be established from the start. A widely used format encodes the essential information directly into the name: Product Identifier ; ASIN ; Campaign Type ; Target Match Type ; Goal. For example, a manual exact-match performance campaign for a blue yoga mat targeting non-branded terms might be named: YogaMat-B0XXXXXX-SP-Manual-Exact-NonBrand-Performance. If you manage multiple marketplaces, include the marketplace code so campaigns are distinguishable across regions without opening each one.

Portfolios should group campaigns by product line, brand, or objective ; launch versus defense, for example ; rather than simply lumping everything together. This makes it possible to see total spend and performance per product cluster without manually summing campaign rows. A brand with three distinct product clusters might run roughly eighteen Sponsored Products campaigns plus separate Sponsored Brands and Sponsored Display layers. Sponsored Brands campaigns, when used, should be organized by brand defense, category capture, and competitor conquest ; each in its own campaign with distinct creative ; because headline placement and video formats convert differently from Sponsored Products and the data needs to stay separated to be readable. Sponsored Display similarly needs its own structural tier: product targeting, audience targeting, and view remarketing each serve different intent points in the funnel, and treating them as one campaign type collapses three different optimization conversations into one undifferentiated dataset.

The most expensive structural errors in Amazon PPC are not bid-level mistakes ; they are architecture mistakes that corrupt the data those bid decisions depend on. Keyword cannibalization between auto and manual campaigns is one of the most expensive and most avoidable. A product with 1,500 reviews and a product with 3 reviews should not share the same campaign structure or budget allocation; lifecycle-based structuring is necessary as a catalog matures. Campaign structures built twelve or more months ago often no longer reflect the current product catalog, keyword data, or business goals, and should be audited and restructured accordingly.

Two other failure modes to monitor: first, adding negative keywords too early when a term has only a few clicks ; the data may be too thin to judge whether it is truly unprofitable or simply slow to convert. Second, product page placements often generate high traffic but low conversion rates, which can drain budget quickly. Before over-negating queries, review placement performance and reduce product page placement bids if that placement is consistently underperforming ; this is a placement lever, not a keyword problem. A weekly optimization routine that checks search terms, negative keywords, bids, budget caps, and campaigns with zero impressions, combined with a monthly review of ACoS, TACoS, ROAS, CTR, CPC, and CVR, gives you the visibility to catch structural drift before it becomes a profitability problem.