Google Ads for banks and credit unions in Australia

A growth lead at a member-owned bank in Geelong can watch “home loan comparison rate VIC” spike after a cash-rate headline and still book almost nothing that underwriting will fund. The click is cheap enough on paper. The enquiry is a refinance tourist who already has a broker, or a rate-chaser who will never pass serviceability. Australian Search for ADIs is not “more form fills.” It is a fight over who owns the next funded home loan, term deposit, or personal loan in a market where comparison-rate rules, responsible-lending copy, and APRA-adjacent brand safety sit on every RSA.

Google Ads still averages about 4.2x ROAS across industries in 2026 benchmarks, while Search CPC has drifted near a global blended average of ~US$2.96 (up roughly 12% year on year) - Australian financial-services CPCs for home-loan terms commonly run $15-$30 AUD, making the gap between global benchmarks and your account significant. Those averages do not describe a mutual in regional Queensland bidding against the majors for “first home buyer loan.” You are paying for intent that looks identical in the auction and splits in the branch: pre-approval ready versus “just looking at offsets.” If your account treats them as one conversion, Smart Bidding will keep buying the cheaper, worse one.

This piece is a practical build for marketing leaders at Australian banks, credit unions, and mutuals. It is Search-first: how queries cluster, how you keep 20-30% of SEM budget from leaking into junk, how landing pages survive ASIC-style scrutiny, and how you measure funded accounts instead of brochure downloads. We will not pretend a homepage plus Performance Max is a strategy. If someone tells you they “just send traffic to the homepage,” that is your cue to leave the meeting.

What Australian savers and borrowers actually type before they pick a lender

The query is rarely “best bank.” It is “offset account vs redraw,” “first home guarantee eligibility,” “personal loan 15k no security,” or “term deposit 12 month rate credit union.” Product language is the targeting. A credit union with a common bond in the Hunter Valley will never win “home loans Australia” against the majors on CPA. It can win “teachers credit union home loan NSW” and “salary packaging car loan members” if the ad and the page say the same thing.

Map intent in four buckets, not fifty ad groups. High-intent product: branded plus product (“[brand] variable home loan”), non-brand product with a number or a constraint (“home loan 90 lvr,” “car loan balloon payment”). Research that still converts if the offer is a calculators-and-appointment page: comparison rate, LMI, offset. Waste: jobs, login, BPAY, “how to dispute a transaction,” broker-only language you cannot serve. Mobile already takes 63%+ of paid search clicks globally, and conversion still lags desktop by 30-40%. For loans, that gap is often a 40-field application on a 4G connection, not “mobile users are less serious.”

Voice and conversational queries matter more than most financial accounts admit. Plenty of 25-34 year olds ask Google Assistant the way they would ask a parent: “can I get a home loan on a casual contract.” Your RSA has to answer that without promising approval. AI Overviews now sit on a large share of searches and have been shown to cut CTR sharply when they appear. You still bid. You just stop treating impression share as the win. The win is a booked appointment with a complete fact-find, not a zero-click summary of the cash rate.

Negatives are not a one-off list. They are a weekly harvest of search terms that look financial and are not your product: “CBA login,” “NAB app down,” “ASIC,” “financial counsellor,” “hardship,” “class action,” “jobs.” High CPC pressure makes people flip new accounts back to manual CPC. That is a signal problem, not a bidding-philosophy problem. If Enhanced Conversions are off and you are losing 20%+ of conversion data to ad blockers and privacy changes, automated bidding will look drunk. Fix the signal before you blame the algorithm.

If your current partner cannot show how home-loan Search is split from deposits and from “general brand,” you do not have a media plan. You have a blended CPA that hides the product that is actually printing margin. A short conversation on Google Ads management is enough to see whether the account is built around funded products or around form volume.

