Amazon PPC Campaign Structure: Why Your Budget Goes Where You Can't See It

One campaign had a $20 daily budget.

It was advertising 163 products.

Divided evenly, that is about twelve cents of daily budget per product, before a single click. It was never divided evenly. A handful of products absorbed nearly all of the spend, some of them spending steadily without producing orders. The rest received almost no exposure at all.

Nothing in the campaign's headline metrics made this obvious. It had spend, some sales, and an ACOS that could be explained away. The problem sat one level down, in a structure that decided where the money went before anyone adjusted a bid.

That is the real job of campaign structure. Neat accounts are not the point. Your structure determines how much control you actually have over where budget goes, what gets tested, and which performance data you can trust. When structure is loose, bids, budgets, and keyword decisions still get made. They are just made on distorted information.

Why Campaign Structure Changes Where Your Money Goes

An Amazon PPC account is a hierarchy:

Campaign (budget is set here)  →  Ad group  →  Advertised products + Targets (bids are set here)

The budget belongs to the campaign. Everything beneath it competes for that budget, and Amazon decides how it is shared. If a campaign contains three ad groups, no setting determines how much each one receives. The same is true for products within an ad group, and for targets within that.

In practice, spend flows toward whatever the auction favors: the targets with the highest bids, the broadest reach, or the most search volume. That is not necessarily what converts best for the business.

A campaign with several ad groups shows the effect clearly. One ad group usually takes the majority of the budget. The others may be starved of impressions, or quietly spending without results, while the campaign-level numbers look acceptable. The only lever for changing the split is bids, which also changes what every click costs.

The pattern repeats at every level. The more entities share one budget, the less control you have over where it is spent, and the harder it becomes to tell what is actually working.

That is why tight structures work well as defaults: one ad group per campaign, a focused set of products per ad group, a manageable number of targets, and separated match types. They are starting points that maximize control, not laws. Some campaigns are deliberately broad, such as an auto catch-all meant to give the whole catalog low-cost exposure. Some catalogs, such as large assortments of near-identical items, can justify grouping. And a legacy campaign with years of performance history may be worth protecting even if its structure is imperfect. What matters is that any departure from the default is a deliberate choice, not an accident that grew over time.

When Too Many Products Share the Same Budget

The 163-product campaign is the extreme version of a common problem.

Too many products in one ad group → Amazon distributes impressions unevenly → a few products dominate spend while others barely show → the seller concludes the quiet products do not convert, when they may never have received a real test.

Campaign-level ACOS makes this worse, because it averages everything together. A campaign can look acceptable while one product inside it wastes money and another, which might perform well, never gets the chance.

Relevance is the second problem. Targets are shared across every product in the ad group. An ad group advertising a jump rope, hand grips, and knee sleeves together cannot have targets that fit all three. Shoppers searching for one product are shown another, and conversion suffers for reasons that have nothing to do with the products themselves.

A strong default is to keep an ad group focused on one product, or one product with its FBA and FBM offers, or a very tight group of closely related items. The more products share targeting and budget, the harder it becomes to see which one is actually driving the result.

Variations deserve particular care. When several child ASINs of the same parent are advertised together, Amazon typically shows only one of them in a given sponsored placement. Advertising every color or size spreads spend across products effectively competing for the same slot. Standalone listings behave differently, which is why some brands deliberately keep products separate to occupy more space on a search results page.

The right answer still depends on the data. Consider an ad group advertising three colors of the same product. Green has 31 purchases. Red has 40. Blue has two, at a ROAS of 0.38. Against a target ROAS of 2, green and red are both earning their place, because different shoppers prefer different colors. Blue is not. Pausing it moves its spend to the variations that convert, without pretending the whole family should be cut down to one.

When Too Many Targets Compete for Attention

The same distortion happens with targets.

Too many targets in one ad group → a few high-volume targets absorb delivery → the rest receive little or no data → the seller cannot distinguish a bad keyword from a keyword Amazon never meaningfully tested.

In one account we reviewed, a product-targeting ad group contained 29 targets. Over more than a month, only one had received any impressions, and only 236 in total. The campaign was technically active and effectively dead.

The important point is what that data does not say. The 28 targets with no impressions had not failed. They had not been tried. Pausing them as poor performers, or keeping them on the assumption that they were part of a working campaign, would both be decisions made on non-information.

A useful guideline is to keep ad groups to roughly ten targets or fewer when the goal is a fair test of each one. It is not a hard limit. But beyond that point, crowding makes it likely that most targets will never generate enough data to judge.

Why Match Types Need Enough Separation to Stay Controllable

Mixing exact, phrase, and broad versions of the same keyword in one ad group creates a specific problem: the tools for controlling one match type start affecting the others.

Mixed match types → targeting control overlaps → negatives or bid changes affect more than intended → optimization becomes hard to isolate.

Suppose the broad version of a keyword starts matching irrelevant search terms. A negative keyword added to stop that traffic applies across the ad group, and can block the exact and phrase versions too. Controlling one match type without damaging the others becomes difficult, and the performance of each becomes harder to read.

Separating match types into their own ad groups or campaigns is a strong default when the goal is clean control and clean measurement. Naming should reflect the contents as well. An ad group labeled "Broad" that holds exact match keywords makes every later decision harder to interpret.

Duplicate Targeting Makes Performance Harder to Read

Accounts that grow through repeated keyword research often end up targeting the same keyword, in the same match type, for the same product, in two different campaigns.

The cost is clarity. Impressions are split between the two. Performance history is fragmented. Attribution becomes unclear, and it is difficult to know which campaign is actually carrying the term. A bid change on one may simply shift volume to the other, so the result of the change cannot be read cleanly.

The check has to be made for the specific product, not just the keyword. A keyword can appear elsewhere in the account for a different ASIN, which is not a duplicate. The question is whether the same ASIN has the same keyword, in the same match type, in more than one place.

Where true duplicates exist, compare the two and keep the one carrying the volume.

How to Fix Structure Without Breaking What Already Works

Structural problems are easy to fix when the affected campaign is not producing sales. Pause what does not work and move on.

The harder case is a crowded or mixed campaign that is performing well. Do not destroy a working campaign just because its structure is imperfect. Its performance history has value, and an abrupt restructure can reset it.

Migrate the traffic instead of cutting it:

• Build the cleaner structure for the portion being separated, typically the less efficient match type or the product being isolated.
• Let it gather traffic before touching the original.
• Shift traffic gradually, for example by bidding slightly higher in the new campaign so volume moves toward it.
• Pause the old target only after the replacement is clearly carrying the volume.

If exact match is the performer inside a mixed ad group, leave exact where it is and move phrase and broad out. The aim is to end with the same sales on a structure you can actually control.

What to Look For

To audit the structure, look for campaigns with multiple active ad groups, ad groups containing large numbers of products or targets, mixed match types, and duplicated ASIN-keyword combinations. A bulk file makes these patterns easier to spot at scale.

The Real Principle

A good campaign structure does not exist because neat accounts look better. It exists because every bid, budget, placement, and keyword decision becomes more useful when you know exactly what that decision affects.

If your Amazon PPC account is spending but you cannot say which product, keyword, or match type each dollar is meant to support, the problem may not be your bids. It may be the structure those bids are operating inside.