Amazon Advertising Strategy in 2026: Structure Before Tactics

Every tips list tells you to add negative keywords, use the right match types, and raise bids on converters. That advice is reasonable. It does not help when the account cannot answer a simpler question: which campaign is supposed to make money, and which one is supposed to buy growth?

When one campaign carries a launch SKU, a mature bestseller, and a clearance item, its reported efficiency blends three different jobs. Any decision based on that number is wrong for at least two of them.

Statisticians call this Simpson's paradox: a trend that holds within each group can reverse when the groups are combined. In ad accounts, the cause is usually a shift in where spend and sales land.

Illustrative example. One campaign advertises a bestseller and a launch SKU. These figures are hypothetical.

Both products became more efficient, but the campaign's combined ACoS increased from 22.7% to 29.4%. The reason was a shift in the sales mix toward the launch SKU, which still carried a higher ACoS.

A manager relying only on the campaign total might reduce bids or budgets across the campaign, including for the improving bestseller. Reviewing performance by ASIN reveals the real cause of the increase. Campaign ACoS is total advertising spend divided by total attributed sales, not a simple average of individual product ACoS values.

Amazon's own Sponsored Products targeting guide recommends defining a business goal for each product before choosing targeting. It also advises against mixing different targeting strategies within the same campaign or ad group, so each tactic can be measured against the right KPI. CMO turns that principle into four planning objectives. These are CMO planning objectives, not Amazon campaign types.

Within each objective, separate products whose margins, conversion rates, or lifecycle stages differ. That keeps every campaign's results readable without untangling several jobs at once.

CMO structure rule: a campaign should answer one question. If you cannot say in one sentence what a campaign is for and how you will know it worked, it is a container rather than a campaign. Containers are where budget quietly disperses.

Sponsored Products usually carries harvest, defend, and scale work, and is judged on cost per sale against the target for each objective.

Sponsored Brands supports brand visibility, product discovery, and video. Judge it on CPC, conversion rate, new-to-brand metrics, and downstream sales against its objective. Its costs are not always higher than Sponsored Products.

Sponsored Display can support audience engagement, product targeting, and remarketing, depending on campaign type. Confirm the billing model and placements in the current console before choosing it.

Amazon DSP is a separate purchase with its own minimums and buying model. Our DSP minimum spend guide covers what it costs to run.

Formats are not a ladder to climb. A brand with conversion problems on its detail pages rarely fixes them by adding another ad type.

Base budgets on contribution margin, objective, demand, and growth potential. Protect proven profitable campaigns from avoidable budget interruptions, and do not automatically put branded demand ahead of non-branded acquisition.

Discovery is a research budget. Decide in advance what you will spend to learn something, and treat overspend as a cap failure rather than evidence the campaign is working. Scale budget should expand only on targets that have proven both conversion and margin, not volume alone.

Worked example: allocating a $20,000 monthly budget. A brand splits its budget by objective rather than by product:

• $9,000 to harvest on proven non-branded category and exact-match terms
• $3,000 to defend branded keywords and its own detail pages
• $3,000 to discover through automatic, broad-match, and exploratory product targeting
• $5,000 to scale validated non-branded targets with expanded budgets and placements

Each line gets its own target. Harvest is judged on ACoS against a margin-based target. Defend is judged on branded visibility and efficiency. Discover is judged on qualified new terms per dollar. Scale is judged on contribution profit and total-sales growth.

Two months later, discovery has produced four converting terms that now run in their own exact-match harvest campaigns. The discovery line drops to $2,000 and harvest rises to $10,000. The split is a decision you revisit, not a fixed ratio. These figures are illustrative, so derive yours from contribution margin and demand.

Four leaks account for a large share of wasted spend, and only one is a bidding problem.

Conversion, not clicks. Ad cost per sale equals CPC divided by conversion rate, so the listing affects every click you buy. See our guides to clicks without sales and Amazon cost per click.

Search term drift. Broad and automatic campaigns gradually pick up traffic that never converts. Amazon's targeting guide recommends evaluating a term after it has received at least 20 clicks before adding it as a negative, and letting negative keywords run for two weeks or more before changing strategy. Treat 20 clicks as a review point, not an automatic exclusion rule. Our keyword research guide covers separating terms that bring traffic from terms that bring buyers.

Stockouts. Sponsored Products ads generally stop serving when a product is out of stock, so an inventory gap is also an advertising gap. Our inventory forecasting guide covers prevention.

Advertising the wrong SKUs. Not every product deserves ad support. Model contribution margin after fees before deciding which ASINs get budget, as our product screening guide explains.

Break-even ACoS comes from your margin, and the target you actually run depends on the objective. Our ACoS guide covers deriving both from your own numbers. Work in this order:

1. Structure. Campaigns grouped by objective and product economics, each answering one question.
2. Targets. One number per objective, derived from margin.
3. Targeting. Match types and product targets aligned to the objective.
4. Budgets. Allocated by objective and reassessed as conditions change.
5. Bids. Last, and the smallest lever.

Sponsored Products offers dynamic bids down only, dynamic bids up and down, and fixed bids, plus rule-based bidding for eligible campaigns. When you combine dynamic bidding with placement adjustments, review the total bid exposure that results rather than judging each setting on its own.

Campaign reports show attributed sales, not total sales. To calculate TACoS, pull total Amazon sales from Seller Central Business Reports and divide total ad spend by that figure. Read campaign ACoS alongside TACoS and total sales, because ad-attributed metrics can look stable while the business moves. Attributed sales also differ from incremental sales: ACoS and ROAS show what ads were credited with, not what they caused.

On January 1, 2026, Amazon introduced a shopping-signal enhanced last-touch attribution model. It applies to Sponsored Brands, Sponsored Display campaigns billed on vCPM, and Amazon DSP campaigns serving in the Amazon store. The model credits ad views during early discovery moments and uses a shorter attribution window. Click-based attribution did not change.

For year-over-year comparisons, Amazon still offers "all views" metrics, such as Purchases (all views), with a 14-day window in the Reporting (beta) tab. Use those columns, or compare only click-attributed results, before concluding that a 2026 campaign got better or worse. Our attribution guide covers how to read the reports.

Before a restructure: Export at least 30 days of campaign, targeting, and advertised-product data. Record spend, attributed sales, orders, ACoS, conversion rate, and budget utilization. Obtain total Amazon sales from the appropriate Seller Central business reports to calculate TACoS. Document inventory availability, promotions, price changes, and listing updates that could affect the comparison.

After implementation: Compare equivalent periods using consistent attribution definitions and sufficiently mature conversion data. Evaluate ACoS by campaign objective, contribution profit by ASIN, TACoS, total sales, and budget allocation. Do not attribute improvements to the restructure alone when pricing, inventory, seasonality, or promotions also changed.

Weekly: review search terms on discovery campaigns, check pacing across objectives, and reduce bids on targets that passed a meaningful click volume without converting.

Monthly: review budget allocation by objective, placement performance, and conversion rate by ASIN. Graduate proven terms into their own exact-match harvest campaigns.

Quarterly: run a structural review. Does every campaign still answer one question? Have new SKUs landed in the wrong campaign? Do targets still reflect current fees and margins?

During launches, promotions, and seasonal peaks, check pacing more often. Treat this schedule as a floor.

A useful first question in any PPC audit is structural: can each campaign's objective be explained in one sentence? The answer helps determine whether the account needs a listing improvement, a campaign restructure, or routine bid optimization.