A £20,000 monthly ad budget can look healthy in a platform dashboard while quietly destroying profit. Revenue may be rising, clicks may be cheap, and your agency may be reporting impressive conversion volume. None of that matters if the margin left after advertising, fulfilment, discounts and returns is too thin. The most effective ways to improve ROAS start with a commercial baseline, not another campaign setting.
For established eCommerce brands, ROAS is not simply revenue divided by spend. It is a decision-making tool. Used properly, it tells you where to protect budget, where to push harder, and which products or audiences are consuming spend without earning their place.
1. Set a ROAS target based on contribution margin
A blanket target of 4x or 5x ROAS is lazy management. The right target varies by product, category and commercial objective.
Start with gross margin, then account for shipping, payment fees, pick-and-pack costs, returns, discounting and any other variable costs. The result is your contribution margin before ad spend. From there, calculate the maximum cost of sale you can afford and turn it into a break-even ROAS.
For example, a product with a 50% contribution margin has a theoretical break-even ROAS of 2x before fixed overheads. That does not mean 2x is a sensible target. It means anything below it loses money at order level. A brand seeking room for overheads and profit may need 3.5x or 4x. A repeat-purchase business may accept a lower first-order return if it can prove strong customer lifetime value.
This distinction matters. Without it, bidding decisions are based on arbitrary platform targets rather than actual profitability.
2. Stop treating every product as equally valuable
Most eCommerce catalogues contain winners, steady sellers, seasonal lines and products that should not receive paid traffic at all. Sending them into the same campaign with the same bidding logic blurs performance and gives automated bidding poor signals.
Segment products around commercial reality. High-margin bestsellers deserve their own budget control. Products with weak margins, frequent returns or limited stock need tighter targets or exclusion. Clearance products may justify aggressive spend if cash recovery matters more than margin. New products may need a controlled testing budget before they earn scale.
This is particularly important in Google Shopping and Performance Max, where product-level performance can be hidden by a respectable account-wide number. A 5x blended ROAS can conceal a small group of products burning thousands at 1.5x.
3. Fix the product feed before increasing spend
Google cannot sell what it does not understand. A thin, inaccurate or generic feed limits relevance before bidding even begins.
Product titles should include the terms real buyers use to search, alongside essential differentiators such as brand, product type, material, size, colour, gender or compatible model where relevant. Descriptions should clarify the offer rather than repeat vague manufacturer copy. Accurate Google product categories, product types, images, availability and pricing are equally non-negotiable.
Custom labels add another layer of control. Label products by margin, price band, bestseller status, seasonality, stock level or return rate. This allows campaigns and targets to reflect the economics of the catalogue rather than a one-size-fits-all approach.
Feed work is not cosmetic. Better data can improve query matching, click quality and campaign control at the same time.
4. Remove wasted search demand with intent controls
Not every click from a relevant-looking keyword is commercially useful. Broad queries can pull in research traffic, bargain hunters, job seekers, DIY audiences or shoppers looking for products you do not sell.
Review search terms regularly, but do not make exclusions blindly. A query with low direct ROAS may still introduce a valuable category or reveal language worth incorporating into titles and landing pages. The question is whether it has a credible path to profitable conversion, not whether it produced a sale yesterday.
Negative keywords, tighter campaign themes and more deliberate use of match types can reduce obvious waste. On Shopping and Performance Max, feed structure, product segmentation and account-level negatives where available play a bigger role. The aim is simple: pay for commercial intent, not just relevance.
5. Improve the landing page, not just the advert
An advert can earn the click. Your product page has to earn the order. If conversion rate is weak, raising bids usually magnifies the problem.
Check the message match first. A shopper clicking an advert for a specific size, bundle or offer should arrive at the exact product and see that proposition immediately. Make price, delivery information, returns, payment options and stock status easy to find. For higher-consideration products, clear comparison points, reviews and product detail can remove hesitation.
Mobile deserves separate scrutiny. Many eCommerce accounts buy the majority of traffic on mobile while product pages are still designed around desktop browsing. Slow image loads, awkward variant selection and an overcomplicated checkout are expensive when every paid visit has a cost.
6. Use first-party data to separate prospects from customers
Remarketing can improve ROAS, but it can also flatter reporting. If existing customers were already likely to return, giving all the credit to an advert creates a false sense of efficiency.
Build meaningful audience groups where consent and platform rules allow: recent purchasers, high-value customers, repeat buyers, product viewers, basket abandoners and lapsed customers. Then tailor exclusions, messaging and bids accordingly. There is little value in paying acquisition-level costs to sell a standard replenishment order to someone who would have purchased anyway.
For Meta, customer lists and value-based audiences can help algorithms identify more commercially valuable prospects. For Google, customer data and conversion values can support smarter bidding. Neither replaces sound campaign structure, but both improve the quality of the signals platforms receive.
7. Audit conversion tracking before trusting ROAS
ROAS is only as accurate as the revenue and spend behind it. Duplicate purchase events, missing consent-mode data, incorrect transaction values, VAT mismatches and untracked refunds can all make a dashboard look better than the bank account.
Check that each order is counted once, purchase values include the correct currency and shipping treatment, and tracking persists across payment providers. Compare platform-reported sales with your ecommerce platform over a meaningful period. They will not match perfectly because attribution models differ, but major gaps need an explanation.
Where possible, send profit-aware conversion values rather than raw revenue. If one product generates £200 revenue at a 20% margin and another produces £150 at a 70% margin, treating both transactions as equal revenue signals encourages the wrong optimisation.
8. Scale budgets gradually and protect efficiency
When a campaign performs well, the instinct is to double its budget. That often forces the platform to find less qualified demand, pushing cost per acquisition up and ROAS down.
Increase budgets in measured steps, then allow enough time and conversion volume to judge the result. The right pace depends on spend level, purchase cycle, attribution delay and campaign type. A high-volume brand may assess changes within days; a business selling considered products may need longer.
Scaling also means broadening intelligently. Test new product groups, new creative angles, additional markets or higher-funnel prospecting separately where possible. Do not dilute a profitable core campaign simply to make a graph rise faster.
9. Make creative prove its commercial value
On Meta in particular, creative is a targeting mechanism. Product demonstrations, social proof, problem-solution angles, founder-led content and offer-led assets can attract very different buyers.
Judge creative beyond click-through rate. The advert with the highest CTR may be attracting curiosity rather than customers. Look at conversion rate, average order value, new-customer rate and blended profitability. A creative that costs more per click but drives larger, lower-return orders may be the better commercial asset.
Refresh creative before frequency and fatigue erode results, but do not replace everything at once. Keep a control asset running so you can distinguish genuine improvement from normal performance fluctuation.
10. Review blended performance alongside platform ROAS
Platform attribution is useful, but every platform tends to claim credit. Google may report a strong return while Meta reports another strong return, yet total business revenue has barely moved. That is why platform ROAS must sit beside blended metrics.
Track total paid media spend against total revenue, new-customer revenue, contribution margin and marketing efficiency ratio. Review performance by channel, but make budget decisions with the wider business outcome in view. If increasing Meta spend lifts branded search and total revenue profitably, last-click reporting may understate Meta’s role. If platform sales rise but total revenue does not, you may be paying to capture demand you already owned.
The best ways to improve ROAS are rarely glamorous. They come from knowing what you can afford to pay, giving each product the right level of attention, fixing bad data and refusing to scale waste. For brands already investing meaningful budget, that discipline is often the difference between advertising that looks busy and advertising that compounds profitable growth.