A Manchester DTC brand can sit at 4.1x last-click ROAS in Google Ads and still miss payroll on the fulfilment side. The Shopping auction priced the click in pence. Royal Mail, a 12% return rate on fashion SKUs, and a 14-day cooling-off window priced the order in pounds. That gap is the real job of PPC for e-commerce companies in the UK, not another dashboard that celebrates revenue before contribution.
UK retail search is expensive enough that you cannot treat Performance Max, Shopping, and brand Search as three separate hobbies. Search CPC has been climbing (the 2026 cross-industry average sits near $2.96, up about 12% year on year, LocaliQ 2026 industry benchmarks; figures in USD; UK GBP equivalents vary by vertical), and 20-30% of SEM budgets still leak on junk queries, missing negatives, and product pages that were never built for the ad that sent the click. Google Ads still posts a 4.2x average ROAS across industries (LocaliQ 2026 industry benchmarks; figures in USD; UK GBP equivalents vary by vertical). Plenty of UK catalogues print that number and lose money after VAT, shipping, and refunds.
This piece is for the marketer who owns the paid number at a UK ecommerce company: how you structure the account so the same SKU is not buying itself three times, how you feed bidding a conversion that actually paid, and how you stop mobile traffic (63%+ of paid search clicks) from dying on a PDP that only works on a laptop.
Why UK retail PPC looks profitable until you net out returns and delivery
You will usually find the first lie in the conversion action, not in the bid strategy. If the pixel fires on “purchase” and you never pass order value net of discount, or you never exclude cancelled and returned orders, Smart Bidding optimises for the wrong pound. Automated bidding then looks greedy. It is not inflating spend for sport. It is chasing the signal you gave it.
UK specifics make that worse. A £42 basket with free next-day delivery and a 30-day returns window is not the same conversion as a £42 basket of non-returnable consumables. Fashion and homeware teams still report ROAS as if every paid order stuck. Then finance asks why contribution shrank while Google looked healthy. Last-click also double-counts when the same order hits Search, then PMax, then a branded remarketing click. That is the “attribution is basically dead” complaint you hear in Slack, and it is fair if you are still using last-click as the CEO answer.
High CPC pressure pushes new accounts back to manual CPC. That can be the right 10-day move while you fix tracking. It is a terrible 90-day strategy. You lose volume, you starve learning, and you never collect the first-party events that Enhanced Conversions and a clean purchase tag are meant to protect. Ad-blockers and privacy changes already knock out 20%+ of conversion data. Starve the remainder and tROAS has nothing honest to work with.
Mobile is the other quiet tax. Most of the clicks arrive on a phone. Conversion rates still lag desktop by 30-40% in a lot of retail accounts, and only 55.9% of origins pass all three Core Web Vitals (HTTP Archive CrUX data, Q1 2026). If your PDP takes a beat too long on INP, you paid UK Shopping prices for a bounce. That is not a media problem. It is a product-page problem wearing a CPC costume.
If paid search is dumping people on a homepage or a sluggish PDP, a dedicated Google Ads programme for ecommerce should start with the offer and the page, not another campaign clone.
Search, Shopping, and PMax without feeding the same SKU three times
Performance Max drives the bulk now in a lot of UK catalogues, and that is fine if you treat it as a demand catcher with guardrails, not as a black box you top up every Monday. The failure mode we still see is three auctions bidding on one hero SKU: a Search exact match, a Shopping product group, and a PMax asset group with the same feed. You pay twice for brand, you cannibalise the query you already own, and you call it “coverage”.
Keep Search for the queries where a human should still write the promise: high-intent non-brand (“women’s waterproof hiking boots size 6”), competitor conquest if the unit economics survive, and brand defence when Amazon and marketplaces squat on your name. Shopping (or a tightly themed PMax feed label) owns the SKU grid. Brand Search should be its own campaign with its own budget so PMax cannot vacuum branded demand and pretend it discovered it.
Signals beat keyword theatre. Upload first-party purchasers, pass Enhanced Conversions, and send value that matches what finance will recognise. “Run broad” only works if creative and the PDP do the targeting. On Meta, creative is your targeting and ads can die in days, not weeks. On Google, the equivalent is feed quality plus query intent: titles that name size, material, and use, GTINs that actually match, and custom labels for margin bands so you can bid down the SKUs that cannot carry UK postage.
