If you're running Amazon PPC right now and your ACOS looks fine but your margins don't, this is probably happening to you somewhere in your account. The fix is rarely a new bid rule pulled from a template. Optimize Amazon ads in an order that tells you what is actually broken.
Start with the business math. Then check the conditions around the ads. Then read the search and purchase data. Only after that should you adjust bids, negatives, budgets, or campaign control.
1. Start with the product economics
ACOS is a ratio between advertising spend and attributed revenue. It is useful for comparing campaigns, but it is not a health score for the business. Before you decide whether a campaign is efficient, calculate what one sale can afford to spend.
Use the selling price and subtract the costs that matter for the SKU: cost of goods, Amazon fees, shipping, refunds, promotions, and other known variable costs. The amount left is contribution before advertising and before any fixed costs excluded from this calculation. Dividing it by revenue gives you a break-even ACOS on that basis. Estimating an affordable click cost also requires a conversion assumption. The same product can support a different click cost when its conversion rate changes.
Write down the assumptions. If your refund rate or fees move, your advertising limit moves with them. For a fuller explanation of targets and margin assumptions, see what a good ACOS on Amazon means.
2. Check the blockers outside the ad console
Alfredo's profitability audit identified four problems that can make healthy-looking campaigns wasteful: buy box loss, removal-order fee leakage, review changes, and fragmented P&L data. These are not bid settings, but they change the result of every click.
- Featured Offer (Buy Box): check your offer eligibility and availability before diagnosing low delivery. Amazon’s Sponsored Products guide explains that ads will not display when the product is out of stock or is not presenting the Featured Offer. Do not assume every ad format handles competing offers in the same way.
- Reviews: a rating drop can reduce conversion, which raises effective advertising cost even when bids stay the same.
- P&L inputs: advertising data, inventory data, fees, and refunds often live in different places. Join them before you judge a SKU.
- Inventory and fulfillment: availability, damage, and reimbursement issues can change the outcome that the advertising report cannot explain.
Do this check before cutting bids on a product that cannot win the offer or before raising bids on a listing whose conversion has just weakened.
3. Separate protection from growth
In a transcript describing an account audit, a campaign appeared to have excellent ROAS because most of its spend went to the brand's own name. A non-branded term with substantial search demand received almost no investment. The dashboard looked good while the account remained hard to discover outside its existing audience.
Split branded and non-branded search terms in your search-term report. Branded campaigns protect demand that already exists. Non-branded campaigns try to earn new demand. They need different budgets, targets, and questions. If you blend them, branded conversions can hide weak growth performance and make a bid change look smarter than it is.
4. Read the funnel before you choose a fix
When sales fall, do not assume the ads are the cause. Use the Search Query Performance Report, where available to your account, to move through four questions:
| Question | Signals to compare | Likely next check |
|---|---|---|
| Is demand changing? | Search query volume and total query impressions | Seasonality, category demand, and market conditions |
| Are shoppers seeing you? | ASIN impressions and impression share | Budget, bids, relevance, inventory, and organic position |
| Are shoppers choosing you? | Total clicks, ASIN clicks, and click share | Main image, title, price, reviews, and offer |
| Are shoppers buying? | Total purchases, ASIN purchases, and purchase share | Detail-page conversion, variations, price, and trust |
A demand problem needs a different response from a visibility problem. A visibility problem needs a different response from a weak offer. This order keeps you from changing the listing, campaign, and budget at the same time and then losing the evidence.
To interpret attribution definitions and report differences, read Amazon Ads attribution.
5. Let campaigns discover, then add control
A lean account can use mostly auto campaigns when the structure still reflects product economics. Group products with similar roles and margins so one bid logic does not govern products that can afford different click costs. Keep seasonal or event-driven products separate when they need a different pacing and review rhythm.
Auto campaigns can collect search-term data. Review that data on a consistent rhythm. Negate terms that waste spend or describe products you do not sell. When a term shows repeat demand and enough evidence to justify control, test it in a manual campaign with its own match type, budget, and bid. Do not create a campaign because one click happened to convert. Use the data to decide when isolation will improve the decision.
The feedback loop is simple: auto discovers, you review, proven terms earn more control, and irrelevant demand is negated. Automation can make the process faster. It cannot choose whether the business goal is growth, maintenance, or profit.
6. Change one meaningful variable and review the result
After the diagnosis, choose the smallest change that addresses the finding. Move budget from branded defense only when the growth campaign can use it. Change a bid only when the product can afford the click. Fix the listing when click share is weak. Investigate the offer when purchase share falls after a healthy click rate.
Record the reason for the change, the date, the segment affected, and the result you expect. Review on a consistent cadence that fits the account's volume. Daily edits can create noise when the account has not produced enough new evidence. A clear log lets you reverse a change and ask why, which is the opposite of black-box optimization.
If you want an operator to run this diagnosis across your account, see Enflet's Amazon PPC management service.
From Enflet’s video library
These guides adapt the ideas from the original videos into a practical reading format.
- 4 Amazon Seller Problems Quietly Killing Your PPC Profitability
- How to Run Amazon PPC with Mostly Auto Campaigns (The Right Way)
- Amazon Auto Campaigns Are Smarter Than You Think (If You Use Them This Way)
- I Audited a $52k Amazon Account. This is Wrecking Sales.
- If Your Amazon Sales Are Dropping, Do THIS First (It Will Help!)
