Everything you need to master Amazon advertising. From PPC campaign structure to DSP and AMC, these guides are built from managing $50M+ in ad spend across $2B+ managed.
Learn About Our PPC Management ServicesDeep-dive guides from our team who manage $2B+ on Amazon.
The definitive guide to Amazon PPC in 2026, covering Sponsored Products, Brands, Display, and budget allocation strategies.
Read the full guide →Proven PPC strategies our team uses daily to drive profitable growth for Amazon brands.
Read the full guide →Drew's personal playbook for structuring PPC campaigns that scale. The exact framework we use for $2B+ managed.
Read the full guide →How we leverage AI and automation to optimize bids, budgets, and targeting at scale.
Read the full guide →When to use DSP vs Sponsored Ads, how to set up campaigns, and what results to expect.
Read the full guide →Stop guessing. Learn to read PPC reports the way our analysts do, with actionable takeaways from every metric.
Read the full guide →Amazon runs three main ad types and they do different jobs. Sponsored Products promote a single listing inside search results and on product pages, and for most brands this is where the majority of spend and nearly all of the early learning happens. Sponsored Brands put your logo, a headline and several products at the top of a search page, which is worth more once people are searching for you by name. Sponsored Display follows shoppers with retargeting on and off Amazon, and behaves more like a demand channel than a search one. New to the terms? Our Amazon PPC glossary defines ACoS, TACoS, match types, placements and more in plain English.
Above those sits Amazon DSP, which is programmatic and reaches audiences who are not currently shopping. That is a different job with different economics, and judging it by the attribution window you use for search ads will always make it look like a failure.
A target ACoS is only meaningful against your break even. Thirty percent is healthy on a product with a 45% margin and a loss on one with a 25% margin. Most accounts we take over have a target that was chosen because it sounded reasonable, and nobody can say what it was derived from.
Write the target down with the reason attached: 25% because the margin is 42%, not 25% because it feels right. That one sentence changes what every later decision is measured against, and it is the difference between managing an account and guessing at one.
A product with a strong subscribe and save rate looks unprofitable on the first order and is strongly profitable across the first year. We have seen an ASIN running 55% ACoS that was 21% on a lifetime basis, comfortably inside the allowable acquisition cost. Cutting bids there does not fix a problem; it starves the best repeat purchase product in the catalogue.
Years of tightening to exact match, negating aggressively and refusing generic terms produces an efficient account with nowhere left to grow. The metrics all look good and the revenue line is flat. That is not a coincidence, it is the predictable end state of optimising one number.
On a plateaued brand doing about $10M a year we did the opposite of what the dashboard wanted: pulled back branded search and opened the account to broader terms. At the same spend, impressions rose by 1.2 million and cost per click fell from around $4.00 to about $0.70. Conversion rate got worse. ROAS got worse. Cost to acquire a customer fell from roughly $40 to roughly $11.
Three of those numbers moved in the wrong direction and the business got better. If your reporting is built to defend conversion rate and ROAS, that test looks like a failure and gets switched off in week two.
Shoppers increasingly search in full sentences and questions rather than two or three word fragments. An account tuned entirely to exact match cannot be found by a question it was never told to expect, which means the tight structures that worked for years now quietly cap discovery.
This does not mean abandoning exact match. It means leaving deliberate room for broader terms so the account can be found by phrasing nobody thought to add, then harvesting what converts back into exact.
The purpose of campaign structure is not tidiness, it is attribution. If you cannot tell which campaign produced which result, you cannot make a decision, and most messy accounts are messy because they were built to launch rather than to be read later.
Separate what you are trying to learn from what you are trying to protect. Branded defence, ranking pushes, launch products and mature performers all want different targets and different tolerances, and mixing them into one campaign averages away the signal you needed.
Automation and AI agents can genuinely run large parts of an account, but only against rules you have articulated. An agent connected to your ad data can see impressions, clicks, spend, orders, ACoS, CPC and conversion rate. It cannot see your margin, your pricing floors, your restock timing or why your target is set where it is.
Before anything touches a live account, write down four things: the target and the reason behind it, the maximum change allowed in a single pass, an explicit list of terms nothing may touch without a human, and what you will check after seven days. Without those an agent improvises on every run, and it improvises well enough that you may not notice for a month.
Advertising into a stockout. Pushing spend at a product weeks from running out buys a faster arrival at zero, and recovering a lost ranking costs far more than the missed sales did.
Judging every SKU by one target. A single account-wide ACoS goal treats a launch product and a mature hero as the same thing. They are not, and the launch product will always look like the failure.
Cutting the spend that feeds another channel. Branded search on Amazon is often created by Meta or TikTok spend. Cut the ad that shows no direct conversions and the Amazon best seller quietly declines with no explanation anyone can find.
Measuring too early. A week is not long enough to read a structural change. Give it a month and discard the weeks distorted by a deal or a competitor stockout.
There is no universal number, because ACoS only means something against your break even. A 30% ACoS is healthy on a 45% margin product and a loss on a 25% margin one. The more useful question is what the target should be and why, written down, rather than a figure that sounds right.
Only after checking what the order is worth. A product with a high subscribe and save rate can look unprofitable on first order economics and be strongly profitable across the first year. Cutting bids there starves the repeat purchase products first.
Usually because it was optimised into a plateau. Years of tightening to exact match and negating aggressively produce an efficient account with nowhere left to go. Growth normally requires accepting worse conversion rate and ROAS in exchange for a lower cost per acquired customer.
Shoppers increasingly search in full questions rather than short fragments. An account tuned entirely to exact match cannot be found by a question it was never told to expect, so leaving room for broader terms matters more than it used to.
Only with guardrails written down first: the target and the reason for it, a maximum change per run, an explicit list of terms it must never touch, and what you will check afterwards. Without those an agent improvises on every run.
90-day minimum, then month-to-month. We earn your business every month.
Book Your Free Strategy Call15 years of Amazon strategy distilled into actionable frameworks. The same playbooks we use for $2B+ managed, now available free.