Retargeting Campaigns for Borrowers Still Comparing Lenders

The Window Between the First Click and the Signature

Almost nobody applies for a mortgage on the first visit. A borrower reads a page, calculates something, closes the tab, and then spends the next several weeks doing exactly what the Consumer Financial Protection Bureau tells them to do: collecting Loan Estimates from three to five lenders and comparing them line by line. That window is the entire opportunity, and it is also the reason a mortgage campaign that measures itself on first-session conversion will conclude, wrongly, that the traffic was bad.

Mortgage retargeting exists to occupy that window. It is not a discount chase and it is not a nag; it is the recognition that the broker who is still present and still useful in week three is the one whose Loan Estimate gets read carefully. The competitive pressure is real — the borrower comparing five lenders is being courted by four others — and the differentiator available to a broker is not price theatre but explanation, because the borrower staring at five documents mostly cannot tell which one is actually cheaper.

What a List May Contain, and What It May Not

The restricted-category rules that govern prospecting do not evaporate at the retargeting stage. On Meta, a housing or credit campaign cannot use Special Ad Audiences, which no longer exist, and cannot narrow by age, gender, or ZIP code even when addressing people who have already visited the site. Custom audiences built from the broker's own first-party signals — site visitors, video viewers, people who opened a lead form — remain available and become the backbone of the account precisely because the prospecting side has so little else. On Google, remarketing lists remain usable under the housing, employment and credit policy provided the list itself was not assembled from prohibited demographic or ZIP signals.

There is a second constraint that is easy to miss. A retargeting list segmented by the loan product a visitor looked at is fine. A list segmented by anything that functions as a proxy for a protected characteristic is not, regardless of how it was labelled in the tag manager. The safe framing is behavioural: what the person did on the site, which pages they read, how far into an application they got. That is also the more useful framing, because a visitor who abandoned an application at the income step needs a different message from one who read the VA page for four minutes and left.

Segmenting by How Far the Borrower Actually Got

A single retargeting audience serving a single ad to everyone who touched the site is the version most brokers run and it is barely worth the spend. The list divides naturally into stages, and each stage wants a different thing.

Someone who read a loan-program page and left is early: they are still deciding what kind of borrower they are, and the useful follow-up is educational — a short explanation of what separates FHA from conventional, or what a VA entitlement actually covers. Someone who used a calculator or a rate widget is deep in rate shopping: they are doing arithmetic, and the follow-up that lands is the one that helps them finish it, including the awkward truth that a no-fee structure is not free and that a discount point costs 1 percent of the loan to buy roughly a quarter point of rate. Someone who began an application and abandoned it is late and expensive to replace: the follow-up here should be a person, not an ad, and paid retargeting should mostly be backing up the phone call rather than substituting for it. And someone who has already received a Loan Estimate is in a documented comparison, which is the highest-stakes segment in the account and the one where being absent is fatal. Segmenting by how far the borrower got is the discipline a marketing plan for mortgage loan officers should specify.

Advertising Into a Rate Shopping Window Without Quoting a Rate

The obvious retargeting creative is a rate, and the rate is the one thing the format handles worst. A number that changes twice a day cannot survive an ad review cycle, and a simple annual rate stated in an ad drags the annual percentage rate onto the creative at equal prominence, which is difficult to achieve inside a small display unit without producing something illegible.

The way out is to compete on the thing the borrower is actually struggling with. During rate shopping the borrower has several Loan Estimates in front of them and no reliable way to compare them, because lender A's low rate is funded by points that lender B did not charge, and lender C's zero origination fee has been baked into a rate that costs more over seven years than the fee would have. An ad that offers to walk through their existing Loan Estimates, page 2 line by line, is not quoting anything, triggers no disclosure obligation, and speaks directly to the moment they are in. It also positions the broker as the person doing the comparison rather than the one being compared — a meaningfully better place to stand when a borrower is holding five documents. What can and cannot be said inside the creative itself is dealt with in running Meta campaigns under the housing restrictions, where the disclosure constraints bite hardest on small square assets.

Frequency, Duration, and the Rate-Lock Cliff

Retargeting windows in most industries are set by habit — thirty days, ninety days, whatever the template said. In mortgage the windows should be set by the borrower's actual clock, and there are two of them. A purchase borrower who requested a pre-approval may be ninety days from a closing and should be held in the audience for the whole of it, because a house hunt that stalls in April can restart in June with no warning. A refinance prospect is on a much shorter fuse: their interest exists because of where rates are this month, and holding them in an audience for six months means paying to show ads to people whose reason for looking has evaporated.

Frequency deserves the same discipline. A borrower who has been contacted by phone and email and is also seeing the brokerage in every ad slot on the internet is not being nurtured, they are being pursued, and the effect on a decision worth several thousand dollars in fees is not neutral. Caps belong in the campaign settings, and exclusions belong in the pipeline: anyone who has applied, anyone who has been declined, and anyone who has closed should be suppressed from prospecting lists the moment the loan origination system knows about it. