Note: All business and company names in our case studies are anonymized for client privacy. All metrics, timelines, and operational details are real and independently verifiable on request.
The starting point
A regional power-tool brand had been running Amazon ads for three years without a structured keyword architecture. Their campaigns had grown organically, auto-targeting layered on top of manual campaigns, broad match covering hundreds of high-volume head terms, zero systematic negative keyword strategy. The result: $0.42 in ad spend for every dollar of attributed revenue. The brand was profitable, but not nearly as profitable as the underlying demand warranted.
When we came in, the account was generating sales, but it was leaving a significant portion of its margin on the table every month. The structure wasn’t wrong on purpose. It had just never been cleaned up.
The diagnosis
We audited every search term that had triggered an ad in the trailing 12 months: 11,000 unique queries. Fewer than 400 had ever produced a confirmed sale. The remaining 10,600+ were pure wasted spend, dollars going toward clicks that had never once converted.
The culprit was broad match on high-volume head terms. “Power drill,” “cordless drill set,” “tool kit”, these pulled in price shoppers, casual browsers, and research traffic. Meanwhile, the longer-tail, higher-intent queries where conversion rates ran 3–4× higher were systematically underfunded. The algorithm was doing exactly what it was told. The problem was what it had been told to do.
The playbook
Negative keyword architecture
Full audit of all 11,000 search terms. Every non-converting term harvested and negated across the entire campaign structure. This alone freed up budget that had been evaporating with no return.
Match type migration
Moved 40% of total budget from broad to phrase and exact match. Tighter relevance, higher conversion rate, more predictable spend. The goal wasn’t to spend less, it was to spend where buyers actually were.
DSP retargeting layer
Layered in Amazon DSP to retarget cart-abandoners and product-detail-page viewers. These shoppers had already self-selected as serious buyers, they just needed a nudge back. Cost-per-acquisition on this layer ran well below the campaign average.
The result
ACoS dropped from 42% to 18% in a single audit cycle. Unit velocity held flat, which is the part that matters. The brand didn’t spend less; it spent smarter. The same budget redirected toward actual purchase intent produced 2.3× more efficient ad spend, with no sacrifice in sales volume.
For context: across our active book, median ACoS runs around 22%. This account went from 11 points above that to 4 points below it, without touching the catalog, the creative, or the pricing.
What worked
Discipline over volume. Most agencies optimize for impression share. We optimize for purchase intent. When you stop subsidizing clicks that will never convert, the math changes fast.
The negative keyword work isn’t glamorous, it’s tedious, methodical, and invisible to anyone who doesn’t know what they’re looking at. But it’s the foundation that every other optimization sits on. You can’t build a 4x ROAS account on a leaky structure.
Running a tool or hardware brand on Amazon with ACoS above 25%?
“Clearsight is the only partner we’ve used that owns the number. When the P&L misses, they’re the first to call it — and the first to have a fix.”
Priya D. · CEO · Auto & Truck Accessories