Google Ads for banks and credit unions

A branch marketing lead at a $1.8B credit union can spend $18,000 in a month on Search and still not know whether those clicks funded checking accounts, HELOCs, or auto loans. The dashboard shows form fills. The CFO asks for funded members. Those are not the same number, and Google Ads will happily optimize for whichever conversion you feed it.

Financial institutions buy paid search because the intent is already in the query: “business checking near me,” “HELOC rates,” “refinance auto loan credit union.” That intent is expensive, regulated, and easy to poison with the wrong negatives, the wrong landing page, and a conversion action that fires on “start application” instead of “submit application.” This piece is about running Google Ads so banks and credit unions pay for funded pipeline, not brochure traffic.

Why banks and credit unions burn Search budget on the wrong intent

You already know branded terms convert. The leak is everything around them. Job seekers type “teller jobs” and “credit union careers.” Existing members type “online banking login” and “lost debit card.” Rate tourists type “best CD rates 2026” with no branch relationship and no intention to open. If those queries sit in a broad match campaign next to “open a checking account,” Smart Bidding treats every click as equally useful. It is not.

Search CPC averaged $2.96 across industries in 2026, up 12% year over year (financial services deposit and lending terms typically run $6-$14 depending on market, with metro HELOC terms pushing higher, per WordStream and Tinuiti FSI benchmarks). Deposit and lending terms sit well above that floor. When 20-30% of SEM budgets routinely disappear on irrelevant queries, neglected negatives, or mismatched landing pages, a mid-size institution can waste a five-figure monthly line item without a single policy violation. Automated bidding does not fix a dirty query report. It accelerates it. High CPC pressure is why plenty of new financial accounts still get pulled back to manual CPC in week one: the algorithm has no clean funded-account signal, so it buys volume.

Mobile accounts for 63%+ of paid search clicks globally, yet conversion rates often lag desktop by 30-40%. For a credit union, that gap is not a “mobile problem.” It is an application-flow problem. A 14-field membership form that asks for SSN, employment, and overdraft preference on a phone will lose the member you already paid $40-plus to attract. Only 55.9% of origins pass all three Core Web Vitals in May 2026 CrUX data. If your apply page fails INP on a mid-range Android, you are paying Google for sessions that never reach submit.

Last-click reporting makes this worse. A member sees a HELOC ad, visits the rates page, calls the lending desk 11 days later, and funds in branch. Google Ads may never see the conversion. Privacy changes and ad blockers already strip 20%+ of conversion data. Feed that incomplete picture into tCPA and the system hunts cheap form starts, not funded loans. You get a pretty CPL and a quiet pipeline.

Performance Max will drive the bulk of spend if you let it. For banks and credit unions, that is often the wrong first move. PMax mixes Search, Display, YouTube, and Gmail. Financial services policy, restricted creative, and “guaranteed rate” claims get you disapproved in places Search never would. Run Search with tightly mapped intent first. Add PMax only when conversion actions are funded-account quality and the asset group cannot wander into “personal loans for everyone” Display.

If your agency’s plan is “we just send traffic to the homepage,” stop. Homepages argue for the brand. Ads argue for one product. Mix them and you teach Google that a click on “open business checking” is a success when the visitor reads the CEO letter and leaves. Dedicated landing pages aligned to ad messaging are table stakes, not a CRO luxury. If you want a second set of eyes on account structure before you scale spend, HeyLead’s Google Ads management work starts with query intent and conversion quality, not a homepage dump.

A Search-first playbook from intent clusters to funded accounts

Do not rebuild the account every quarter. Constant structure changes reset learning. Build fewer campaigns, map them to products that actually fund, and keep negatives ruthless. Below is the operating checklist we use when a bank or credit union already has a Google Ads account that “looks fine” and still cannot explain funded accounts.

A regional bank CMO we worked alongside had one Search campaign covering “checking,” “mortgage,” and “wealth.” tCPA looked efficient at $41 per lead. Lending ops said 8 of 10 “leads” were CD rate screenshots and login help. Splitting products, adding login/jobs negatives, and moving the primary conversion to submitted application dropped lead volume 37% in 18 days and raised funded checking plus booked mortgage consults enough that they restored budget. The mechanism was not a new channel. It was refusing to let Google treat a login click like a new household.

Where funded-account tracking actually breaks after the click

Paid search for banks and credit unions fails most often after Google has done its job. The visitor lands. The headline on the page does not match the ad. The rate table is three clicks deep. The apply button opens a PDF. Or the Instant Form equivalent is a 19-field membership app that asks for mother’s maiden name before it explains overdraft. Traffic did not fail. The handoff failed.

Watch the page the way a member uses it. HeyLead Insights is built for that: session recordings, heatmaps, scroll depth, form abandon, click patterns. You want the exact field where HELOC applicants stall, not a monthly slide that says “bounce rate 62%.” If legal requires a long disclosure, put it next to the CTA, not as a wall before the first field. If membership eligibility is a credit union constraint, ask it early so you do not pay for a full application from someone outside the charter.

