The Complete Guide to Google Ads for eCommerce in 2026

1. Why eCommerce PPC is Different from Other PPC

eCommerce PPC isn't lead generation with a checkout button bolted on. It is a structurally different discipline that lives or dies on three things generic PPC doesn't depend on: product-feed dependency, catalogue-scale bidding, and product-level policy enforcement. A B2B lead-gen account can survive a mediocre landing page if the offer is strong enough. An eCommerce account cannot survive a broken feed - no feed means no Shopping ads and no Performance Max, full stop.

We define the eCommerce PPC hierarchy of needs as: 40% Feed Engineering, 30% Campaign Structure, 20% Bidding, and 10% Ad Copy. That ratio surprises most marketers who come from a search-engine-marketing background, where copy and keywords dominate the conversation. In eCommerce, your "keyword" is your product title, your "landing page" is your product feed data, and your "ad copy" is a combination of image quality, price competitiveness, and review stars that Google surfaces automatically. If your feed is broken, no amount of smart bidding can save you. If your structure is messy - one giant Shopping campaign covering 4,000 SKUs with no priority tiers - your budget bleeds into low-margin products while your bestsellers starve.

The three defining characteristics of eCommerce PPC are:

• Product-feed dependency: No feed means no Shopping ads and no Performance Max. Your ads are only as good as the XML or Content API payload behind them.
• Catalogue-scale bidding: When managing 500+ or 50,000+ SKUs, individual keyword bidding becomes meaningless. You must manage performance via groupings, custom labels, and algorithmic signals rather than manually adjusting bids product-by-product.
• Merchant Center policy enforcement: Unlike Search ads, which are usually approved at the account or ad level, Shopping ads are scrutinised at the individual product level. A single pricing mismatch on one of 1,000 products can trigger a warning, a suspension of that item, or - repeated often enough - a full account suspension.

Where a specialist eCommerce PPC agency wins compound value is at the intersection of these three things: feed engineering that keeps data clean, policy monitoring that catches issues before Google does, and measurement that tells you which of your 4,000 SKUs are actually profitable once shipping, returns, and margin are factored in. Generalist PPC agencies typically own one of those three; rarely all three.

2. The 5 Things Google Ads Does Automatically for eCommerce

Google has moved aggressively toward automation across every layer of the eCommerce stack. In 2026, fighting these features head-on is a losing strategy; the winning move is engineering your inputs - feed data, audience signals, creative assets - so the automation performs better on your behalf.

1. Auto-bidding via Smart Bidding: Strategies like Target ROAS (tROAS), Target CPA, Maximise Conversions, and Maximise Conversion Value use millions of real-time signals - device, location, time of day, browser, historical intent - to set a bid for every individual auction. Here's why this matters: manual bid management at catalogue scale is functionally impossible once you pass a few hundred SKUs, so Smart Bidding isn't optional, it's the only viable mechanism.
2. Automatic asset generation in PMax: Given a landing page and a product feed, PMax will generate video, display creative, and search headlines without any manual input. Here's why this matters: the auto-generated assets are usually mediocre, so supplying your own high-quality creative is one of the highest-leverage things you can do inside a PMax campaign.
3. Product feed matching via Merchant Center to Google Ads sync: The connection between Merchant Center and Google Ads is near-instantaneous once configured. Here's why this matters: price and stock changes on your website flow through to ads within minutes via the Content API, which reduces the risk of advertising an out-of-stock or mispriced item.
4. Automatic responsive Shopping ads: Google tests combinations of your title, image, and price presentation to see which yields the highest click-through rate for a given searcher. Here's why this matters: the quality of the inputs (title structure, image resolution) directly caps the ceiling of what this automation can achieve.
5. Data-driven attribution (default since 2023): Google no longer defaults to last-click; it distributes conversion credit across the touchpoints in a user's path using a machine-learned model. Here's why this matters: this changes how you should read campaign-level ROAS, because assist-heavy campaigns like Display and DSA will show a lower directly-attributed ROAS while still contributing real incremental revenue.

3. The 6 Things Google Ads Gets Wrong by Default

Google's in-platform "recommendations" are optimised to increase your spend, not your margin. Our job as an eCommerce PPC agency is to sit between the platform's defaults and your P&L. Below are the six defaults we override on nearly every account we take on.

