Amazon Sponsored Ads Tips: The Math, the Mistakes, and the Fixes

The one number every tip list skips: your break-even ACOS

ACOS (spend divided by attributed sales) tells you nothing about profit on its own. It's a ratio, not a verdict. What matters is where that ratio sits against your margin.

Say a product sells for $30 and, after cost of goods and Amazon fees, keeps 35% gross margin — $10.50. If you want to protect 15 points of that margin after ad spend, your ceiling is a 20% ACOS (35% margin minus 15% target profit). Push past that and you're paying to lose money, no matter how good the campaign looks in the dashboard.

Run this math per SKU before you set a single bid. A single account-wide ACOS target is a shortcut that quietly kills your best-margin products and props up your worst ones.

Automatic vs. manual targeting: use both, in sequence

Automatic targeting isn't a beginner setting you graduate out of — it's a discovery tool. Let Amazon match your listing to real search terms for two to four weeks. You're not trying to win yet; you're trying to see what people actually type before they buy your product.

Once a term has enough clicks to mean something — fifteen to twenty is a reasonable floor, not three — move the winners into a manual exact-match campaign with a bid based on that term's own conversion rate, not a category average. Leave the losers in auto or negative them out.

The common failure is leaving auto campaigns running indefinitely on default bids. You end up paying the same amount for a branded search as for a vague, low-intent discovery term, because nothing ever separated the two.

When the fix doesn't work: diagnose before you adjust again

The instinct when ACOS goes up is to cut the bid. Often that's the wrong lever. A bid change only fixes a bid problem — and a lot of what looks like a bid problem is a price change, a stockout, a lost Buy Box, or a listing edit that happened somewhere else in the account.

Before adjusting a bid a second time, check what actually changed: the listing, the price, the inventory, the competitive set. The table below covers the symptoms we see most.

The mistake worth naming — because we've made it too

We once cut bids on a launch keyword because ACOS spiked in the first week. The real cause was a stockout that hadn't surfaced in the dashboard yet. The bid change did nothing, because the bid was never the problem — and we lost several days chasing the wrong fix before checking inventory.

That's why any change worth making needs three things attached before it runs: the evidence behind it, a measurement plan, and a rollback trigger. If you cut a bid, decide in advance what result reverses the decision and by when. Write it down before you make the change, not after it disappoints you. Otherwise you're guessing twice instead of once.

Other honest mistakes worth avoiding: negative-exacting a search term after three clicks and no sale (that's noise, not data); harvesting so aggressively that your manual campaigns fragment into hundreds of single-digit-impression keywords nobody checks again; and treating Sponsored Display as an afterthought when it's often the cheapest way to defend your own listing from competitor targeting.