The Amazon PPC Guide That Does the Math With You

What Amazon PPC Costs You, in Real Numbers

Amazon PPC is pay-per-click advertising inside Amazon search and product pages: you bid on keywords or products, and Amazon charges you when someone clicks, not when someone sees the ad. The question that actually matters isn't "what is it" — it's "what ACoS keeps me profitable," and that number is different for every product.

Work it out from the product, not from a benchmark you read somewhere. Say you sell a $25 item. Amazon's referral fee takes $3.75 (15%). FBA fulfillment costs $5.50. Cost of goods is $6.00. What's left before ad spend is $9.75 — a 39% margin. That 39% is your break-even ACoS: spend more than that relative to sales, and the ad is losing you money on that unit, full stop.

Set your target ACoS below break-even, not at it — overhead and returns eat into that margin too. Targeting 25% ACoS on that $25 item with a 12% conversion rate, the math runs backward into your maximum cost per click: 0.25 × $25 × 0.12 = $0.75. Bid meaningfully above $0.75 on that keyword and no amount of "optimization" fixes a bid that's structurally too high for the economics.

Any guide that stops at defining ACoS without walking through this math is skipping the only part that tells you whether the account is actually working.

The Three Ad Types, and When Each One Actually Earns Its Placement

Sponsored Products put a single listing into search results and on other product pages. They carry most PPC budgets because they sit closest to the purchase decision. Start here on a new product.

• Sponsored Products — bids on keywords or ASINs, appears where organic listings appear. Best for direct sales and new-product visibility.
• Sponsored Brands — requires Brand Registry, shows a logo and up to three products at the top of search. Better for cross-catalog awareness than for one SKU's ACoS.
• Sponsored Display — follows shoppers off Amazon and retargets people who viewed a listing without buying. Weakest at driving first-touch sales, strongest at recovering ones you already paid to generate once.

Sellers who run Sponsored Brands expecting Sponsored Products-level ACoS are usually disappointed for the wrong reason — it isn't underperforming, it's doing a different job.

Building a Structure You Can Actually Debug Later

Start every new product on an automatic campaign. You don't have search-term data yet, and Amazon's matching will surface terms you wouldn't have bid on manually. Let it accumulate real clicks — judging it after two days tells you almost nothing.

Once the search term report has volume, split winners into a manual campaign on exact match, and set those same terms as negative exact in the automatic campaign so you're not bidding against yourself for the same click. This one step — negating what you've promoted — is the most commonly skipped part of the whole setup.

Keep campaign names literal: product, campaign type, match type. You will be reading this account at 11pm eventually. Write for that version of you.

When the Numbers Are Bad: What to Check, in Order

A bad number is information, not a verdict. Before touching a bid, work out which of three things happened: not enough people saw the ad, enough saw it but didn't click, or enough clicked but didn't buy. Each has a different fix, and fixing the wrong one wastes a week.

Low impressions usually means your bid lost the auction or the budget capped out early in the day — check hour-by-hour spend before touching keywords. High CTR with low conversion usually means the ad is matching the right search but the listing isn't closing — that's an images-and-price problem, not a PPC problem, and no bid change fixes it.

Whatever you change, attach three things to it before you run it: the evidence that justifies the change, a way to measure whether it worked, and a point at which you reverse it if it didn't. Every proposed change carries three things before it runs: the evidence behind it, a measurement plan, and a rollback trigger. That's the difference between optimizing and just moving numbers around.

Mistakes That Cost More Than a Bad Bid

Setting up and walking away. A campaign that ran fine in month one drifts as competitors' bids move and your own catalog changes. No account runs itself for six months without a search-term review.

Changing five things at once. Raise a bid, add ten negative keywords, and switch to dynamic bidding in the same afternoon, and when performance shifts you'll have no idea which change did it. Change one lever, wait for enough clicks to mean something, then change the next.

Chasing ACoS to zero. We've throttled bids down to hit a target ACoS and quietly killed a search term's organic rank momentum along with it — the campaign looked cheaper on the weekly report and the product ranked worse a month later. Sponsored and organic rank interact; optimizing one in isolation can cost you the other.

Treating Sponsored Display like Sponsored Products. Judging a retargeting campaign against a search campaign's ACoS and pausing it for "underperforming" throws away a channel that was never supposed to compete on that metric.

Where This Goes When You Want It Off Your Desk

Everything above is manual work: pulling search term reports, doing the break-even math per SKU, deciding what to change and in what order. Some sellers do this well themselves. Others run enough SKUs, or spend enough on ads, that the hourly cost of doing it by hand exceeds the cost of handing it off.

Dr. PPC, built by Full Circle — an agency that has managed more than $500M in revenue across 100+ brands — reads the account, writes a strategy against that product's real margin (the same break-even math above, done per SKU, continuously), and proposes each change with the evidence, measurement plan, and rollback trigger attached. You choose how much runs on autopilot: everything queued for your click, routine changes automatic with the bigger ones held for review, or fully autonomous inside guardrails you set. Pricing is $300 a month plus 3% of ad spend, capped, month-to-month, with the first 30 days free — and Orbit, the analytics and tracking suite underneath it, is included rather than sold separately. Whether or not you ever use it, the discipline it enforces — evidence, measurement, rollback, before the bid changes — is the right way to run PPC by hand too.