A Sussex homeware seller we'll come back to in a moment was spending £2,000 a month on Google Shopping and looking at £8,000 in tracked sales. On paper that's a 4x return on ad spend, the kind of number most agencies would put in a case study. The seller was also quietly losing money every single month, and didn't work that out until three months in.
That gap between what the dashboard says and what actually lands in the bank account is where most small UK ecommerce sellers come unstuck with PPC. This article assumes you're past the traffic problem: you've got a working store, a Google Shopping feed or an Amazon listing, and clicks are coming in. What we're fixing here is profit, not visibility.
The Traffic Trap: Why Clicks Are Killing Your Ecommerce Margins
Most sellers judge a PPC campaign by the numbers Google and Amazon put in front of them: clicks, impressions, conversions, 'sales generated'. Those numbers feel like success because they're easy to read and they go up. None of them tell you whether you made money.
Take that Sussex homeware seller again. £2,000 spent, £8,000 in tracked Shopping sales, a headline ROAS of 4x. Sounds healthy. But that £8,000 includes 20% VAT, so roughly £1,333 of it was never the seller's money to begin with. Strip out cost of goods, card processing fees, and postage on products that are heavy and awkward to ship (think lamps, throws, ceramics), and the real margin left before you even pay for the ads is thin. Once the £2,000 ad spend comes out of that thin margin, the seller was in the red, despite a ROAS that looked like a win.
ROAS is simply revenue divided by ad spend. Spend £1,000, generate £4,000 in sales, that's a 4x ROAS. It's a useful number for comparing campaigns against each other, but it says nothing about profitability on its own, because it never asks what that revenue cost you to produce. Break-even ROAS is the number that matters: the minimum ratio you need to hit before a single pound of profit exists. Anything below it, and every 'successful' campaign is quietly draining your bank account.
Every decision covered in the rest of this guide, campaign structure, bid strategy, pricing, search term audits, only makes sense once you know your break-even ROAS. Get that number wrong, or skip it altogether, and you can optimise a campaign brilliantly and still go backwards.
Work Out Your Real Break-Even ROAS Before You Spend a Penny
The formula looks intimidating written out, but it's just accounting for everything that eats into a sale before ad spend gets a look-in:
Break-even ROAS = 1 ÷ (gross margin − VAT impact − marketplace/referral fees − fulfilment costs)
All of those figures need to be expressed as a percentage of the VAT-inclusive selling price, because that's the number Google and Amazon use when they calculate your ROAS or ACoS. Work through a real product and it stops being abstract.
A worked example
Say you sell a kitchen gadget on Amazon for £40 including VAT.
- VAT element: £6.67
- Cost of goods (COGS): £12
- Amazon referral fee at 15%: £6
- Pick, pack and postage: £4
Add those up and £28.67 has gone before you've spent a penny on advertising. That leaves £11.33 of true margin, which is roughly 28% of the VAT-exclusive revenue (£33.33). Plug that margin figure into the formula and you get a break-even ROAS of about 3.5. Below that, every ad-driven sale is costing you more than it earns.
The VAT trap, spelled out
This is the part that catches out even sellers who think they've done their sums properly. Google Ads and Amazon report revenue, ROAS and ACoS using the full VAT-inclusive sale price, because that's what the customer paid. Your actual profit, though, has to be calculated on the VAT-exclusive figure, because a fifth of that revenue was never yours; it's owed to HMRC. The practical effect is that most VAT-registered UK sellers are overstating their real ROAS by somewhere around 17-20% without realising it. A dashboard showing 4x might genuinely be closer to 3.3x once VAT is stripped out, and if your break-even sits at 3.5x, that 4x you were celebrating is actually a loss.
Once you know your break-even ROAS, don't just aim to scrape past it. Build in headroom of 20-30% above break-even as your actual target. That buffer covers returns (which run high in categories like clothing and homeware), promotional discounting, and the seasonal dips where cost-per-click rises but conversion rate doesn't keep pace, Black Friday and the January sales being the obvious UK examples.
