Google Ads for UK banks and credit unions: a practical guide

The first query that burns a UK bank or credit union Search budget is rarely “current account”. It is “mortgage advisor jobs Leeds”, “ISA comparison Which?”, or “credit union loan calculator” from someone who will never sit in a branch. Marketing leads at mutuals and challenger banks still get judged on funded products and booked appointments, not on clicks that look cheap in the Google Ads UI. That gap is the whole job.

Search CPCs in UK financial services routinely sit in a much higher band than cross-industry averages - typically £6-£18 for mortgage and ISA head terms, per WordStream’s 2024 UK industry report. Twenty to thirty percent of SEM spend still leaks to irrelevant queries, neglected negatives, or a homepage that does not match the ad. Mobile takes 63%+ of paid clicks, then converts 30-40% worse than desktop if the ISA or mortgage form is a PDF-era maze. Average Google Ads ROAS sits around 4.2x across industries (WordStream, 2023 cross-industry median); UK FS ROAS is typically measured differently given multi-week KYC cycles. Financial services will not hit a generic cross-industry figure if you treat every conversion as a form fill and let Performance Max invent demand you cannot underwrite.

This piece is a UK-specific Search playbook for banks and credit unions: how you cluster intent in pounds, how you keep FCA-flavoured copy honest, how you stop smart bidding from chasing brochure downloads, and what you actually do in the account this month. It is not another “buy funded accounts, not forms” sermon. You already know volume is a dead metric. The work is making Search pay for products you can actually open.

What UK savers and borrowers type before they ever pick a branch

People in Manchester, Glasgow, and Cardiff do not search like a national brand deck. They search like someone comparing a 4.6% easy-access ISA against a local credit union’s payroll deduction, or like a first-time buyer who typed “Help to Buy ISA still available 2026” at 22:14. High intent clusters around named products: current accounts with switching bonuses, fixed-rate ISAs, personal loans with representative APR, mortgages with LTV bands, and business current accounts for limited companies. Low intent looks similar in the auction: “best bank UK”, “credit union near me jobs”, “FCA register”, “how to complain to financial ombudsman”.

You will also see comparison-intent that you cannot win with a product page. “Best cash ISA 2026 MoneySavingExpert” is a research session. Bidding on it without a comparison-style landing page and a clear eligibility box is how you buy expensive bounce. Regional credit unions get a different mix: “credit union loans Birmingham”, “payroll saving scheme”, “join [town] credit union”. Those queries convert when the ad names the common bond (live or work in the borough, employer list, membership fee in pounds) instead of a generic “community banking” line.

AI Overviews now sit on a large share of money queries. Google Search serves over 8 billion queries daily; AI Overviews were reported to reach over 1 billion users per month as of May 2024 (Google I/O 2024), and zero-click behaviour is the default on many “what is an ISA” questions. Impression share without click-through is not a vanity win if your CPL is measured in booked interviews. Treat AIO as a reason to tighten exact and phrase match around product-plus-location and product-plus-rate language, not as a reason to abandon Search. The clicks that remain are more expensive. They also closer to a decision.

A realistic week for a UK credit union marketer: 40-80 paid sessions to a membership page, 8-12 application starts, 2-4 completed memberships, and a handful of “call us” taps from people who still want a human to explain the common bond. A high-street bank’s digital team might see thousands of ISA clicks in ISA season (April and the run-up to tax year end) with a conversion rate that looks healthy until operations flags that half the applications fail KYC. Your Google Ads structure has to reflect that product reality, or Target CPA will happily buy more of the wrong applications.

Where UK financial Search accounts quietly waste 20-30% of spend

The leak is usually not “Google is expensive”. It is query junk, homepage landings, and conversion actions that teach Smart Bidding the wrong lesson. Job-seekers click “careers” adjacent terms. Students click “student bank account” when you only offer a standard current account. Debt-help traffic hits “credit union loan” when your underwriting will decline them. Comparison-site brand terms (MSE, Compare the Market, GoCompare) soak budget if you leave them un-negatived. That is the 20-30% waste band you already hear about in SEM. In FS it shows up as applications that never fund, not as “bad leads” in a CRM dump.

Automated bidding does inflate spend when the conversion you import is a PDF download or a newsletter tick. Plenty of UK banks still fire a primary conversion on “ISA guide downloaded”. The algorithm then hunts more guide readers. You asked it to. High CPC pressure then pushes teams back to manual CPC on new campaigns, which feels like control and usually just slows learning. The fix is not “never use tCPA”. It is a conversion stack: micro events for diagnosis, one primary action that maps to a started application or booked branch appointment, Enhanced Conversions on, and a weekly negative list that reads like a risk committee memo.

Landing pages are the other silent tax. Sending “2-year fixed ISA 4.1% AER” traffic to a rates hub that also sells mortgages, business banking, and a careers widget is how you lose the message match. Mobile conversion lag of 30-40% versus desktop is often INP and form length, not “mobile users are browsers”. Only 55.9% of origins pass all three Core Web Vitals in May 2026 CrUX data. A 14-field membership form with a scanned ID upload on 4G in a Tesco car park will not clear that bar. If you need proof of where people stall, session behaviour on the application step beats another dashboard argument. HeyLead Insights is built for that post-click picture: scroll depth, rage clicks on the APR footnote, abandon on the address lookup.

