Branded versus non-branded Amazon ads: stop judging them as the same job

A shopper searching your brand name is not in the same position as someone searching a broad product category. Both can be worth reaching, but the same efficiency target can hide very different commercial questions.

Branded campaigns may capture existing demand, protect a useful placement or help a shopper find the right product. Non-branded campaigns may introduce the product to someone still comparing alternatives. Neither description proves how much additional revenue the ads create.

Start by separating the jobs before evaluating the results.

Build a classification you can audit

Create three practical groups:

• Brand-led searches: your brand, product-line names and genuine variants.
• Category searches: the problem, product type, use case or attribute.
• Competitor-led searches: another brand or identifiable competing product.

Keep ambiguous terms in a review bucket. Some brand names are also common words, so a simple text match can misclassify intent.

Classify actual search terms where the report provides them, not only the campaign name. A campaign called “generic” can still attract brand-led traffic. Keep product-target/ASIN rows and inferred non-search terms separate from literal shopper queries, as explained in the search-term review guide. Record what you cannot observe rather than pretending classification is complete.

Compare each group against its purpose

For brand-led activity, ask:

• Are we helping shoppers reach the intended product?
• Is there a credible competitive or assortment reason for the placement?
• How much spend is required to maintain the chosen presence?
• What happens to total brand sales and contribution when exposure changes?

For category activity, ask:

• Does the offer fit the shopper's intent?
• Are we paying a sensible amount to reach that demand?
• Which queries produce contribution rather than merely attributed revenue?
• Does the product have the stock and economics to support expansion?

For competitor-led activity, add a harder relevance test. A competing brand's popularity does not mean its shoppers want your product at your price.

Why blended ACoS can flatter the account

Here is a hypothetical example:

• Brand-led ads: $2,000 spend and $20,000 attributed sales.
• Category ads: $8,000 spend and $20,000 attributed sales.
• Combined: $10,000 spend and $40,000 attributed sales.

Blended ACoS is 25%. The underlying groups are at 10% and 40%.

The blended result is mathematically correct. It is strategically incomplete. Moving more spend toward existing brand demand could improve the headline without expanding the business.

The opposite mistake is declaring all brand advertising wasteful. That conclusion also needs evidence. Your marketplace, competitive environment and product range matter.

New-to-brand is useful, but it answers a narrower question

Amazon classifies a purchase as new-to-brand when the shopper has not purchased from that brand on Amazon during the previous 12 months. Metric availability and attribution details depend on the reporting product. Amazon's metric definitions.

That does not establish that the ad caused the purchase. A shopper can be new to the brand and already intend to buy because of a recommendation, an external campaign or a prior experience outside Amazon.

Use new-to-brand data as an additional view of the audience, not as a substitute for an incrementality test.

Test the question without pretending the test is perfect

If brand spend is material, agree a bounded experiment. Specify the campaigns, budget change, observation period and business metrics before starting.

Keep major confounders visible: promotions, price, stock, competitor activity and changes elsewhere in the media plan. Avoid changing several of them at once.

Compare total brand revenue and contribution as well as ad-attributed sales. If attributed sales fall after reducing ads, that alone does not tell you whether the business lost the same amount of revenue.

A before-and-after comparison is directional evidence, not a randomized causal result. Where a suitable controlled measurement method is available, evaluate its fit and requirements. If the evidence remains weak, say so and make a decision proportionate to the uncertainty.

Put the distinction into the report

The weekly view should show spend, attributed sales and the agreed economic measures by intent group, followed by the account total. Add a note when classification coverage is incomplete.

Then discuss the next decision: maintain, reduce, expand or investigate. This makes it much harder for a flattering blended average to replace actual management.

Use the budget allocation framework to compare competing uses of spend, and the profitability guide to avoid confusing attributed revenue with contribution.

If you want a second view of the account's demand mix, request an account review.

‍