In ecommerce, you don’t really write ads — you optimise data. The product feed is the campaign, the margin is the target, and the algorithm does the bidding. Get the feed and the profit signals right and Google’s AI will scale you profitably; get them wrong and no amount of clever campaign structure will save you. That’s truer in 2026 than ever, because Google has handed almost the entire machine to automation and left you in charge of exactly two things that matter: the quality of your product data and the accuracy of your profit signals.
This guide takes you from your first Shopping campaign to scaling Performance Max into the millions while return on ad spend climbs instead of collapsing — which, done right, it can. The throughline is profit, not revenue: the businesses that win optimise to margin, not top-line ROAS. If you’d rather we apply this to your account directly, get a free audit.
This is the deep dive; for the wider 2026 picture start with the complete Google Ads guide, and for done-for-you delivery see our Google Ads for ecommerce service.
What this guide covers
- How ecommerce Google Ads actually works
- Merchant Center and product feed optimisation
- Campaign structure for ecommerce in 2026
- Winning with Performance Max
- Bidding and profitability
- Conversion tracking for ecommerce
- Remarketing and retention
- Scaling profitably as budgets grow
- Common mistakes
- FAQs
How Ecommerce Google Ads Actually Works
The mechanics are different from Search. You don’t bid on keywords and write text ads; you upload your catalogue to Google Merchant Center as a product feed, and Google matches your products to shoppers’ queries using the data in that feed — your titles, images, prices, product types and attributes. Shopping and Performance Max campaigns then serve those products across Google’s surfaces and optimise toward your conversion goal.
The practical implication is profound: your feed is your keyword strategy, your ad copy and your relevance signal, all at once. An hour spent improving product titles usually moves performance more than an hour spent inside the campaign settings.
The channels in play. Shopping puts your products directly in commercial results. Performance Max serves those same products across Search, Shopping, YouTube, Display, Discover, Gmail and Maps from a single campaign. Search (brand and non-brand) captures text queries. Remarketing and Demand Gen bring back and expand your audience. In a mature account these stack into a full funnel — but for most stores, feed-driven Shopping and Performance Max do the heavy lifting.
Merchant Center and Product Feed Optimisation
Most ecommerce underperformance is a feed problem wearing a bidding costume. This is the foundation — get it right before touching anything else.
Merchant Center health first. Verify your store, configure shipping and tax accurately, and stay on top of policy compliance — disapprovals and account suspensions are the fastest way to kill an ecommerce account overnight. In 2026 there are hard deadlines to meet: Google’s updated product data specification raised the minimum product image resolution (with a 500×500 pixel floor), added a new video attribute and product-level shipping controls, and — critically — the legacy Content API for Shopping stops functioning on 18 August 2026, making Merchant Center Next the only supported feed-management path. If you’re on an older feed setup, migrating is not optional.
The product feed is your number-one lever. Within the feed, prioritise in this order:
- Titles. The single most important match signal. Front-load the words shoppers actually search — brand, product type, key attributes (size, colour, material, model) — because Google reads titles to decide which queries you’re eligible for. “Bosch Professional GSB 18V-55 Cordless Combi Drill” beats “Great Value Drill — Free Delivery” every time.
- Images. High-quality, compliant, meeting the new resolution minimum. Images drive click-through in Shopping more than anything else.
- Descriptions, GTINs, brand and product types. Complete, accurate structured data. Google prioritises products with complete, high-quality data — thin feeds get throttled.
Custom labels are your profit strategy. This is where feed work becomes margin work. Tag products with custom labels for margin band, best-sellers, seasonality and price tier, so you can bid on and segment by profit rather than treating every SKU the same. A best-margin range and a loss-leader shouldn’t share the same target.
Feed tools. Use supplemental feeds and feed rules to fix and enrich data at scale, and consider a dedicated feed-management tool for large or messy catalogues. In 2026, Merchant Center can even auto-generate product videos from your images and feed data via its built-in Video Builder — useful raw material for YouTube and Demand Gen. See how we run this in Google Shopping Ads.
Campaign Structure for Ecommerce in 2026
The structural landscape shifted meaningfully this year, so start from where things actually stand.
