The Amazon PPC Guide for 2026: Campaign Structure, Bids & ACoS

Most sellers don't lose money on Amazon ads because their bids are too high. They lose it because their account has no structure — every keyword piled into one campaign, no separation between research spend and proven winners, and no way to read which dollar did what. Fix the architecture first and the bidding gets easy. This guide walks the campaign types, the structure that scales, and the profit math that keeps spend honest.

The three campaign types, and what each is for

Amazon's ad console gives you three core placements. They're not interchangeable — each does a different job in the funnel, and treating them as one bucket is the first mistake.

• Sponsored Products (SP) — the workhorse. Single-product ads that appear in search results and on product pages. This is where the majority of your spend and nearly all your conversions will live. If you only run one type, run this.
• Sponsored Brands (SB) — banner-style ads with your logo, a headline, and multiple products, plus the Sponsored Brands Video format. Requires Brand Registry. Best for defending your brand term, introducing a catalog, and capturing top-of-search real estate.
• Sponsored Display (SD) — audience and product-targeting ads that can follow shoppers on and off Amazon. Strong for retargeting browsers who didn't buy and for going after competitor detail pages.

Start with Sponsored Products. Layer in Sponsored Brands once you have a hero product and Brand Registry. Add Sponsored Display when you have enough traffic to retarget meaningfully — running it cold rarely pays.

Account structure that actually scales

The structure that survives growth separates discovery from harvest. You run broad, loose campaigns to find what converts, then promote winning search terms into tight, controlled campaigns where you can bid them aggressively without dragging unrelated terms along.

1. Discovery layer — automatic campaigns plus broad and phrase match keywords. Let Amazon show you search terms you'd never have guessed. Keep bids modest; this layer's job is data, not volume.
2. Harvest layer — when a search term from discovery proves it converts, add it as an exact-match keyword in a dedicated campaign and add it as a negative in the discovery campaign. Now your winners get their own budget and you've stopped paying twice for the same click.
3. Brand-defense layer — a separate campaign on your own brand terms. Cheap clicks, high conversion, and it keeps competitors from buying their way onto your name.

Keep one product (or one tight variation group) per campaign. The moment you mix unrelated ASINs, your reporting turns to mush and you can't tell which product is actually carrying the spend.

How to think about bids

Your bid is a ceiling on what you'll pay for a click — Amazon usually charges less. The lever that matters more than the number is the bidding strategy: fixed bids hold steady, while dynamic strategies let Amazon raise or lower in real time based on conversion likelihood. Down-only is the safe default while you learn a campaign; it won't overspend on placements that aren't converting.

A sane bidding loop looks like this:

• Launch new keywords near the suggested bid, then adjust on data — not on day one.
• Raise bids on terms that convert below your target ACoS and have room to grow.
• Cut or pause terms with meaningful click volume and zero or near-zero conversions.
• Use placement modifiers to pay up for top-of-search only where it earns back the premium.

Give every change a window — a handful of conversions or a couple of weeks — before you judge it. Reacting to a single day's noise is how good campaigns get strangled.

ACoS vs TACoS — the number that protects profit

ACoS (Advertising Cost of Sales) is ad spend divided by ad-attributed sales. It tells you how efficient a campaign is in isolation. But it ignores the sales your ads *help* drive organically, so optimizing ACoS alone can quietly shrink the business.

TACoS (Total Advertising Cost of Sales) is ad spend divided by *total* sales — ads plus organic. Watch it over time: if TACoS falls while revenue climbs, your ads are buying durable organic rank, which is exactly what you want. We go deep on this in What Is TACoS?.

The trap is chasing a low ACoS as a vanity number. A 15% ACoS on a product with a 12% net margin still loses money. The only ACoS that's "good" is one your actual unit economics can absorb — which means you have to know your true profit per unit before you set a target at all.

Scaling without torching margin

Scaling isn't "raise every bid." It's widening the funnel where it's working and tightening it everywhere else. Pour budget into harvest campaigns that convert under target, keep feeding the discovery layer just enough to surface new winners, and let underperformers starve. Review search-term reports on a regular cadence, promote and negate relentlessly, and let TACoS — not a single day's ACoS — tell you whether growth is actually compounding.

Frequently asked questions

What's a good ACoS target on Amazon?

There's no universal number — it depends entirely on your net margin. A useful frame is your break-even ACoS, the point where ad spend equals the profit on that sale. Anything below it is profitable on a per-click basis; above it you're paying for rank or volume on purpose. Calculate your true margin first, then set the target.

Should I run automatic or manual campaigns?

Both, on purpose. Automatic campaigns are your discovery engine — they surface converting search terms cheaply. Manual exact-match campaigns are where you control and scale those proven terms. Run autos to find, manuals to harvest, and negate the winners out of autos so you're not bidding against yourself.

How long before I judge a new keyword?

Wait for enough clicks to be statistically meaningful — generally a couple of weeks or a handful of conversions, whichever comes first. Judging a keyword after a day or two of data leads to pausing terms that just needed time, or scaling ones that got a lucky early sale.