What This Looks Like In Practice
Label Weavers, a US manufacturer of custom woven and printed labels, name tapes and care labels, came to us with an account that had been left to an automated bidding system. It prioritised scale over control, and control was exactly what was missing: campaigns had duplicated over time, variation families were grouped inconsistently, and Sponsored Brands and Sponsored Display were barely used. The account leant almost entirely on lower-funnel spend, with nothing above it building visibility or bringing shoppers back.
We rebuilt it as a manual, segmented structure, one campaign type per job: auto and broad match for discovery, phrase match to refine intent, exact match for the terms already converting, and manual product targeting for competitor and ASIN-level control. Every legacy campaign was paused so the data behind future decisions was clean rather than inherited. Base bids came down, and top-of-search placements were prioritised against historical performance data rather than left to an algorithm's guesswork. Sponsored Brands and Sponsored Display were then built out properly for the first time: product collections around hero lines, Store Spotlight for category presence, and retargeting to recover shoppers who'd already shown intent.
ROAS improved by 35%. Revenue grew 34%. Sponsored Brands revenue, from a near-standing start, grew 382%.
Read the full Label Weavers case study.
What this means for you:
If your Amazon optimisation consists primarily of increasing and decreasing bids, you're only solving part of the problem. Bids influence efficiency, but account structure determines growth. Discovery campaigns uncover new opportunities, exact campaigns maximise proven demand, Sponsored Brands and video build awareness, and Sponsored Display recaptures shoppers who didn't buy first time. When those layers are missing, no amount of bid tweaking will deliver the growth you're looking for. Eventually, performance stagnates because you're optimising efficiency instead of building demand.