A £20 cost per acquisition can be excellent for one ecommerce business and destructive for another. The difference is rarely the Google Ads interface. It is the margin behind the order, the likelihood of a second purchase, the cost of fulfilment, and whether GoogleAds for ecommerce is being judged on revenue alone rather than profitable growth.
For ecommerce teams with an established catalogue and real demand, the question is not whether to run Shopping, Search or Performance Max. It is whether the account can direct spend towards products, customers and queries that improve contribution after advertising costs. That requires a measurement model and account structure built for commercial decisions, not a race for the lowest cost per sale.
Start with the number that makes growth worthwhile
Revenue is visible and reassuring. It is also incomplete. A campaign can report a strong return on ad spend while concentrating budget on discounted products, low-margin lines or orders with expensive returns. Conversely, a campaign that appears expensive at first purchase may acquire customers with high repeat value.
Before changing bids or launching more campaigns, define the economic threshold each order must meet. At minimum, calculate contribution before ad spend: net product revenue minus cost of goods, payment fees, fulfilment, shipping subsidy, expected returns and any promotion cost. The remaining amount is the pool available to acquire the customer and still make a profit.
A simple working calculation is:
Allowable CPA = first-order contribution + expected future contribution – target profit
If a £120 order produces £45 of first-order contribution, and a realistic repeat-purchase contribution is £20, an allowable CPA may be materially higher than the £15 target implied by a revenue-only model. If repeat purchasing is weak, the opposite is true. Do not use an optimistic lifetime value to excuse unprofitable acquisition. Use cohort data, a conservative time window and separate new from returning customers where possible.
This is also where the target metric changes. ROAS can be useful when margins are broadly consistent. With wide margin variation, profit on ad spend, contribution after advertising cost, or a margin-weighted conversion value is more reliable. The right choice depends on the quality and speed of your data, but the principle does not: bidding should receive a signal that resembles the commercial outcome you want.
GoogleAds for ecommerce needs an intent-led structure
Campaign structure should make budget, search intent and performance diagnosable. It should not exist because a platform recommendation suggested another campaign type.
Search campaigns are usually the clearest place to control high-intent demand. Brand search, category search, product-specific queries and competitor terms behave differently. Separating them allows you to see whether apparent growth comes from capturing demand you already own or creating incremental sales. Brand terms may deserve protection, but they should not obscure the performance of non-brand acquisition.
Shopping and Performance Max can scale product discovery, particularly where feeds are complete and conversion signals are dependable. Their trade-off is reduced visibility and control at query, placement and product level. That does not make them unsuitable. It means they need stronger guardrails: clear product segmentation, accurate feed attributes, exclusions where appropriate, and regular checks against business data rather than platform reporting alone.
A practical starting structure might separate profitable hero products from lower-margin or clearance inventory, bestsellers from long-tail products, and new-customer acquisition from remarketing where measurement permits. Do not fragment campaigns simply to make the account look sophisticated. Split when products have materially different margins, conversion rates, stock positions, customer value or budget requirements.
The product feed is a commercial asset
For Shopping-led accounts, feed quality is not admin work. Titles, product types, categories, images, availability, price and promotional information influence eligibility and relevance. A vague manufacturer title may be accurate, yet fail to match the language buyers use. A feed without reliable custom labels makes it harder to manage seasonal products, margin bands, stock levels or strategic ranges.
Build labels around decisions you will actually make. For example, margin tier, bestseller status, seasonal priority, stock risk and newness can help direct reporting and budget. Keep the logic documented. If a label cannot be explained to a finance or merchandising colleague, it is unlikely to support a better advertising decision.
Fix conversion measurement before scaling spend
Most ecommerce Google Ads problems eventually become measurement problems. The platform may record a purchase, while finance sees cancellations, duplicate orders, tax differences, refunded sales or revenue assigned to another channel. Small discrepancies are normal. Unexplained, persistent discrepancies are a reason to pause scaling.
The core event should be a completed transaction with a unique order ID and transaction value. Test it across devices and payment methods. Ensure duplicate purchase events cannot inflate results when a confirmation page reloads. Then compare platform conversion data with analytics, the ecommerce platform and finance records over the same date range and attribution assumptions.
For stronger optimisation, send better values back into the account. This can include net revenue rather than gross revenue, cancellation-adjusted values, or product-margin-informed values where the operational setup supports it. The more closely the conversion value reflects realised commercial value, the less likely automated bidding is to scale the wrong sales.
Be cautious with micro-conversions such as newsletter sign-ups, add-to-baskets and checkout starts. They are useful diagnostics, especially when purchase volume is low, but they should not silently replace purchases as the primary success metric. A high volume of checkout starts can reveal friction. It cannot prove profitable demand.
Diagnose performance in the right order
When revenue stalls, teams often change bids, budgets, creative and landing pages at once. That creates activity without learning. Work from the commercial outcome back towards the mechanism.
Use this diagnostic sequence:
- Check whether net revenue, margin and customer acquisition cost meet the agreed threshold.
- Compare new versus returning customer performance, brand versus non-brand demand, and product groups with different economics.
- Review stock availability, pricing, promotions, delivery promises and returns before blaming media performance.
- Check conversion tracking against order and refund data.
- Review search terms, product-feed eligibility, campaign overlap and budget allocation.
- Only then test bidding, creative, landing-page changes or a revised campaign split.
This order matters because a falling ROAS may reflect a stock-out of the best-selling product, a sitewide discount ending, weaker conversion rate from a delivery issue, or an attribution change. None of those are solved by increasing a target ROAS.
Know when to use automation and when to constrain it
Automated bidding is valuable when conversion tracking is accurate, the account has enough relevant data and the target aligns with commercial reality. It is weaker when values are distorted, campaigns combine radically different product economics, or changes happen too frequently for the system to learn.
Give automation a stable objective, then make controlled changes. If you raise budgets, change targets, alter feed titles and introduce a promotion in the same week, you will struggle to identify what caused the result. A disciplined test has a stated hypothesis, a meaningful evaluation period and a decision rule decided before the data arrives.
The same applies to creative. Asset testing in Performance Max can improve coverage and message fit, but it cannot compensate for poor product-market fit, an uncompetitive offer or a checkout that loses buyers. Treat creative as one lever within a system, not the system itself.
Connect paid search to the rest of demand capture
Ecommerce search growth is stronger when paid and organic teams work from the same buyer language. Search-term data can expose category phrasing, comparison questions and product attributes worth addressing in category pages, buying guides and on-site navigation. Organic visibility can identify high-value topics where paid coverage is expensive or incomplete.
AI-assisted discovery adds another consideration. Product information, policies, comparisons and category positioning need to be clear enough for people and systems to interpret. That is not a substitute for Google Ads management. It is a way to reduce friction across the discovery journey and improve the commercial pages paid traffic eventually reaches.
For teams unsure whether the constraint is account structure, tracking, feed quality, landing pages or unit economics, begin with a Search Diagnostic. The useful output is not a longer task list. It is a ranked view of what is limiting qualified revenue, what can be tested quickly, and what should not receive more budget until the evidence improves.
The best next move is usually modest: establish the profitable acquisition threshold for one priority product group, validate the purchase value being reported, then decide whether additional spend has a credible route to contribution. That is how ecommerce advertising becomes a managed growth decision rather than an expensive act of faith.