Why is my ACoS good but my profit bad?

ACoS measures ad spend against ad attributed revenue only, so it can look healthy while the account loses money. It ignores four things that decide whether a click was worth buying: organic sales the ad cannibalised, returns on the units it sold, Amazon fees on those units, and cost of goods. A campaign at 22% ACoS on a product with a 26% gross margin after fees is losing roughly four cents on every dollar it books, and the advertising console will show that campaign in green the whole time.

The metric is not broken. It is answering a narrower question than the one you are asking.

What is the difference between ACoS, TACoS and profit per click?

ACoS is ad spend divided by ad attributed sales. It tells you how efficiently a campaign converts spend into sales it can claim credit for.

TACoS is ad spend divided by total sales, organic included. It is a better health signal for a product line, because a falling TACoS at flat revenue means organic is carrying more of the load. It is still a ratio of revenue, not of profit, so it can fall while contribution falls too.

Profit per click is the one that decides. Take contribution per unit after fees, returns and cost of goods, multiply by conversion rate, subtract cost per click. If the result is positive, buy more clicks. If it is negative, no ACoS target will save the campaign.

ACoS tells you how the ad performed. Profit per click tells you whether it should have run.

How does a campaign at 22% ACoS lose money?

Take a $42 product. Referral and fulfilment fees take $14.20. Returns average 5%, costing another $2.60 across all units. Landed cost is $11.60. Contribution before advertising is $13.60 per unit, about 32% of price.

At 22% ACoS the campaign spends $9.24 in ads per unit sold. That leaves $4.36. Looks fine, until you check the search term report and find that 61% of the ad attributed orders came from branded terms, where the customer typed your brand name and would have bought anyway. Strip those out and the incremental spend per genuinely new order is $23.70 against $13.60 of contribution. The campaign is not profitable. It is expensive attribution on sales you already had.

How do you find search terms with spend and no profitable orders?

This is the single highest yield hour in an Amazon advertising account, and it needs no tooling beyond a spreadsheet.

1. 01Pull the search term report for the last 60 days, all campaign types.
2. 02Filter to terms with spend above one target cost per acquisition and zero orders. That is pure waste, and it is usually 8 to 15% of spend.
3. 03For terms that did convert, compute contribution per order using landed cost, not revenue per order.
4. 04Flag any term where spend exceeds contribution. That is the second, larger pile.
5. 05Split the survivors into branded and non branded. Judge them on separate targets, because branded clicks are cheap defence and non branded clicks are growth you are paying for.

Contribution per order comes straight out of the Keep Rate work. If you have not built that yet, start with the full definition of Keep Rate and the waterfall behind it, because every advertising decision below depends on knowing what a unit actually earns.

Is pausing always the right answer?

No, and reflexive pausing is how accounts shrink. A term can lose money per order and still be worth funding in three cases.

• It is a launch term feeding rank on a product where organic position pays back over months, and you have measured the organic lift rather than assumed it.
• It is defensive coverage on your own brand where a competitor is bidding, and the alternative is handing over customers who already decided to buy from you.
• It converts into subscriptions or repeat purchase, so first order economics understate lifetime contribution. Check the actual repeat rate before you use this one, because it is the most abused excuse in the category.

Everywhere else, the choice is not pause or keep. It is bid down, cap the placement, tighten the match type, or move the term to a campaign with its own budget so it cannot quietly consume the product line's spend.

What should you judge campaigns on instead?

Three numbers, checked weekly, per product line rather than per campaign.

• Contribution after advertising, in dollars. Not a ratio. Ratios can improve while the business shrinks.
• Incremental cost per new to brand order, which separates growth spend from defence spend.
• Share of spend on terms with zero profitable orders in the trailing 30 days. Target under 5%. Most accounts we open sit between 11 and 18%.

None of this is complicated. It is just reconciliation, run often enough to catch waste while it is still small, which is the actual difference between an operator with a system and a monthly reporting cycle.