Shopping vs Performance Max — what changed. As of early 2026, Performance Max is the only format available to new Shopping advertisers; existing Standard Shopping campaigns keep running but can no longer be created. On top of that, AI Max came to Shopping campaigns in April 2026, bringing generative text customisation that turns your feed data into dynamic ads answering conversational queries. In practice this means most ecommerce spend now flows through Performance Max, with Shopping inventory absorbed into it.
| Factor | Performance Max | Standard Shopping (legacy) |
|---|---|---|
| Availability (2026) | Default for new advertisers | Existing only — can’t create new |
| Inventory | All Google surfaces | Shopping + Search partners |
| Control | Improved in 2026 (see below) | High, granular |
| Best for | Most stores, scaling | Existing accounts wanting manual control |
Segment by margin and priority — even inside automation. The way to keep control in a PMax-dominated account is to split campaigns along lines that matter to profit: separate your best-margin and best-seller ranges from the rest so you can set different targets, split brand from non-brand, and separate proven products from new ones that need a learning budget. This is where those custom labels pay off.
Always protect your brand. Run a dedicated brand Search campaign. It’s cheap, defends against competitors bidding on your name, and captures shoppers already sold on you — and keeping it separate stops its efficiency from masking how your prospecting is really doing. Run non-brand Search alongside Shopping to capture text queries your feed doesn’t cover. See Google Search Ads.
Winning With Performance Max
PMax is where most ecommerce budget now lives, so making it earn is most of the job. The good news: 2026 made it far less of a black box.
Asset groups and audience signals. Structure asset groups by product theme rather than dumping the whole catalogue into one, and feed audience signals — your customer lists, high-intent segments — that give the AI a head start on who converts. Strong signals shorten the learning curve.
Control it — the levers you now have. 2026 handed PMax real controls: channel-level reporting showing exactly where budget goes across Search, Shopping, YouTube, Display, Discover, Gmail and Maps (with a cost-per-channel view), negative keywords at campaign and account level, customer-list exclusions so you spend on new customers instead of re-buying existing ones, a budget-projection report, and placement reports for brand safety. Use them. Decide between feed-only and asset-rich campaigns deliberately, apply brand exclusions, and use the new reporting to see inside what used to be opaque. See Performance Max.
Feed conversion value it can optimise to. PMax is only as good as the value signal you give it. Feed accurate revenue — and ideally margin — so Target ROAS optimises toward profit rather than turnover. This is the hinge the next section turns on. It’s exactly the approach behind our ecommerce case studies, where accounts scaled on Shopping and Performance Max with ROAS climbing as budgets grew — see House of Tiles (68% more online revenue at a 31% lower cost per acquisition) and Scotts for Tools (57% more online sales at 4.9× ROAS).
Bidding and Profitability
Here’s where “profit, not revenue” stops being a slogan and becomes maths.
tROAS vs Maximise Conversion Value. With limited data, Maximise Conversion Value builds volume; as data accumulates, add a Target ROAS to hold a return threshold. Choose based on how much conversion history the campaign has — targets set too aggressively on thin data just destabilise learning.
Know your break-even ROAS. A “good ROAS” is meaningless in the abstract — it depends entirely on your margin. Break-even ROAS is simply 1 ÷ gross margin: at a 50% margin you break even at 2.0x, at 25% margin you need 4.0x just to stand still. Calculate yours, set your target above it by whatever profit you require, and stop comparing your ROAS to someone else’s in a different-margin business.
Optimise to profit, not revenue (POAS). The most advanced — and most profitable — move is to feed margin rather than revenue into your conversion value, so the algorithm chases profit on ad spend (POAS), not top-line ROAS. Two products with identical prices but different margins should not be bid on identically, and only a margin-aware value signal teaches Google that. This is the single biggest differentiator between an account that grows revenue and one that grows profit.
Conversion Tracking for Ecommerce
Accurate data or the whole machine misfires — because every bidding decision above depends on the values you report.
Get the foundations right: GA4 and Google Ads purchase tracking properly configured, Enhanced Conversions switched on to recover conversions lost to browser restrictions, and Consent Mode v2 implemented if you serve any EEA or UK traffic (mandatory, enforced at session level, with modelling to preserve measurement for users who decline cookies). As spend grows, move to server-side tagging, which recovers a large share of otherwise-lost conversions and sharpens every automated decision.