3.1 PMax Cannibalising Brand Search

By default, PMax will bid on your own brand name alongside generic and competitor terms. This makes reported ROAS look excellent, because you're paying to serve an ad to someone who was already typing your brand name into Google and would have clicked your organic listing for free. The problem is diagnostic, not moral: a campaign that shows 15x ROAS built substantially on brand-name clicks tells you almost nothing about whether PMax is driving incremental revenue. The fix is to apply brand exclusions at the account level, or run a dedicated brand Search campaign and let PMax report on true prospecting performance.
Deep-dive: How to set up PMax brand exclusions >

3.2 Default Asset Group Structure is One-Per-Campaign

Google's default guidance nudges advertisers toward a single, broad asset group covering the whole catalogue. In practice this blends creative and audience signals across products that have nothing in common - a £20 accessory and a £2,000 sofa end up in the same asset group, shown the same generic imagery. We recommend segmenting asset groups by product category, price tier, or margin band, so the creative and audience signal actually match what's being sold.
Deep-dive: The Visionary Asset Group Framework >

3.3 Standard Shopping Without Priority Tiers

Running Standard Shopping as a single campaign with no High/Medium/Low priority structure wastes the one genuine advantage Standard Shopping has over PMax: granular control. A 3-tier priority structure lets you force budget toward proven, high-margin SKUs while still giving new or long-tail products a chance to gather data in a lower-priority, lower-bid tier.
Deep-dive: 3-Tier Priority Strategy >

3.4 Feed Quality Gaps Go Unflagged

Google will happily run ads against a feed with vague titles, missing custom labels, and thin descriptions - it simply charges you more for the privilege via lower Quality Score-equivalent signals and reduced impression share on competitive terms. Nothing in the standard Merchant Center dashboard proactively tells you "this title is costing you clicks." That diagnostic work has to be done manually or via a feed audit.
Deep-dive: Feed Engineering 101 >

3.5 Attribution Defaults Hide 20-40% of Revenue

Google Ads' self-reported conversions and GA4's cross-channel view frequently disagree, and both can diverge meaningfully from what actually clears in your payment processor once returns, cancelled orders, and cross-device journeys are accounted for. For longer consideration-cycle categories - furniture, jewellery, high-ticket beauty devices - we regularly see 20-40% of true revenue sitting outside what any single platform reports in isolation.
Deep-dive: Measurement & Attribution in 2026 >

3.6 Merchant Center Policy Audit Missing from Onboarding

Most brands - and most agencies - only look closely at Merchant Center policy when a suspension email arrives. By then, revenue has already stopped. A proactive policy audit at onboarding, especially for regulated categories like healthcare, supplements, alcohol, and financial products, should be a standard first step, not a reactive fire-drill.
Deep-dive: Fixing Disapprovals >

Campaign Type Comparison

Choosing the right campaign structure depends on your business goals, catalogue size, and desired level of control. This table compares the six core campaign types available to UK eCommerce advertisers across eight practical dimensions - click a column header to sort.

4. Merchant Center & Feeds - The Foundation

Merchant Center is the brain of your eCommerce ads. If the brain is foggy - missing attributes, mismatched prices, vague categorisation - every downstream campaign type suffers, no matter how well-structured your bidding strategy is.

4.1 Merchant Center Setup

Google Merchant Center is the free product database Google uses to run Shopping and Performance Max ads. Setup requires business verification, website verification, tax and shipping configuration, and feed submission. UK-specific setup includes VAT handling and UK-specific policy compliance (such as the CAP Code for advertising standards). Most accounts take 2-4 hours to set up correctly the first time.

When setting up for the UK market, ensure your tax settings reflect the standard 20% VAT rate, or the relevant reduced or zero rate for specific goods such as children's clothing or certain food items. Failure to align the price in your feed with the final price shown in the checkout, including tax, is the single most common cause of automatic item disapprovals. You must also verify website ownership via a meta tag, DNS record, or a linked Google Analytics 4 property before Google will let you advertise products from that domain.