The returns problem most ROAS calculations ignore
Returns are the line most break-even calculations skip entirely, and in categories like clothing, footwear and homeware they can quietly wipe out a campaign that looks profitable on paper. Say that same £40 kitchen gadget has a return rate of 8%, which is fairly typical for small kitchen appliances sold online. For every 100 units sold through PPC, 8 come back. You've paid the ad cost to acquire that sale, you refund the VAT-inclusive price, you often eat the return postage, and in a lot of cases the returned item can't be resold at full price, if it can be resold at all. Build an 8% return rate into the worked example above and the true margin per completed sale drops from £11.33 to somewhere closer to £9.80 once you spread the cost of the returned units across the 92 that stick. That's not a rounding error, it moves your break-even ROAS from roughly 3.5 up towards 4. A campaign hitting 3.7x looks comfortably profitable until you factor in returns, at which point it's marginal at best.
The fix isn't complicated, it's just a step most sellers skip: pull your actual return rate by SKU from your store or marketplace reporting (Shopify, Amazon Seller Central and most order management systems all report this) and build it into the break-even formula as its own line, not folded vaguely into "margin". A product with a 2% return rate and a product with a 15% return rate should never be treated as having the same break-even ROAS, even if their gross margin percentage looks identical on a spreadsheet.
Using price as a lever, not just bids
UK sellers tend to treat the selling price as fixed and the ad account as the only dial they can turn, but price is very often the faster fix. If your break-even ROAS on a product sits at 4x and you're currently running at 3.2x, you don't have to solve that entirely through bid strategy and negative keywords. Raising the price by even 5-8% on a product with genuinely low price sensitivity (anything reasonably unique, hard to compare directly on Google Shopping, or where you're not fighting three other sellers on the exact same listing) can move the break-even ROAS target down by a full point or more, because a bigger slice of each sale is margin before the ad even gets clicked.
This matters more on Amazon than most sellers expect, because of the Buy Box. If you're not winning the Buy Box, Sponsored Products spend is largely wasted, because the ad can drive the click but a competitor's listing gets the sale. Cutting price to chase Buy Box share and then running PPC hard on top of that price cut is a common way UK Amazon sellers end up with strong sales figures and a shrinking bank balance, because they're paying to acquire customers at a margin that was already thin before the ad spend went on.
Free shipping thresholds are the other lever worth testing deliberately rather than copying from competitors. A threshold set too low (free shipping on anything over £20, say) trains customers to expect it on orders that don't actually cover your postage cost once you strip out VAT and packaging. Testing the threshold upward in small steps, £25 to £35 to £45, while watching conversion rate and average order value together, usually finds a point where basket size increases enough to absorb the postage cost without meaningfully denting conversion. That's a pricing decision, not an ad account decision, but it changes the break-even ROAS on every single campaign sitting underneath it.
The other mistake worth naming directly: doing this calculation once, at account level, and assuming it applies across the board. It doesn't. A £15 phone accessory with a 60% margin can happily run at a much lower break-even ROAS than a £300 appliance with wafer-thin margins and expensive courier costs. If you're only tracking one blended ROAS figure for your whole account, you are almost certainly overspending on some products and underspending on others without knowing which is which. This is exactly why the campaign structure has to follow product-level economics, not the shape of your catalogue.
Structuring Google Shopping Campaigns Around Profit Tiers, Not Your Whole Catalogue
Once you've got break-even ROAS worked out per product, the obvious next step is to stop treating your Shopping feed as one undifferentiated block. Google Merchant Center lets you add custom labels to your feed, and this is where the profit tiering actually gets built.
Set up something like custom_label_4 with three values: hero, mid, and longtail. Hero products are your high-margin, high-velocity sellers, the ones that comfortably clear break-even ROAS with room to spare. Mid-tier covers products that are profitable but tighter. Long-tail and testers are new or unproven lines where you're gathering data rather than expecting a return yet.