Two ecommerce-specific pitfalls to watch: deduplicate your platform’s conversion numbers (Shopify, WooCommerce, etc.) against Google’s so you’re not double-counting, and pass first-party value and margin data rather than a flat revenue figure, so bidding can optimise to profit. This measurement layer is the backbone of how we run PPC management.
Remarketing and Retention
New-customer acquisition gets the attention, but recovered carts and repeat buyers are where margin hides.
Run dynamic remarketing off your feed so the exact products someone viewed follow them back, prioritise cart and checkout abandoners (your highest-intent audience anywhere), and use Customer Match to drive repeat purchase and win-back from your existing customer base. Demand Gen can prospect for new audiences visually. The balance to manage is acquisition versus retargeting spend — don’t let automation quietly pour budget into re-buying customers who’d have returned anyway; that’s what customer-list exclusions in PMax are for. See Google Remarketing.
Scaling Profitably as Budgets Grow
This is the payoff — and the part most stores get wrong by chasing revenue at the expense of margin.
The counter-intuitive truth, and the one our own accounts demonstrate, is that ROAS can climb as spend scales, not fall — if you scale the right way. Expand product coverage into ranges that are working, open new geographies and markets deliberately, and increase budgets gradually so campaigns don’t lurch back into learning. Watch profit as you grow, not just revenue: marginal ROAS on the last chunk of spend matters more than blended ROAS, and as long as your marginal return still beats break-even, that spend is making money. Push to the point where the next pound of spend stops clearing your margin threshold — and hold there while you improve the feed and creative to push the threshold outward.
A relevant proof point: premium and multi-market stores scaling profitably across regions — see Vitality Pro, scaled across the UK and US, and The Server Store, driving high-value sales through Search and Shopping.
Common Ecommerce Google Ads Mistakes
- Neglecting the feed. The root cause of most underperformance — titles, images and data quality outrank campaign settings.
- Chasing top-line ROAS while losing on margin. A high ROAS on low-margin SKUs can still lose money; optimise to profit.
- One giant Performance Max campaign. No margin segmentation means no control over where budget flows.
- No brand protection. Leaving your brand terms undefended hands cheap conversions to competitors.
- Ignoring Merchant Center disapprovals. They compound into suspensions — and in 2026, missing the feed-spec deadlines throttles you.
- Over-crediting last click. It undervalues upper-funnel and remarketing’s real contribution.
- Pausing winners too early. Judging products or campaigns before the learning period finishes wastes their potential.
Frequently Asked Questions
What’s a good ROAS for ecommerce?
There’s no universal number — it depends entirely on your margin. Break-even ROAS is 1 ÷ your gross margin (a 50% margin breaks even at 2.0x; a 25% margin needs 4.0x). A “good” ROAS is comfortably above your break-even. Comparing your ROAS to a business with different margins is meaningless.
Performance Max or standard Shopping?
As of 2026, Performance Max is the only Shopping format available to new advertisers; standard Shopping continues only for existing campaigns. Most ecommerce spend now runs through PMax — which gained real controls this year — so the practical answer for new accounts is Performance Max, structured and segmented by margin.
How do I fix a Merchant Center suspension?
Identify the specific policy violation flagged, fix the underlying issue (misrepresentation, missing policies, data mismatches between feed and site are common causes), then request review. Prevention matters more than cure: keep feed data accurate, shipping and returns policies clear, and stay current with the 2026 product-data specification.
How much budget do I need to start?
Enough to generate the conversion volume Smart Bidding needs to learn — for most stores that means a budget capable of producing a few dozen conversions a month. Start focused on your best-margin, best-selling products rather than spreading thin across the whole catalogue.
Why are my Shopping ads unprofitable?
Usually one of three things: a weak feed (poor titles/images limiting relevance), bidding to revenue instead of margin (winning low-margin sales that lose money), or no segmentation (best- and worst-margin products treated identically). Fix the feed first, then feed margin-based values, then segment.
How long until Performance Max stabilises?
Expect a learning period of a few weeks as the campaign gathers conversion data, and avoid major changes during it — every significant edit resets learning. Judge performance once it’s past learning and you have enough conversions to read the signal reliably.