We recommend the Content API integration method for Shopify and similar platforms, since it provides real-time price and inventory updates rather than waiting on a scheduled fetch. For smaller stores without developer resources, a scheduled fetch of a Google Sheet or hosted XML file is workable, but the fetch frequency should match - or ideally beat - your stock update cycle to avoid advertising sold-out items.

4.2 Product Feed Fundamentals

A product feed is the data file Merchant Center reads to know what to advertise. Google requires 11 attributes as a minimum: id, title, description, link, image_link, price, availability, condition, brand, and either gtin or mpn, plus google_product_category. Additional recommended attributes - custom_label_0 through custom_label_4, sale_price, product_type, and promotion_id - unlock advanced segmentation, seasonal promotions, and richer reporting further down the funnel.
Master your feed data >

4.3 Disapprovals + Policy

Product disapprovals come from three sources: feed data issues (missing GTIN, price mismatch, landing page mismatch), policy issues (restricted content, promotional text in the title, prohibited products), and website issues (broken product URLs, missing checkout flow, non-compliant landing pages). The fix workflow is consistent regardless of cause: identify the issue at the feed source, correct it there rather than patching Merchant Center directly, resubmit, and allow 24-72 hours for re-review.
Fixing Merchant Center issues >

4.4 Product Identifiers (GTIN, MPN, Brand)

Global Trade Item Numbers (GTINs) are the single most powerful identifier signal you can give Google. A valid GTIN allows Google to place your product into a catalogued auction where it can be directly compared against competitors selling the identical item, which - counter-intuitively - often improves match quality and impression volume even in competitive categories. If you sell an own-brand product with no existing GTIN, register GS1 barcodes rather than setting identifier_exists to false; the identity lift is worth the modest cost.
Deep-dive into GTINs >

4.5 Google Product Category Taxonomy

Google will use AI to infer your product category if you leave the field blank, but explicit categorisation via the google_product_category attribute removes the guesswork and prevents misclassification into the wrong vertical. This is especially important in the UK, where VAT treatment and applicable advertising policy can vary meaningfully by category - a supplement classified incorrectly, for example, can trigger unnecessary health-claims policy reviews.
Category mapping guide >

5. Performance Max for eCommerce - The Dominant Campaign Type

Performance Max is no longer "new." In 2026 it is the default engine for eCommerce growth, combining Search, Shopping, Display, YouTube, Gmail, and Discover placements into a single campaign driven entirely by your conversion goal. Unlike traditional campaign types, you don't choose where your ads appear - Google's model decides, allocating impressions to whichever placement it predicts will convert.

Success in PMax isn't about "optimising keywords"; there aren't any to optimise. It comes down to four levers, ranked by the impact we typically see:

1. Audience signals: Uploading first-party customer lists (Customer Match) and defining custom segments based on competitor URLs and relevant search terms gives the algorithm a "seed" audience to find similar high-value users. This is the single highest-leverage input in a new PMax campaign.
2. Asset variety: Supplying genuine, high-quality video and lifestyle photography - rather than relying on auto-generated assets - measurably improves creative performance in the auction. We recommend at minimum one 15-second horizontal video and one vertical video optimised for Shorts and Reels-style placements.
3. Negative keywords: Account-level negative keyword lists prevent PMax from wasting budget on irrelevant queries or, as covered above, cannibalising branded search. This is one of the few direct levers of control PMax still offers.
4. Listing groups: Segmenting products by performance tier inside the asset group stops a handful of bestsellers from consuming 90% of the budget while hundreds of other SKUs receive close to zero impressions. Deliberate listing-group structure gives every category room to be tested fairly.

We regularly see a ROAS lift of 20-30% simply from cleaning up PMax asset groups and improving creative inputs - no bidding-strategy change required. For AB Ellie, a high-end bridal brand, we used PMax to target high-intent wedding planners and brides-to-be by layering first-party customer data over Google's in-market wedding segments. This produced a 145% increase in conversion value while protecting their premium brand positioning.

The result was a 34% ROAS lift in that four-week window, achieved with zero change to the daily budget. The lesson generalises: PMax punishes fragmentation. Every extra asset group splits your conversion data and slows the algorithm's ability to learn, so the discipline of consolidating rather than proliferating asset groups is worth more than most bid-strategy tweaks.