Build separate Shopping campaigns around those labels rather than lumping everything into one:
- Hero products get Target ROAS bidding with a generous budget, because you know they can absorb the spend and still return a profit.
- Mid-tier products get a lower, more protective Target ROAS target, so Google isn't tempted to chase volume at the expense of margin.
- Long-tail or tester products get manual CPC or a small fixed daily budget, purely to collect conversion data cheaply, not to scale.
Standard Shopping campaigns also let you set a priority (low, medium or high). This matters a lot once you're running standard Shopping campaigns alongside Performance Max, because Google will otherwise let whichever campaign it prefers pull budget away from the other. Setting your hero-tier standard Shopping campaign to high priority, and letting Performance Max run at a lower priority for broader reach, stops your low-margin SKUs quietly draining the budget that should be protecting your best performers.
This is exactly why dumping an entire catalogue into a single Performance Max campaign is one of the most common ways UK sellers waste money. Performance Max is built to chase volume and conversions across the account, with limited visibility into which SKUs it's actually spending on. Left unchecked, it will happily bid up branded search terms and low-margin products that would very likely have converted anyway, without any ad spend at all, while starving your genuinely profitable long-tail products of the budget needed to prove themselves.
Branded traffic is a particular blind spot inside Performance Max, because brand searches convert at a high rate and make the whole campaign's ROAS look excellent, while masking the fact that you're paying for clicks you'd have got for free organically. The fix is to exclude brand terms from your Performance Max asset groups, or apply a brand exclusion list, and run branded search as its own separate, tightly controlled campaign. That way your blended ROAS figure reflects what Performance Max is actually doing for genuinely new, non-branded demand, rather than being propped up by customers who typed your business name into Google anyway.
Auditing search terms before you scale budget
Standard Shopping campaigns don't let you target keywords directly, Google matches your feed to search queries automatically, which means the search terms report is the only place you can see what's actually triggering your ads and spending your budget. Pull it monthly, not annually. A homeware seller running Shopping ads for "ceramic table lamps" will, without fail, also show up for searches like "cheap table lamps" or "table lamp repair parts", neither of which converts at a rate that covers the click cost, but both of which will happily eat budget every day if left alone.
The practical process is: export the search terms report, sort by spend, and work down from the top. Any search term with more than, say, 15-20 clicks and zero conversions in a 30 day window is a candidate for a negative keyword. Add it as an exact-match negative at the campaign level, not the account level, because a term that's irrelevant for your lamps campaign might be exactly what you want to show up for on a separate accessories campaign. Do this consistently and most sellers find that somewhere between 10% and 20% of their Shopping spend has been going to search terms that were never going to convert, money that can be redirected straight into the hero-tier campaigns that are actually clearing break-even ROAS.
A Worthing-based tool and hardware supplier we'd use as a template here had a 400-SKU feed treated as one undifferentiated mass inside a single Shopping setup. Their self-declared 'best seller', a well-known cordless power drill, was returning a genuine ROAS of 1.8x once VAT and fulfilment costs were factored in, comfortably below break-even. Meanwhile a niche fixing kit buried on page three of the catalogue was quietly returning 6x, but had almost no budget behind it because nobody had ever looked at it. Splitting the feed into hero, mid and long-tail tiers and reallocating budget away from the drill and toward the fixing kit and a handful of similar niche products didn't just improve the account's blended ROAS, it changed which products the business actually wanted to keep stocking.
Amazon Ads: Structuring Sponsored Products Around ACoS, Not Rank
Everything above works because Google Shopping and Amazon Sponsored Products are, underneath the different dashboards, solving the same problem: you're paying for clicks and need to know which ones are actually profitable. But Amazon has its own quirks, its own fee structure, and its own version of the trap that caught the Sussex homeware seller, and treating an Amazon account the way you'd treat a Google Shopping account is where a lot of UK sellers lose money a second time.