Do not rebuild the account every fortnight. Constant structure changes reset learning. Fewer campaigns, broader targeting, and a conversion action you trust will beat a 2024 maze of match types. AI Max and similar automated Search setups are showing 12-18% lower CPCs versus standard Search in some 2026 benchmarks. That only helps if the landing URL matches the query and the margin label is honest.
A Leeds outdoor retailer we mapped had PMax, standard Shopping, and a catch-all Search campaign all eligible for the same “down jacket” queries. Branded CPCs looked cheap because PMax was stealing them. Non-brand CPA looked awful because Search was left with leftovers. They carved brand into its own campaign, labelled the feed by contribution after average returns, and capped PMax so it could not outbid brand. Revenue from paid did not jump overnight. Contribution per session did, which is the only number that funds the next inventory buy.
Get a free marketing audit - we review your search, ads, and landing pages and send back what to fix first.
Get a free audit
Rebuild the conversion signal and the PDP before you raise budgets
Raise budgets last. If 20-30% of spend is already wasted, more budget just scales the leak. Work the checklist below in order. Skip a step and tROAS will look “broken” when the account is actually doing what you asked.
Audit scorecard
-
Purchase value that finance would signFire one primary purchase conversi
Purchase value that finance would signFire one primary purchase conversion with revenue after discount. Exclude test orders, staff orders, and failed payments. If you later import refunds from Shopify or your OMS, do it on a delay that Smart Bidding can use, not a spreadsheet two months late.
-
Enhanced Conversions on, not "we'll do it next quarter
"Turn on the unified Enhanced Conversions path so hashed first-party data patches the 20%+ you lose to blockers. If Event Match Quality is weak, fix the fields (email, address, phone) before you argue with the algorithm.
-
Feed labels by margin, not by merchandising vanityCustom labels for cont
Feed labels by margin, not by merchandising vanityCustom labels for contribution after typical UK shipping and a realistic return rate. Bid up replenishment and accessories. Bid down hero SKUs that look good in the lookbook and lose money in the warehouse.
-
Negatives that match how Brits actually searchJobs, PDF, free, DIY, sala
Negatives that match how Brits actually searchJobs, PDF, free, DIY, salary, wholesale (if you are DTC), and marketplace intent you cannot fulfil. Review search terms weekly at first, then on a cadence that does not reset learning with daily nips.
-
Brand carved out of PMaxSeparate brand Search with its own budget
Brand carved out of PMaxSeparate brand Search with its own budget. If PMax still captures brand, you are paying a discovery tax on demand you already own.
-
Query-to-PDP message matchThe ad promised size 6, waterproof, next-day
Query-to-PDP message matchThe ad promised size 6, waterproof, next-day. The PDP must say those three things above the fold, in pounds, with stock and delivery to UK postcodes. If they said they just send traffic to the homepage, that is the leak.
-
Mobile PDP that survives INPSticky add-to-bag, size guide without a new
Mobile PDP that survives INPSticky add-to-bag, size guide without a new tab, and images that do not block interaction. Check Core Web Vitals on the money URLs, not the blog.
-
UTMs that survive the checkout hopShopping to PDP to cart to Shopify che
UTMs that survive the checkout hopShopping to PDP to cart to Shopify checkout drops parameters if you are sloppy. Build links properly or GA4 and the ads account will tell different stories about the same order.
DIY playbook for the next build
Action checklist
- Export 90 days of paid orders with SKU, channel, discount, shipping, and return flag. Rank SKUs by contribution, not revenue.
- Pause or demote the bottom contribution decile in the feed before you touch bids.
- Confirm one primary conversion, Enhanced Conversions, and value rules that match GBP including VAT the way you report internally.
- Split brand Search. Give PMax a ceiling. Keep non-brand Search only where a query still needs a written promise.
- Rebuild the top 10 paid landing PDPs: price, delivery, returns, size, and proof in the first screen on a mid-range Android.
- Watch session behaviour on those PDPs with HeyLead Insights so you see rage taps on size charts and form abandon, not just a blended conversion rate.