ROAS Performance: Before vs After

Below is the anonymised ROAS distribution across 12 Visionary client accounts, comparing the 90 days before engagement against the most recent 90-day period under our feed-first methodology. Three named accounts (LADC, Biopreventative, AB Ellie) are shown with consent; the remainder are anonymised.

6. Standard Shopping - When to Use it Instead of PMax

PMax is powerful, but the "just use PMax for everything" advice repeated across most PPC content is wrong for a meaningful minority of catalogues. We recommend retaining or reverting to Standard Shopping in three specific catalogue patterns, and we regularly see Standard Shopping outperform PMax by 20-40% on a ROAS basis in these situations:

1. New product launches: PMax needs conversion data to learn. For brand-new products with zero purchase history, Standard Shopping with manual or portfolio bidding lets you force impressions and gather data quickly, rather than waiting for PMax's machine-learning model to gradually discover the product is worth showing.
2. Hyper-competitive, commoditised niches: If you sell a product that is identical to dozens of competitors' listings - think generic phone cases or standard household consumables - you need SKU-level bid control to protect margin. PMax will happily bid you into an unprofitable auction chasing conversion volume; Standard Shopping lets you cap exposure per product.
3. Low conversion volume accounts: If your account generates fewer than roughly 30 conversions per month, PMax's automated bidding struggles to exit the learning phase reliably. Standard Shopping on Manual CPC or a conservative tROAS target is typically more stable and predictable at low volume.

In these three scenarios, the loss of PMax's cross-network reach is more than offset by the control gained. The right answer for most accounts is not "PMax or Standard Shopping" but a deliberate split, with Standard Shopping carved out specifically for the SKUs that fit one of the three patterns above.

<?xml version="1.0" encoding="UTF-8"?>
<Campaigns>
  <Campaign name="[ECOM] - Shopping - High Priority - Catch-all">
    <Priority>High</Priority>
    <Budget>10.00</Budget>
    <BiddingStrategy>Manual CPC</BiddingStrategy>
    <NegativeKeywords>
      <Keyword>[Brand Term]</Keyword>
    </NegativeKeywords>
  </Campaign>
  <Campaign name="[ECOM] - Shopping - Medium Priority - Brands">
    <Priority>Medium</Priority>
    <Budget>50.00</Budget>
    <BiddingStrategy>tROAS</BiddingStrategy>
  </Campaign>
  <Campaign name="[ECOM] - Shopping - Low Priority - Clearance">
    <Priority>Low</Priority>
    <Budget>20.00</Budget>
    <BiddingStrategy>Max Conversions</BiddingStrategy>
  </Campaign>
</Campaigns>

7. Search Campaigns for eCommerce

While Shopping and PMax carry the volume, Search campaigns remain the specialists inside an eCommerce account. We use dedicated Search campaigns for three jobs Shopping formats can't reliably do alone:

• Brand protection: Ensuring you own the top organic and paid position when someone searches for your brand name directly, denying competitors the chance to intercept your own customers with a cheaper offer.
• Generic high-intent terms: Targeting category-level queries - "best silk pillowcase UK" or "waterproof running jacket women's" - that route to a collection page rather than a single product, capturing shoppers earlier in the comparison stage.
• Competitor conquesting: Bidding on competitor brand names to capture customers actively comparing alternatives, provided your ad copy and landing page genuinely earn the click rather than relying on the brand term alone.

A well-structured eCommerce Search account is typically 5-15% of total spend, but disproportionately efficient, because every click is driven by explicit keyword intent rather than algorithmic placement guessing.

8. Dynamic Search Ads + Remarketing

Dynamic Search Ads (DSA) are the bridge between your website's SEO and your PPC account. Google crawls your site and automatically generates ads for queries that match your content but aren't covered by your keyword-based campaigns. For catalogues with strong on-page SEO and thousands of long-tail product variants, DSA fills keyword gaps that would otherwise take months to build manually.

Remarketing, by contrast, is about closing the loop rather than opening new demand. In 2026 we focus heavily on dynamic remarketing - showing a shopper the exact product they viewed but didn't buy, at a price point and creative treatment matched to how far through the funnel they got (browsed vs added to cart vs began checkout).

