I've held Total ACOS at 11.1% and net margin at 24.9% for twelve straight months across UK, US, Canada and the EU on combined Amazon revenue of £225,716, by treating the account as a data problem rather than a marketing one.

Decision · PPC efficiency

Splitting defensive and discovery campaigns, and sweeping negatives every month

The situation. When you let Amazon Ads aggregate everything into a single "ACOS" number, you can't tell whether you're defending a hero ASIN at a healthy 5% or burning money on long-tail discovery terms at 80%. Both end up in the same average. On a small-to-mid Amazon account that one piece of laziness is often the difference between a 22% net margin year and a 12% one.

The data. The monthly Search Term Report and the campaign-level views in Seller Central, broken out by ad type, Sponsored Products, Sponsored Brands, Sponsored Brands Video and Sponsored Display each have very different cost structures and different jobs to do.

The decision. Defensive campaigns are kept tight, targeted at the hero ASIN and own-brand search terms, and held to a very low ACOS. Discovery campaigns are allowed to run hotter while a new keyword proves its conversion rate, with anything underperforming the contribution-margin threshold negated at the monthly sweep.

The result. Over the last twelve months the account has spent £25,024 on advertising against £225,716 of sales, for a Total ACOS of 11.1%. More importantly, the account has been profit-positive every single month, even in the seasonally soft months.

A note on attribution. The headline 11.1% Total ACOS divides total ad spend by total revenue across the whole account, which is the honest denominator but quietly understates how much of that revenue ads actually produced. The on-screen sponsored figure inside Amazon Ads uses same-day attribution, a conservative window that only credits a sale to PPC if the click and the order happen inside one calendar day. On that definition sponsored looks like roughly 20% of revenue, with the remaining ~80% booked as organic. In reality a meaningful share of that "organic" revenue is halo: customers who first discovered the listing via a sponsored placement and came back days later to buy. Re-run the same data on a 7-day or 14-day window and the organic/sponsored split moves materially closer to 50/50. I default to same-day in this case study because it under-claims rather than over-claims the contribution of ad spend; an interview conversation is the right place to walk through the multi-touch view.

Decision · Inventory policy

Holding roughly three months of cover, no stockouts, almost no long-term storage fees

The situation. Inventory policy on Amazon is a knife-edge. Carry too little and a hero SKU goes out of stock; the Best Seller Rank drops within days and takes weeks of paid traffic to rebuild. Carry too much and Amazon's Long-Term Storage Fees start chewing through margin on units that haven't even sold yet, and cash is tied up in stock instead of going into ads or new product development.

The data. Sellerboard's demand forecast combined with each SKU's lead time and a safety-stock buffer gives a target reorder point. The FBA inventory age-bucket report is reviewed weekly to catch anything drifting into the 181+ day band before it triggers a fee.

The decision. Target a band of roughly sixty to one-hundred-and-twenty days of supply on every active SKU. Reorder when a SKU enters the lower half of the band; submit a removal or run a clearance promo on anything entering the upper half with a slowing sell-through rate.

The result. 30 active SKUs, median 195 days of supply, only 0.42% of units aged into the older buckets. Long-Term Storage Fees over the full year came in at £113 against more than 23,743 units shipped, which is essentially a rounding error on a P&L of this size.

Decision · Listing quality & CX

Treating NCX rate, return rate and review wording as leading indicators

The situation. Star ratings are a lagging indicator. By the time a listing's average rating has moved, the damage is already done and recovery takes months. NCX rate, return rate and the specific wording of negative reviews tend to shift first, and if you're reading them every week you can usually catch a quality drift before it reaches the rating.

The data. The Listing Summary Report covers 35 SKUs in this snapshot, with NCX rate, return rate and the CX Health classification for each. The overall refund rate across the account this year held at 3.5%.

The decision. Read negative reviews every week, in full, not as aggregated sentiment. Look for patterns of three or more, that's where the real packaging, copy or expectation-setting fixes live. Any SKU dropping out of the "Excellent" CX Health band triggers a forty-eight hour review covering the listing, the packaging and the most recent ten negative reviews.

The result. The CX Health distribution across audited SKUs is shown to the right, and the account's total refund cost over the year came in at £6,152 on £225,716 of revenue.

Decision · International expansion

Running 12 Amazon storefronts as four marketplace groups, with ad spend concentrated where it earns

The situation. Selling on Amazon UK alone caps the addressable market roughly at a third of what's available in English-speaking territories and perhaps a tenth of what's available across the EU. Expanding into the EU, US and Canada multiplies the demand pool but also multiplies the operational overhead: VAT registration in each jurisdiction, EPR compliance, language localisation, separate PPC accounts, and the basic question of which markets are even worth the advertising spend.

The data. Revenue and contribution are tracked per storefront in the Sellerboard dashboard, then grouped here into four operational categories, UK, US, Canada and EU. The nine EU storefronts (Germany, France, Italy, Spain, Netherlands, Sweden, Belgium, Poland, Ireland) are reported as a single group because operationally they share inventory through Pan-EU FBA (the Amazon programme that holds inventory in one EU fulfilment centre and lets it ship cross-border to all the others at no extra fulfilment fee), a single translated listing set and a single VAT / EPR / IOSS compliance workflow.

The decision. Concentrate ad spend on the three markets that actually justify it on contribution, the United Kingdom, the United States and Canada, which together absorb around 87% of total Amazon advertising. The EU group gets a small holding budget on its larger storefronts (Germany, Italy, Spain, France) to keep the Buy Box defended and feed Pan-EU spillover; the smaller EU storefronts (Ireland, Netherlands, Sweden, Belgium, Poland) are kept open with translated listings but run on organic and cross-border spillover only. VAT and EPR compliance is handled in-house in every jurisdiction rather than outsourced.

The result. Twelve-month revenue is split across the four marketplace groups as shown. The UK is dominant at roughly 43% of Amazon revenue, the United States and Canada together contribute another 46%, and the EU group collectively contributes a meaningful and profitable secondary stream on a fraction of the ad spend per pound of revenue.

Where this goes next. The same playbook extends into the Spanish- and Portuguese-speaking marketplaces, Mexico, Spain and Brazil. I speak both languages (C1) and have lived and worked in Brazil and across Spain and Latin America, so listing localisation, customer messaging and marketplace-specific compliance are handled in-market rather than translated. This would be helpful for a brand or agency expanding into LatAm or Iberia.