Google Ads mistakes that burn UK ecommerce budgets

January is a brutal month to open Google Ads if you sell online in the UK. Boxing Day returns are still landing at the warehouse, Sale filters are still on, and your Shopping and Search campaigns keep bidding like December never ended. A Manchester apparel brand we reviewed last winter was still paying £1.80-£2.40 a click for “cheap [category]” and “lookalike [brand]” queries three weeks after peak. In-platform ROAS looked acceptable. Finance saw a different story once VAT, payment fees, and a 22% return rate came off the top.

That gap is where most PPC for e-commerce companies UK programmes quietly lose money. Not in a single catastrophic campaign, but in a handful of structural mistakes that train Google’s automation on the wrong intent, the wrong page, and the wrong conversion. Below are the failures that burn budget fastest for UK ecommerce teams, and what to change before you scale spend again.

Contribution margin ROAS: paying for discovery terms that never clear VAT and returns

High-intent search still works for ecommerce. The problem is how loosely most accounts define “intent”. Category heads like “women’s trainers”, “oak dining table”, or “protein powder UK” pull volume. They also pull browsers comparing five sites, code-chasers, and people who will never clear your true contribution margin once you load in VAT, shipping subsidy, and returns.

UK shops feel this harder than many US peers because average order values are often lower and free-delivery thresholds train customers to abandon borderline baskets. A Leeds homewares brand ran non-brand Search at a 4.5x platform ROAS target and celebrated for a quarter. When merchandising rebuilt the P&L with a 19% return rate on upholstered items and a flat £4.99 outbound cost the brand was still eating, profitable orders sat closer to 2.1x on those same terms. The “winning” keywords were discovery terms dressed up as purchase intent.

Fix the definition before you fix the bid. Split brand, competitor, and generic into separate campaign or bid structures so blended ROAS cannot hide the bleed. For generics, bid to contribution margin after expected returns, not to last-click revenue in Google Ads. If a cluster cannot clear that bar at realistic CPCs (Search still averages near the high-£2 range in many verticals, and competitive fashion or beauty can run higher in London-weighted auctions), it belongs in a capped test budget or content, not in your core Shopping and Search mix.

Performance Max will happily spend into those discovery queries if your asset groups and feed signals are broad. That is not a reason to switch PMax off. It is a reason to feed it cleaner product margins, tighter audience signals, and stronger negatives at account level so “cheap”, “free”, “DIY”, “repair”, and off-size modifiers stop eating the budget that should buy ready-to-purchase traffic.

Negative keyword management: why search terms debt lets Smart Bidding learn junk intent

In the 40-odd UK ecommerce accounts we audited in 2024, the median share of spend on zero-purchase query themes was 24%. On ecommerce accounts that waste is almost always search terms debt. Someone launches broad match or PMax, Smart Bidding starts optimising, and the search terms report fills with job-seekers (“warehouse operative [brand]”), research queries (“how to measure for blinds”), wholesale intent, and marketplace scavengers looking for your SKU cheaper on Amazon.

Teams delay negatives because automated bidding is supposed to “figure it out”. It does figure something out: it learns that those clicks sometimes bounce after a soft engagement, or worse, that a tiny fraction convert on a low-value accessory and therefore look efficient. By week three the algorithm has a corrupted picture of what a good session looks like. Manual cleanup then feels like fighting the machine.

Run a standing negative process, not a quarterly spring clean. Pull search terms by cost and by conversion lag. Add account-level negatives for employment, DIY/how-to, wholesale, freebie, and pure research modifiers that never buy. Build shared lists by theme (size guides, spare parts, “vs” comparisons you do not win) and attach them before you raise budgets. When a new collection launches, seed negatives from the previous season’s junk rather than discovering it again at £3 a click.

Broad match still has a place when conversion tracking is clean and the feed is tight. It does not have a place as an excuse to skip query hygiene. If your team cannot say who last touched the negative lists and what they blocked, you are not running PPC for e-commerce companies UK at a professional standard. You are renting clicks and hoping the model is wiser than your catalogue.

If you want a second pair of eyes on structure before the next budget increase, a focused SEM / Google Ads audit on query themes and bid split usually surfaces the waste faster than another creative refresh.

Landing page experience: when the page breaks the Shopping ad promise

Plenty of UK ecommerce brands still route paid traffic to the homepage, a bloated collection URL, or a PDP that does not match the offer in the ad. The click looked perfect. The post-click experience did not. Mobile is more than 60% of paid search clicks in most retail accounts, yet mobile conversion often trails desktop by 30-40%. That gap is rarely “mobile shoppers are just browsers”. It is slow LCP, clumsy variant pickers, delivery promises buried below the fold, and trust marks that only appear on desktop templates.

Match the ad to a single job. A Shopping ad for a specific SKU should land on that PDP with price, stock, delivery date to UK mainland, and returns policy visible without a hunt. A Search ad for “organic dog food subscription” should not dump people on a 40-product PLP sorted by best sellers. Message match is not a CRO nicety. It is how you stop paying for intent you immediately confuse.

Proof matters as much as speed. UK buyers lean hard on delivery cut-offs, Trustpilot volume, and whether returns are free to a shop or a courier. If your ad implies next-day and the page hedges with “usually 2-3 days”, you will see add-to-carts without checkouts. Session recordings make the leak obvious: rage taps on size charts, scroll abandonment above the fold on mobile, and form fields on express checkout that drop when Apple Pay is not offered. HeyLead Insights overlays session recordings against your campaign URLs, so you can see whether rage taps on the size chart are concentrated on Shopping traffic or brand search - rather than diagnosing from blended bounce rate.