Attribution will not be pretty. CM360-style last-click from Google Ads has been getting noisier. Last-click counts the same conversion across channels. In-platform CPA will disagree with the core’s funded-account file. That is normal. Do not freeze the account every time the two reports diverge by 12%. Pick a business answer: funded accounts and booked lending appointments inside a 30- or 60-day window, with Google Ads as a directional input. Unified marketing measurement and simple incrementality (brand Search on/off in a non-core DMA, or geo holdouts) beat another dashboard argument. A funded checking relationship generating $800 in annual revenue over a 4-year average tenure implies a rational max CPA well above what a $41 tCPA target allows - model your own before anchoring to industry composites. Bid like it, but still require a clean application event so Smart Bidding is not guessing.

Zero-click search and AI Overviews now sit on a large share of queries. 68.01% of US Google searches ended without a click in early 2026 in one dataset, and AI Overviews on more than 20% of searches have cut CTR sharply on informational terms. That is mostly a problem for “what is a HELOC” content, not for “apply for HELOC [city].” Protect the apply cluster. Do not pour budget into educational queries that AI Overviews will answer without a click unless those pages are a deliberate remarketing pool with a real follow-up offer.

Reporting cadence matters more than another tool. Monthly PDFs bury a disapproved RSA for 25 days. Weekly: search terms, disapprovals, conversion volume vs funded lag, and which landing URLs dropped below a minimum CVR. No one can guarantee a specific ROAS or CPL before seeing your account, landing pages, and unit economics. Realistic lift in a regulated deposit or lending program takes 3-6 months of clean signals, not a week of Performance Max. If someone promises funded-member volume from the first reel-style YouTube asset, they are selling a 2024 playbook in 2026.

Keep the stack boring: Google Tag Manager, Enhanced Conversions, CRM or core import of funded status, UTMs that humans can read. “Clean conversion setup” is the phrase buyers already use because they have lived the opposite. When the apply path is this fragile, a short landing page pass on the three highest-spend URLs usually returns more funded accounts than adding Bing or another social network.

Two patterns we see repeatedly

Institutions that leave tCPA on “begin application” for months often hit the platform target every week while funded checking stays flat. Switching the primary conversion to submitted application and adding login and careers as negatives typically makes spend look worse for about two weeks, then membership actually has work.

PMax can look like 4x in the UI until names are matched to the core. A large share is often existing members hitting the debit-card page. Pulling PMax back and putting that budget on exact match “open business checking” with a dedicated apply URL is an uglier account and a better pipeline.

FAQs

Should we start with Performance Max or Search campaigns?

Start with Search mapped to products that fund. Add Performance Max when conversion actions are submitted applications or qualified appointments, asset text is policy-safe, and you can exclude brand cannibalization. PMax can reach 4x-8x ROAS in optimized ecommerce-style accounts. Deposit and lending accounts need tighter control first.

Is automated bidding a bad idea in financial services?

It is a bad idea on dirty conversions. It is a good idea once the primary action is quality and Enhanced Conversions are on. If you only have 12 conversions a month, stay on manual or maximize conversion value with a conservative cap until volume supports tCPA. Daily bid panic resets learning. Leave the algorithm alone for a full conversion cycle, often 2-3 weeks for applications.

How do we prove ROI when a checking account funds two weeks later?

Import offline conversions with a click ID and a realistic lag window. Report two numbers to the CFO: platform CPA on submitted applications, and cost per funded account from the core. Do not pretend last-click will reconcile them. Quality leads, not more leads, is the language that survives a 90-day lending cycle.

What if legal will not approve dedicated landing pages?

Get one template blessed: product name, required disclosures, eligibility, fee or APR caveats, and a single form. Reuse it for checking, savings, and HELOC with swapped modules. A homepage with 12 nav items is not a compromise legal should win. It is the most expensive page you can buy clicks for.

Who should own the Google Ads account?

The institution should. Agencies should have admin access, not ownership you cannot recover. Account lock-in is a real objection. Keep billing, MCC, and conversion actions under the bank or credit union’s Google customer ID.

Putting it to work

This week

• Pull 90 days of search terms and mark fundable vs login, jobs, and rate tourism.

• Demote “begin application” so it cannot train bidding.

• Point the top three spend ad groups at product URLs, not the homepage.

• Run Core Web Vitals and a SERP snippet preview on those URLs.

• Write a negative list for careers, login, routing number, and ATM.

• Confirm Enhanced Conversions is actually on, not just “planned.”

Next 30 days

• Split mixed product campaigns so checking and HELOC cannot share a bid strategy.

• Import a lagged funded-account or booked-appointment conversion from the core or CRM.

• Staff a weekly 30-minute review of disapprovals and search terms, not a monthly PDF.

• Watch apply-form abandon on mobile and cut one field or one step that does not affect underwriting.

• Only then test a tightly capped Performance Max asset group for one product.

Start by reconciling last month’s Google Ads “leads” to funded accounts in the core. The gap you find is the brief. HeyLead will take the unglamorous loop off your plate: query hygiene, policy-safe RSAs, apply-page alignment, and conversion imports so bidding optimizes for funded members instead of form starts. Chat with us on WhatsApp