How a compliant Australian Search account is actually structured

Fewer campaigns, broader match, tighter assets - and a compliance sign-off path that does not require three rounds of legal review every time a rate changes. That is the 2026 version of “signals over keywords,” not the 2020 version of 80 exact-match ad groups. You still need keywords. Google has not replaced them. You feed the machine product clusters, geo that matches your licence and branch footprint, and conversion actions that underwriting recognises. Performance Max can drive the bulk of spend in some retail books. For regulated lending, keep a Search core you can explain to risk, then let PMax sit on brand-plus-remarketing only after the conversion taxonomy is clean.

AI Max style campaigns are showing 12-18% lower CPCs versus standard Search in some 2026 accounts. That saving disappears if the landing page is the corporate homepage and the primary conversion is “newsletter.” Dedicated pages aligned to the ad: comparison rate in the first screen, eligibility in plain English, a calculator or a time-bound appointment, not a PDF of the PDS buried under a mega menu. Agencies that skip that step are why 20-30% of SEM budgets still die on irrelevant queries and mismatched pages.

Creative in Search is RSA plus sitelinks plus the page. “Creative is your targeting” is a Meta line, but the same idea applies: the headline that names First Home Guarantee or a member-only car loan is doing the audience work. Do not rotate “trusted since 1954” against a query that asked for a rate. Compliance will still own the legal line. Marketing owns whether the first 90 characters match the keyword.

Where Australian loan ads lose the lead after the click

The auction is not where most books bleed. The page is. You paid for “home loan 90 percent lvr” and dropped people on a brand film with a “explore our range” carousel. Last-click still counts a form. The lender interview never happens. Session recordings make this painfully obvious: rage clicks on a rates table that is an image, abandon on the employment-type field, scroll that never reaches the comparison rate. That is the job of HeyLead Insights on the apply path, not another dashboard of CPC.

A mid-size mutual in South Australia ran Search into the main home-loans URL for eleven months. Cost per enquiry looked “acceptable” at a blended figure the board liked. Branch managers said the calendar was full of tyre-kickers. The fix was not more budget. It was a single page that asked for occupancy, estimated deposit, and a 20-minute video call, plus a primary conversion that fired only when those three fields were complete. Enquiry volume dropped. Funded settlements over the next quarter did not. One mechanism: stop bidding on incomplete tyre-kicker forms.

Mobile lag is operational. If the desktop apply path converts and mWeb does not, you do not have a “channel problem.” You have a keyboard and identity-verification problem. Prefill from a logged-in member area when you can. Cut fields that credit will not use until conditional approval. 55.9% of sites passing Core Web Vitals is the wrong benchmark for a bank. Your apply origin should be in the pass column, full stop.

Attribution will not save a bad offer. Last-click double-counts across brand Search and PMax. In-platform CPA will diverge from the core banking extract. That is not a reason to switch off tracking. It is a reason to treat Google Ads as a tactical scoreboard and to reconcile weekly against funded accounts in AUD, by product. Unified marketing measurement is the CFO conversation. Enhanced Conversions and a clean CRM ID are how you keep Smart Bidding from going blind while you have that conversation.

When the page and the RSA finally match, you can afford to “run broad” inside the product cluster. Broad without a strong funnel is how you buy “loan calculator excel.” Broad with a page that names LVR, occupancy, and a booked call is how you let Google find the queries you would never think to add, including conversational ones. The risk is real if creatives and funnels are weak. The answer is not 2019 exact match forever. It is a funnel that deserves broad.

Two Australian Search patterns that keep showing up in ADI accounts

Pattern one: cash-rate week. A regional bank in Victoria dumped extra budget into “home loan rates” the morning the RBA moved. CPCs jumped, brand Search soaked most of the increment, and non-brand converted at a fraction of the prior week because every aggregator and broker was in the same auction. The better move is a prepared RSA set that leads with “talk to a lender today” and a same-week appointment inventory, not a 0.10% rate brag the legal team will anyway strip. Hold non-brand CPA caps. Let brand absorb the curiosity traffic. You are buying interviews, not news-cycle vanity.

