I’m Onur Demirpençe. I spent fifteen years writing ads for the world’s biggest brands: Coca-Cola, Mastercard, Vodafone, Starbucks. Now I engineer the growth behind them.

The cleverness problem

When a campaign burns its budget by noon and sales are flat, the first instinct is more budget. The second instinct, the one that feels smarter, is a tactic. Add negatives. Cut the bottom placement. Try dayparting, because nobody is really buying at four in the morning.

There is a whole genre of Amazon advice built on that second instinct. Someone finds a lever, pulls it, screenshots the before and after, and calls it a system. The lever usually does something. It rarely does what the account needed, and the screenshot rarely survives anyone doing the arithmetic.

None of it is diagnosis. A campaign hitting its budget is a symptom.

High ACoS might be a PPC problem. The same reading also comes from a weak main image, the wrong price, thin reviews, a listing that fails to explain the product, irrelevant traffic, or a product that attracts customers with expectations it cannot meet. More spend does not fix any of that. It sends more traffic into it.

The reverse trap is quieter. A low ACoS does not prove the business is healthy. Amazon has never published a universal good ACoS, for the obvious reason that it depends on your margin, your lifecycle stage and what the campaign is for.

So an audit is not a hunt for wasted spend. Wasted spend is the easiest thing in the account to find and the least likely reason growth stopped. An audit is a search for the constraint.

1. Can Amazon put you in the right fight?

Before you ask whether shoppers click, ask whether Amazon can put the product where the demand is.

That gets decided below the title, in the part of a listing nobody screenshots. Product type, browse node, required attributes, backend terms. File a javelin in the wrong node and Amazon does exactly what you told it to do, which is hide it. That is a filing problem, not an advertising one, and no bid in history has fixed a filing problem.

One warning before you spend a dollar chasing demand. The search term report only contains terms that already produced a click. It proves you can buy that demand. It says nothing about whether you own it.

2. Do shoppers pick you?

The product is showing. Nobody chose it.

CTR tells you that, but only against the right comparison. Top of search is not a detail page. A supplement is not a face cream. Sponsored Brands is not Sponsored Display. Which is why “a good Amazon CTR is 0.4 percent” is worth what you paid for it. Amazon Ads publishes benchmarks by category, ad product, format and goal. Use those.

If the number is weak against the right benchmark, raising bids buys more impressions from people who were already looking past you.

Everything that decides a click happens before the click. Main image, title, price, star rating, review count, and whether the product answers the question in the search box. In the javelin account the pre-click problem was price, and price arrives in the console dressed as a bidding problem.

3. Does the page close the click?

They clicked. They left.

Search Query Performance is the only report that shows the whole walk in one line: impressions, clicks, cart adds, purchases, query by query. Read it left to right and it tells you where the shopper stopped. Plenty of impressions and almost no clicks sends you back a section. Clicks that die at cart add send you into the page.

Then the work is the page itself. Images, A+ content, price, reviews, variation structure, and whether any of it answers the objection the shopper walked in carrying. Most detail pages describe a product. Very few argue with the reason not to buy it.

One rule while you are in there. Do not add SQP figures to Ads Console figures. Two instruments, same street, different windows, different clocks. The sum of them is not a number.

4. Are you buying the right customer?

No listing makes every visitor valuable. This is the first question that genuinely lives inside the ad account.

Auto targeting is there to find queries you would never have thought of. Manual is there to keep the ones worth keeping. So the search term report is not a scoreboard. It is a guest list, and it tells you exactly who your money invited.

Three things to look for. Terms that eat spend and return nothing. Discoveries that have earned controlled targeting. And terms that convert while describing a need your product does not serve.

The third one is the expensive one. Search evidence proves demand. It does not prove product truth. A term can convert and still carry the wrong expectation, and rewriting a listing to chase it is how a product starts lying about itself, one bullet at a time.

And stop negating so early. Amazon’s own threshold sits around twenty clicks. A keyword with four clicks and no sales has not failed. It has not spoken yet. But most importantly, can your catalog cover for those clicks?

5. Did the product keep the promise?

Most audits stop at the purchase. That is one step too early. A conversion means somebody bought. It does not mean the sale was any good.

Return reasons are the honest metric, written in plain language by the only person who actually held the product. Not as described. Wrong size. Not what I expected. Read them in a run and they point at the listing far more often than at the warehouse. If shoppers keep buying and sending it back, the campaign produced a conversion and the business paid for the wrong customer twice, once in ad spend and once in return cost.

One detail, because most of the internet has it wrong. Amazon Ads does remove purchases from attribution when an order is cancelled, returned or declined within 72 hours, then restates the data. Ad reporting does not ignore returns. It also does not replace a returns and profitability analysis. A sale can be a clean conversion and a broken promise at the same time.

6. Can this sale afford to be bought?

Now ACoS is worth arguing about.

ACoS is spend divided by ad-attributed sales. It measures one thing, which is how efficiently you bought that revenue. Profit lives somewhere else entirely: price, COGS, fees, refund rate, the rest of the variable cost, and the margin the business has to keep. Those numbers give you a break-even and a target. Without them, a target ACoS is a preference with a decimal point in it.

TACoS shows the other half of the picture, spend against total revenue. It is how you catch a brand quietly turning into a paid traffic habit, where the sales hold up while the organic share of them keeps shrinking. Read it with lifecycle context or do not read it at all.

The question was never whether an ACoS is good. It is what this product can afford to pay for a customer it keeps. In the javelin account, at the old prices, the honest answer was almost nothing. Nothing you can do inside a campaign survives that.

7. Now spend more

Seven things have to be true at the same time. The listing is eligible for the demand you want. Shoppers choose it. Clicks convert. The traffic brings the right customer. Returns reveal no broken expectation. The margin can pay for acquisition. Inventory can take the volume.

Then a campaign running out of budget at noon is a real constraint on a healthy business, and more money is the correct answer. It is almost never the first one.

The order is the method:

1. Get filed correctly before you buy traffic.
2. Earn the click before you buy more impressions.
3. Close the click before you buy more clicks.
4. Buy the right customer before you buy more volume.
5. Keep the promise before you call the sale a win.
6. Know the margin before you call the growth profitable.
7. Scale last.

The ad account is not where the problem starts. It is where the problem becomes measurable. Almost everything that limits an Amazon business sits upstream of the console: the product, the price, the page, the promise, the margin. The console only keeps the receipts.

The javelins needed better campaigns, and they got them. They also needed a reason for $250 to look like the sensible choice, and only one of those two things was ever visible in the Ads Console.

Advertising is an art. Its job is to move somebody to act, and at its best it does that with enough craft that they are glad afterwards. Talking someone into a product that cannot keep its promise is not advertising. It has another name. Online, Amazon is where most of that now gets decided.

So the work is not to find new ways to give Amazon money. It is to build something that deserves more of yours.