Free tools - try these yourself
PMax in cleaned-up accounts still lands in a 4x-8x ROAS band when the feed and the conversion are honest. That is not a promise you can print before you have seen the account, the PDP, and the unit economics. Anyone who guarantees a ROAS on a sales call has not opened your feed yet. Realistic lift takes months, not a weekend of campaign cloning.
If the page is the bottleneck, you already know it: high CTR, weak add-to-bag, and a mobile conversion rate that makes desktop look like a different shop. Fix that before you argue about bidding.
What marketing leaders are seeing
Paraphrased from client conversations; details anonymised.
“We were celebrating 5.2x in Google while finance had us at 1.8x after returns. The conversion action counted every Shopify paid order, including the ones that came back in the poly bag two weeks later.” - Head of Growth, UK fashion ecommerce
“PMax looked like it found new demand. Search term reports showed it was mostly our brand plus the same three jackets Shopping already owned. We cut the overlap and contribution per click finally moved.” - Performance marketing lead, home and living retail
FAQs
Should UK ecommerce companies pause PMax if brand is getting stolen?
Not as a first move. Split brand Search, set a PMax budget ceiling, and use account-level brand exclusions where they still behave. Pausing PMax often dumps the leftover queries into a messy Search campaign and you lose Shopping inventory coverage overnight.
Is manual CPC safer while CPCs keep rising?
Use it as a short diagnostic, not a philosophy. Manual CPC on a new account can stop a runaway tROAS while you fix value tracking. Leave it on and you will under-buy the SKUs that can actually carry UK delivery costs.
Do we need Meta if Google already takes most of the budget?
Only if you have creative that can target. Meta’s reported average across industries hovers around 2-3x (Databox/Revealbot 2025 benchmarks), with higher variance at sub-£5k/month spend. If you cannot refresh UGC before ads die in a few days, keep the money in feed-based Google until the PDP and the offer are tight.
What reporting cadence actually helps a marketing lead?
Weekly on wasted spend, search terms, feed disapprovals, and contribution, not a monthly ROAS slide. Monthly-only reporting buries a bad product group for 30 days of UK CPCs you cannot get back.
Can we guarantee a ROAS before the first month?
No. You can guarantee a conversion setup, a feed labelled by margin, and landing pages that match the query. The auction and your return rate still get a vote.
Putting it to work
Execution sprint
This week
- Pull 30-90 days of performance for UK e-commerce (Search Console, ads, CRM, or call logs - whatever you have).
- Flag the top leak: wrong intent, weak page, slow response, or dirty conversion tracking.
- Ship one fix on the highest-traffic money path (page, campaign split, or response rule).
- Run the free tools below on that same URL or account and log the findings.
Free tools for this sprint
Next 30 days
- Expand the fix to the next one or two money paths only after the first one shows cleaner bookings.
- Align creative, keywords, or content with the same offer the page now states.
- Review booked outcomes weekly; cut anything that still only produces unqualified volume.
Pull the last 90 days of paid orders with SKU-level contribution (after discount, shipping, and a realistic return rate), then mark every campaign that is still bidding on the bottom decile. That single cut often funds the tracking and PDP work the auction has been waiting for.
This week
-
Confirm one primary purchase conversion in GBP with value after discount.
-
Turn on Enhanced Conversions and spot-check a handful of real orders against Google Ads.
-
Carve brand Search out of PMax and cap PMax so it cannot vacuum branded queries.
-
Add negatives for jobs, free, PDF, and any marketplace intent you will not fulfil.
-
Run the Core Web Vitals checker on the ten PDPs taking the most spend.
-
Build UTMs that survive Shopify checkout with the UTM link builder.
Next 30 days
-
Label the feed by contribution, not by merchandising season.
-
Rebuild message match on those ten PDPs (size, delivery, returns, price in £).
-
Reconcile Google Ads revenue to OMS, including refunds.
-
Only then raise budgets on the SKUs that still print money after postage.
If you want a second pair of eyes on the overlap between Shopping, Search, and PMax, and on the conversion value that is teaching the bids the wrong pound, HeyLead can take that feed, tracking, and PDP alignment work off your plate. Chat with us on WhatsApp
Free marketing audit, or reach Martin directly:
Get a free audit Connect on WhatsApp · martin@heylead.com