Google Search Ads for Purchase and Refinance Intent

The Most Expensive Clicks Google Sells

Mortgage is one of the handful of categories where a single click can cost more than a month of advertising in an ordinary local business. Bids are pushed there by the size of the prize: a $250,000 file at a broker fee of 1 to 2 percent is worth $2,500 to $5,000, so every lender, aggregator, and lead reseller in the country can justify a bid that would be insane anywhere else. A broker entering this auction with a generalist's instincts — broad match, a homepage as the destination, a conversion goal set to page views — will spend a month's budget in a week and have nothing in the pipeline to show for it.

What makes the channel worth the price is that intent is legible at the query level in a way it is nowhere else. A person typing a question about VA eligibility has disclosed their loan type, their veteran status, and their stage of research in four words, and no amount of demographic targeting elsewhere would have found them. The entire discipline of mortgage search campaigns is deciding which of those disclosures are worth paying for and building the account so that the ones that are not never cost anything.

Five Tiers of Query, Priced Differently

Brand queries are the cheapest and the most misunderstood. Somebody searching the brokerage by name has usually been referred by an agent or a past client, and the click costs almost nothing — but the broker is often paying to appear above an organic listing that would have been clicked anyway, and the honest answer is to bid on brand mainly to keep a competitor from sitting on top of it.

Loan-type queries are the core of the account. Questions about FHA down payment minimums, VA entitlement, USDA eligibility maps, and jumbo thresholds separate borrowers into groups that want completely different conversations, and each deserves its own ad group, its own headline, and its own landing page. Pre-approval queries are the highest-value purchase intent available: someone searching for a pre-approval is usually inside a real house hunt with a deadline, and this is the query worth losing money on for a quarter. Rate-shopping queries are the trap. They carry enormous volume, they are dominated by comparison sites with budgets no brokerage can match, and the searcher is frequently three months and several lenders away from doing anything. Refinance keywords are their own tier, and their value swings entirely with the rate environment — the same query that funds three files in a falling market will fund none in a flat one.

The Negative Keyword List Is the Profit Centre

More money is saved in this vertical by the queries an account refuses than by the ones it wins. Mortgage attracts an enormous volume of traffic that looks commercial and is not: people researching how to become a loan officer, students working through amortization homework, homeowners hunting for a servicing phone number to make a payment, journalists reading rate news, and job seekers looking at broker salaries. Every one of those searches contains the words a mortgage campaign bids on, and every one of them will happily cost thirty dollars.

A serious negative list starts before launch rather than after the first invoice. Employment terms, education terms, definition terms, calculator terms, payoff and servicing terms, and the customer-service queries attached to the big national lenders all belong there on day one. Broad match with automated bidding, which works acceptably in low-stakes categories, is genuinely dangerous here until the account has been trained on downstream conversion data, because the algorithm will pursue cheap conversions with total sincerity and cheap conversions in this vertical are almost always the wrong ones. Starting on phrase and exact match, harvesting the search terms report weekly, and expanding deliberately is slower and it is the only version that survives contact with the invoice. The negative keyword list carries more weight here than the bid strategy, which is true of commercial mortgage digital marketing as well.

What Google's Credit Policy Takes Away, and How the Account Compensates

Google files mortgage advertising under its housing, employment and credit policy, which strips the targeting levers most search accounts lean on. Gender, age, parental status, marital status, and ZIP-code targeting are unavailable. Geography is limited to broader units — city, county, DMA, state — and privacy-safe radius targeting, so a broker who wanted to concentrate spend on three affluent postcodes cannot. Remarketing lists are usable, but only where they were not assembled from the prohibited signals.

The compensating mechanism is the query itself, which is why keyword structure carries so much weight in this channel and audience structure carries so little. A campaign cannot ask Google for high-income households, but it can bid on jumbo loan queries, and the jumbo query is a better signal than the income proxy ever was. It cannot exclude renters, but a pre-approval query filters them out by definition. Alongside the targeting rules sit the financial-services advertising requirements, which oblige the advertiser to be verified and to be transparent about who they are, and the Regulation Z rules that govern what a headline may claim — an ad that names a rate has to carry the annual percentage rate at equal prominence, and an ad that names a payment or a down payment figure has pulled the full disclosure set onto itself. The practical consequences for what can actually be written into thirty characters are worked through in headlines built around Regulation Z triggering terms.

Bidding Toward the Closing Rather Than the Form

Smart bidding optimizes toward whatever conversion the account tells it to value, and the default choice — a form submission — is the one that produces the worst outcome. The algorithm will find the cheapest available form submissions, and the cheapest form submissions in mortgage come from rate-curious browsers who submit forms on four sites in an evening. An account bidding this way looks superb in the interface and starves in the pipeline.

The alternative is to feed the platform the events that happen weeks later. When an inquiry becomes a credit-pulled application, and later a funded loan, those stages get uploaded back into Google as offline conversions against the original click, and the bidding model gradually learns which queries produced files rather than which produced forms. This requires the click identifier to survive from the first search into the loan origination system and back out again, which is a data plumbing job rather than a marketing one and is set out in carrying the click ID from first search through to closing. Until that pipe exists, the honest approach is manual: run maximize conversions with a hard budget cap, review the search terms report against actual closings each month, and accept that the account is being steered by a human rather than by the machine.

Two Campaigns, Two Budgets, Two Very Different Clocks

Purchase campaigns run on the housing calendar and on a ninety-day lag between pre-approval and funding, which means their budget should be steady, their conversion event should be the pre-approval, and their performance should be judged a quarter after the spend. Refinance campaigns run on the rate environment and on nothing else. When rates drop, refinance keywords go from dormant to fiercely contested inside a week, and a campaign that cannot absorb several times its normal budget on short notice will watch the window close from the sidelines.

Running both from a shared budget guarantees the volatile side eats the stable one, and the specific damage is invisible for months: the purchase pipeline that starves in a refi surge does not show up as a problem until the quarter it was supposed to fund. Separate campaigns, with separate daily caps and a written rule about what happens to each when rates move, are the structural fix. Both sides also need a landing page that matches the query — a VA loan click that lands on a homepage has wasted a click that cost more than most businesses pay for fifty. Advisers selling a different fee-based financial service run into the same tension between expensive clicks and slow closings, and their approach is worth comparing in how advisors buy leads for a fee-based service. Managed properly, mortgage search campaigns end each month with a number the broker can defend: not clicks, not leads, but files that funded and what each one cost to find.