Does Remarketing Deserve the Credit?

Two separate reasons.

1. Different attribution windows. Sponsored Products credits conversions within 7 days of a click and 14 days of a view. Sponsored Brands and Sponsored Display use 14-day click and 14-day view windows. Sponsored Display therefore gets credit for orders that occur on days 8-14 after a click - orders Sponsored Products is not allowed to claim. A campaign with a longer window will look more efficient on ACOS for reasons that have nothing to do with the ad.
2. Overlapping credit. Each campaign is credited for the orders it influenced, so a shopper who saw a Sponsored Products ad, clicked a Sponsored Brands ad, and later converted can appear in more than one campaign's numbers. Campaign-level totals are not a deduplicated ledger, which is exactly what the 1,450 versus 1,750 gap is showing you.

Practical rule: compare campaigns on the same window (pull the 7-day and 14-day columns in the report and align them), and judge the account on blended TACOS against total business sales rather than by ranking campaign ACOS. A 45% ACOS on the discovery layer that feeds the 12% remarketing layer is not automatically the worst performer - it may be the reason the other two have anyone to talk to.

The campaign structurally reaches warm demand. Its audience is defined as people who already viewed the hero ASIN in the last 30 days. A high conversion rate is expected by construction: these are shoppers in the funnel, some of whom would have returned and bought anyway. The 12% ACOS is not evidence of incrementality.

Test design:

• Holdout: pause Sponsored Display for 3-4 weeks and compare total sales of the hero ASIN, and total account sales, against the preceding period and against the year-ago period if seasonality is a concern. Budget is not reallocated to Sponsored Products during the pause, otherwise you cannot attribute the change.
• Benchmark against organic: compare Sponsored Display's conversion rate with the ASIN's baseline session conversion rate. If remarketing converts at roughly the same rate as ordinary returning traffic, it is largely buying conversions that were already coming.
• Check new-to-brand share on the Sponsored Display campaign. If almost none of its orders are new-to-brand or new-to-customer, it is harvest.

What would change my mind: if hero ASIN sales fall materially during the pause and recover when it resumes, the campaign is incremental - keep or scale it. If sales hold flat, the spend is duplicative and should be redirected to discovery or to the catalog rather than defended on ACOS.

Freeze first, then re-cut.

1. Stop the escalation. A 5-point TACOS move on 6% sales growth is a buying problem, not a reporting problem. Hold budgets flat until you know where the 40% went.
2. Re-cut Sponsored Products by search term, not by campaign ACOS. Somewhere in that 45% is a tail of terms above target and a core performing well below it. Cut or negate the tail; protect the terms that are carrying volume. Sponsored Products is usually the discovery layer, so cutting it wholesale to fund remarketing shrinks the audience the remarketing campaign depends on - a self-defeating reallocation.
3. Hold Sponsored Display flat during the holdout rather than scaling it, and evaluate Sponsored Brands on new-to-brand orders rather than on ACOS alone, since its role is acquisition.
4. Normalize the measurement: align attribution windows across campaigns, evaluate on rolling 30-day windows (the most recent week is understated by conversion lag), and report against a TACOS guardrail - for example, hold the account at or below 11% and require any budget increase to justify itself on incremental sales within four weeks.
5. Write the rule down: budget moves require an incrementality rationale, not a cross-campaign ACOS ranking.