9. Feed Optimisation - The Ongoing Lever

Feed optimisation is not a one-time setup task; it is a continuous process that compounds in effect the longer it's maintained. The four primary levers, in the order we typically prioritise them:

1. Title engineering: Structuring titles as Brand + Product Type + Key Attributes (colour, size, material) in a consistent, front-loaded order. This alone can lift impressions by well over 100% on poorly-titled catalogues, because the title functions as the primary matching signal against search queries.
2. Custom labels: Segmenting the feed by margin, season, or price tier (e.g. "Bestseller," "High Margin," "Clearance") so campaign structure and reporting can be built around business priorities rather than raw category alone.
3. Image enhancement: Testing lifestyle imagery against white-background product shots. In fashion and homeware, lifestyle context frequently wins on click-through rate; in electronics and commoditised goods, a clean white-background shot tends to convert better because shoppers are comparing specs, not aesthetics.
4. Description enrichment: Moving beyond copy-pasted manufacturer descriptions to include the specific attributes and use-case language Google's semantic matching engine relies on, which improves match quality on longer, more descriptive search queries.

10. Measurement + Attribution - The Metrics that Predict Revenue

In a world without reliable third-party cookies, measurement has become an engineering problem rather than a reporting one. We use a three-source reconciliation framework so decisions are based on reality rather than platform-inflated numbers.

• Google Ads: Useful for real-time bidding signals and platform-specific performance, but it will often claim view-through and assisted conversions that may have happened regardless of the ad.
• GA4: Provides a cross-channel view of the customer journey. We compare "first-user" against "session-source" attribution to understand how PPC contributes at the top versus the bottom of the funnel, using the Advertising Workspace to compare models side by side.
• Store analytics (Shopify or equivalent): The source of truth for actual, bank-cleared revenue net of refunds and cancellations. We reconcile this against Google Ads' reported spend to calculate the Marketing Efficiency Ratio (MER) - total revenue divided by total marketing spend - which is far harder to game than platform-reported ROAS.

We also implement Enhanced Conversions as standard on client accounts. This securely hashes customer data such as email or postal address at the point of purchase and matches it back to a signed-in Google user. In our experience this recovers between 5% and 15% of conversion data that would otherwise be lost to Apple's App Tracking Transparency framework or browser-level cookie blocking - data that, without recovery, makes Smart Bidding models less accurate simply because they're training on an incomplete picture of what actually converted.

11. eCommerce PPC by Vertical

Tactics that work for a fashion brand will underperform, or actively fail policy review, for a supplement company. Vertical context changes creative strategy, campaign mix, and compliance risk simultaneously.

• Fashion: Heavy weighting toward visual assets, lifestyle-led PMax creative, and aggressive seasonal budget scaling around key retail moments. Our work with LADC demonstrates what feed-first fashion Google Ads scaling looks like at volume - moving from a fragmented catalogue to a structured PMax and Shopping split delivered the +1,386% performance lift referenced above. AB Ellie, in the adjacent bridal-fashion space, shows the same principles applied to a higher-consideration, lower-frequency purchase.
• Beauty: Strong emphasis on social proof signals in feed and creative, YouTube and Shorts placements for "get the look" style content, and shopping bundles that lift average order value in a category where per-unit prices are typically low.
• Healthcare and supplements: A compliance-first approach is mandatory. Biopreventative's turnaround - from constant disapprovals to 7x ROAS in 30 days - came from removing unsupported health claims at the feed and ad-copy level before scaling spend, rather than trying to scale first and fix policy issues reactively.

Budget Allocation Calculator

Use the calculator below to model a starting budget split across Performance Max, Standard Shopping, Search, and DSA/remarketing based on your monthly spend and catalogue size. Treat this as a launch-day starting point, not a fixed rule - revisit the split every 30 days once real conversion data comes in.

12. The 30-Article Strategy Map

This guide is the anchor for the wider eCommerce PPC topical map. Every campaign type, feed attribute, and measurement concept above has a dedicated deep-dive article linked below, grouped into seven clusters.

FAQs - eCommerce PPC Questions Answered