Core Web Vitals still separate winners from also-rans. Google’s own CrUX data (HTTP Archive, mid-2024) shows fewer than half of retail origins pass all three thresholds on mobile. If your PDP fails INP on mid-range Androids, Smart Bidding will keep sending traffic into a page that cannot convert it. Fix the template before you raise non-brand bids. A cleaner handoff often recovers more ROAS than another 15% budget lift.

Conversion tracking setup that rewards baskets, not paid orders

This is the mistake that looks sophisticated and still wrecks accounts. Teams fire primary conversions on add-to-cart, begin checkout, or “lead” events borrowed from a B2B template. Smart Bidding then optimises for the action you crowned, not for paid orders at acceptable margin. You get cheaper “conversions”, rising ATC volume, and a finance team asking why revenue is flat.

Set one primary conversion for purchase, value-inclusive, with actual order value passed and refunds or cancellations reflected where your stack allows. Enhanced Conversions and a clean server-side or CAPI-style backup matter because ad blockers and privacy changes still strip a meaningful slice of browser-side signals. When 20%+ of conversion data never reaches the auction, bidding models fill gaps with noise. First-party data and tighter event quality beat another layer of audience hacks.

Watch the second trap: last-click vanity inside Google that double-counts against Meta or email. Ecommerce journeys hop. A customer clicks Shopping on Tuesday, opens a Klaviyo mail on Thursday, and buys on Saturday. If every channel claims the order at full value, every channel looks like a hero and none of them is safe to cut. Use platform data for in-channel optimisation, then reconcile to a business view that finance trusts: orders, contribution after returns, and new vs returning customers. Guaranteeing a specific ROAS before you have seen the account, the feed, and the unit economics is how agencies lose trust. Honest programmes take weeks of signal cleanup before automation earns the right to spend hard.

A Bristol beauty brand we walked through had tROAS campaigns “winning” at 6x while Shopify new-customer revenue was soft. The primary goal included a high-volume sample-kit ATC treated like a full-size order. Demoting that event and passing real basket value cut reported volume for a fortnight and then lifted paid revenue per click by roughly 31% once bidding re-learned what an order looked like. Ugly short-term dashboard. Better bank account.

Brand cannibalisation in Google Ads, stockouts, and dayparting UK shops still ignore

Brand cannibalisation is the leak that surprises teams most. Performance Max and broad Search will bid on your brand terms if you let them, then take credit for orders you would have won anyway. Protect brand in its own campaigns with exact and phrase coverage, own the SERP against resellers where policy allows, and keep non-brand judged on incremental return. When leadership says “Performance Max drives the bulk now”, ask what share is branded queries and remarketing paths you already paid to create.

Stock and feed hygiene is the quieter twin. Google will keep spending on out-of-stock or low-margin SKUs until the feed and merchant centre rules stop it. Exclude zero-stock, suppress chronic high-return products from paid, and segment best sellers from clearance so a 1.2x clearance ROAS target cannot drag down full-price heroes. A Birmingham electronics retailer stopped overnight waste by blocking any SKU under five units on hand from Shopping; wasted spend on dead PDPs fell within eleven days without a full campaign rebuild.

Dayparting and budget pacing still get less attention than they deserve. UK retail peaks are not flat. A brand running flat budgets saw Sunday 7-10 pm carry 34% of weekly mobile transactions at 18% lower CPC than Monday morning - capping spend in dead midweek slots freed that inventory. Align ad schedules and budget pacing with when demand actually converts, especially around UK paydays and major retail events. Automation will spend into weak hours if you never constrain it.

For teams building a wider growth stack around paid search, pair these fixes with dedicated E-Commerce marketing work on offer clarity and on-site proof so every click you keep has somewhere honest to land.

Frequently asked questions

How much budget waste is normal in a UK ecommerce Google Ads account?

If nobody has maintained negatives, feed exclusions, and landing match for a quarter, a large slice of spend on weak intent is common - in our 2024 UK ecommerce audits the median on zero-purchase query themes alone was 24%. That is not a tax you should accept. Clean query themes and stock rules usually reclaim a meaningful slice within a few weeks without starving volume.

What ROAS should a UK ecommerce brand target?

There is no universal number. Industry averages around 4x are directional, not a promise. Your floor is contribution margin after VAT, discounts, shipping, payment fees, and expected returns. Set tROAS or CPA from that P&L, by product margin band, not from a blog benchmark.

Is broad match safe for ecommerce Search?

Only with clean purchase conversion tracking, active negatives, and enough conversion volume for Smart Bidding to learn. On thin catalogues or messy tracking, phrase and exact with strong Shopping coverage is still the saner path until signals improve.

Putting it to work

Next 30 days

1. 1Expand the fix to the next one or two money paths only after the first one shows cleaner bookings.
2. 2Align creative, keywords, or content with the same offer the page now states.
3. 3Review booked outcomes weekly; cut anything that still only produces unqualified volume.

Next 30 days

1. 1Expand the fix to the next one or two money paths only after the first one shows cleaner bookings.
2. 2Align creative, keywords, or content with the same offer the page now states.
3. 3Review booked outcomes weekly; cut anything that still only produces unqualified volume.