Break-even ACoS is not the same calculation as break-even ROAS
ACoS (Advertising Cost of Sale) is just ROAS turned upside down, ad spend divided by sales rather than sales divided by ad spend, but the fee structure sitting underneath it is different from a standard Shopify or WooCommerce store. Amazon referral fees vary by category, typically 8-15% for most product types but can run higher for some categories like jewellery. On top of that, if you're using FBA (Fulfilled by Amazon), there's a pick and pack fee based on size and weight tier, plus monthly storage fees that increase sharply in the run up to Christmas, October through December storage rates on Amazon UK roughly double compared to the rest of the year. Fail to build the seasonal storage fee increase into your Q4 break-even ACoS calculation and a campaign that was comfortably profitable in September can quietly turn unprofitable in November, with the ACoS dashboard showing no change at all, because storage fees don't show up in the advertising report, they show up separately in your seller account statement.
The formula mirrors the Google version: break-even ACoS = gross margin, expressed as a percentage of VAT-inclusive selling price, after referral fee, fulfilment fee, storage cost and VAT impact are all stripped out. Run that £40 kitchen gadget example through Amazon's fee structure rather than a generic marketplace fee and the number moves around enough that copying a break-even ROAS figure from your Google Shopping account straight over to Amazon, assuming it holds, is a reliable way to either overspend or leave profitable keywords underfunded.
Separate exact match from discovery, the same way you'd separate hero from long-tail
Amazon Sponsored Products campaigns should be split the same way the Google Shopping feed gets split into tiers, just using match type instead of custom labels. An exact match campaign, built around the keywords you already know convert (pulled from your search term report or Brand Analytics), gets a higher bid and a protective ACoS target, because you already have the data to know it works. A separate broad or phrase match discovery campaign, running on a modest daily budget, exists purely to surface new search terms you haven't thought of yet, not to drive volume.
The mistake that mirrors the Performance Max problem on Google is running everything through Amazon's automatic targeting or a single broad-match campaign and letting Amazon's algorithm decide where the budget goes. It will, predictably, spend heavily on your own brand name (which converts at a very high rate and makes ACoS look excellent) and on close variations of your best-selling product, while giving almost nothing to keywords that would have opened up genuinely new demand. Run a branded campaign separately, with its own tightly controlled budget, for the same reason you'd exclude brand terms from Performance Max: so the ACoS on your non-branded campaigns reflects real new customer acquisition, not customers who already knew your product name.
Harvesting negative keywords before they bleed your ACoS
Amazon's Search Term Report (or the Search Query Performance report inside Brand Analytics, if you have brand registry) works the same way as Google's search terms report, and needs the same monthly discipline. A seller running Sponsored Products for a stainless steel water bottle will inevitably pick up clicks for "water bottle with straw" or "kids water bottle", close enough to trigger a broad match but not what the listing actually is, and the conversion rate on those mismatched terms tends to sit near zero. Harvest them into negative exact match entries at the campaign level, on a set schedule, monthly at minimum, weekly if the account is spending more than a few hundred pounds a day. Skipping this step is one of the most common reasons a Sponsored Products account that looked fine at launch quietly drifts from a 22% ACoS to a 40% ACoS over six months, with nobody noticing because the account is only ever glanced at, not audited.
TACoS: the number that tells you whether PPC is growing the business or just moving spend around
ACoS only measures the campaigns you're actively running. TACoS (Total Advertising Cost of Sale, total ad spend divided by total sales across the whole listing, not just the PPC-attributed portion) answers a different and arguably more important question: is advertising growing your overall sales, or is it just buying back sales you'd have got organically anyway? A listing with a healthy 15% ACoS but a TACoS that hasn't moved in six months is a warning sign, it suggests the PPC spend is largely cannibalising organic ranking rather than adding incremental revenue. Tracking TACoS alongside ACoS, monthly, is the Amazon equivalent of the branded-traffic exclusion problem on Google Shopping: both are ways of checking whether the headline number is measuring genuine growth or just measuring spend that would have converted anyway.