If your agency or in-house team still parks every ad on the homepage, treat that as a red flag you already know how to read. Dedicated pages per product cluster, with representative APR, eligibility, FSCS wording, and a single CTA, are table stakes. A short look at Google Ads management for Search programmes is useful when the account has grown into 40 campaigns that all share the same “Apply now” conversion.

A UK Search build: product clusters, negatives, and signals that underwriting can live with

Keep the account boring on purpose. Fewer campaigns, tighter product themes, broader match inside those themes, and ruthless negatives. “Signals over keywords” is the 2026 slogan. Google still needs keywords in financial services because the regulator and the product catalogue are specific. You feed better signals. You do not delete the query map.

Checklist

Two UK account stories: ISA season versus a regional common bond

A performance lead at a mid-sized UK bank ran one “Savings” campaign through March. Broad match on “ISA” pulled “lifetime ISA first time buyer”, “help to buy ISA closed”, and “ISA allowance leftover”. CPC looked efficient. Application starts were up 41% week on week. Operations then reported that 3 in 5 starts were Lifetime ISA traffic the product team did not offer on that brand. The change that mattered was not a new bid strategy. They carved Lifetime and cash ISA into separate campaigns, added “lifetime”, “help to buy”, “HTB”, and “LISAs” as negatives on cash ISA, and pointed each ad group at a rate page dated for that tax year. Starts fell. Funded cash ISAs in April were the number the ALCO pack actually wanted.

A credit union marketing manager in the West Midlands had the opposite problem. “Credit union near me” was cheap and noisy. Membership applications arrived from postcodes outside the common bond, which membership officers had to reject by hand. They added a postcode checker as the first step, imported only “eligibility pass + application start” as the primary conversion, and used location targeting plus negatives for neighbouring towns they cannot serve. Call-only ads ran 09:00-17:00 when the office could pick up. Cost per started application rose. Cost per accepted member dropped because the algorithm stopped celebrating ineligible postcodes.

Neither story needed a new platform. Tighter positioning, a clearer offer, and a conversion that operations would sign off beat adding Meta or another Search partner. Performance Max can sit beside this as a brand-and-remarketing layer once Search terms are clean. Let it drive bulk only after the conversion action is something a funded account, not a PDF.

If the landing page still dumps people after the rate table, fix that before you scale. Insights on the application step will show whether the leak is the APR accordion, the document upload, or a cookie banner covering the CTA. That is cheaper than another £8k-£25k conversation about “the channel is tired”.

FAQs

Should UK banks still use keywords if “signals over keywords” is the 2026 line?

Yes. Use keywords to fence products and compliance. Use first-party conversions, Enhanced Conversions, and CRM lists as the signals that power bidding. Dropping the query map in FS is how you buy jobs, complaints, and comparison-site research.

Is Performance Max a substitute for Search on current accounts and ISAs?

Not at the start. PMax can take brand and existing-customer lists once Search negatives and landing pages are honest. If PMax “drives the bulk” before you have a clean primary conversion, you will scale the wrong applications. 4x-8x ROAS shows up in optimised accounts, not in accounts that convert on brochure clicks.

Why does automated bidding feel like it inflates budget?

Because the conversion you gave it is too cheap to obtain and too far from a funded product. Fix the event, wait through a learning window, and stop daily structural tweaks. Manual CPC on a new account is a short diagnostic, not a philosophy.

How do we prove ROI when last-click looks broken?

Import funded accounts and completed mortgages as offline conversions. Report Search as a contribution to those outcomes, not as a last-click scoreboard. Attribution at channel level is still useful for negatives and copy. It is a weak CEO answer on its own. Plan 3-6 months before you call a structure “done”.

What should we refuse to promise in a pitch?

A guaranteed CPL or ROAS before anyone has seen the account, the KYC drop-off, and the landing pages. No serious Search team can honestly lock that. Ask who will actually manage the account, how negatives are handled, and whether traffic ever lands on the homepage.

Putting it to work

This week

• Pull 90 days of search terms and tag job, comparison, complaint, and out-of-area rows for pause or negative.

• Confirm the primary conversion is application start or booked appointment, not a guide download.

• Point each live RSA at a product URL and rewrite headlines to name the product and a factual £ or % figure.

• Run the Core Web Vitals checker on the mobile application URL and note INP on the first form step.

• Add UTMs per product with the UTM link builder so ISA, loan, and current account traffic stay separable in GA4.

Next 30 days

• Import one offline funded-account conversion and compare it to in-platform CPA.

• Split mixed “savings” campaigns into cash ISA versus other savings if search terms show Lifetime or HTB leakage.

• Add a postcode or eligibility gate if you are a credit union with a hard common bond.

• Give bidding two conversion cycles without a restructure, then cut or scale on funded outcomes.

Start by exporting those search terms and lining them up against products you can actually open in the UK this quarter. If the messy part is the handoff from query and ad copy to an application that underwriting will accept, HeyLead can run that Google Ads build, conversion design, and landing-page work as an ongoing Search programme rather than a one-off restructure. Chat with Martin on WhatsApp