Pattern two: the common-bond account that copies a major. A credit union with a health-sector bond copied “home loans Australia” structure from a template. Half the spend hit postcodes they do not serve and occupations they cannot verify. After they geo-fenced around hospitals and member postcodes, and after RSAs named the bond, cost per funded loan fell ~35% over two quarters while impression share dropped from 18% to 11%. Volume is a dead metric here. Information that matches who you can actually write is the only thing that moves the needle.

In both cases the team wanted a ROAS promise on day one. No one can guarantee a specific ROAS or CPL before running campaigns in your account, on your pages, against your credit policy. Performance Max ROAS of 4x-8x shows up in optimised retail accounts. Lending is slower and messier. Plan for 3-6 months of signal hygiene, then scale. Paid media retainers in this work are not cheap because the compliance and tracking load is not cheap. Treat that as a feature of the category, not a procurement failure.

What marketing leaders are seeing

Two things we hear repeatedly from ADI marketing teams:

We were reporting cost per application start and calling it a win until credit showed us 4.2 funded loans from a month that looked ‘on target’ in Google Ads. The conversion action was a 3-field form. We killed it as primary the same afternoon.

Cash-rate Tuesday used to mean we doubled non-brand. We stopped. Brand Search and appointment inventory take the spike now. Non-brand keeps the CPA cap. Funded volume is less jumpy than the news cycle.

FAQs

Are generic home loan keywords worth bidding on for credit unions in Australia?

Only with a hard CPA cap, tight geo, and a page that states the common bond in the first screen. Otherwise you are subsidising aggregator traffic you cannot fund. Brand plus member-occupation modifiers usually produce cleaner interviews.

Does automated bidding just inflate spend without better quality?

It will if your primary conversion is a weak form and Enhanced Conversions are off. With funded-stage or booked-interview signals, it usually beats panic manual CPC after the account has enough volume. Do not flip strategies every week or you reset learning.

Where should Performance Max sit in a regulated book?

After Search is clean. Use PMax on brand, deposits, and remarketing first. Keep high-scrutiny lending in Search so legal and credit can see the query. “Performance Max drives the bulk now” is true for some retail. It is a later chapter for most ADIs.

How do we handle comparison rates in RSAs without killing CTR?

Put the product and the next step in headline 1. Put the comparison-rate and fee line where policy requires it. Pin only what legal insists on. Preview the SERP snippet before you ship so you are not truncated into a claim you did not mean.

How often should we report on Google Ads performance for a bank or credit union?

Weekly on spend, CPA by product, search-term waste, and booked interviews. Monthly on funded accounts and margin. Monthly-only media reporting is how issues hide for 30 days in a category where CPCs move with the RBA.

Putting it to work

This week

• Export 90 days of search terms and mark waste versus fundable home-loan, personal-loan, and deposit intent.

• List every primary conversion in Google Ads and demote anything that is not an application start, booked interview, or qualified call.

• Confirm Enhanced Conversions and UTMs on the apply URLs with the UTM link builder.

• Run the apply URL through the Core Web Vitals checker on mobile.

• Draft one RSA set per product cluster that a compliance officer can approve in a single pass.

Next 30 days

• Launch or rebuild dedicated landing pages so ads never hit the corporate homepage.

• Reconcile Google Ads conversions to funded accounts in AUD by product, not to MQL counts.

• Freeze campaign structure long enough for bidding to learn, then scale only the clusters that fund.

• If the plumbing stalls, use a free audit as a second set of eyes on tracking and intent.

Start by lining last quarter’s Google Ads “leads” against settlements in the core banking extract, by product, in AUD. The gap you see is the whole job: Search that buys interviews credit will write, not forms the dashboard likes. HeyLead can own that Australian banking Search loop, from query clusters and negatives through apply-path conversion and funded-account reporting, so your team is not rebuilding RSAs and conversion actions between cash-rate announcements. Chat with us on WhatsApp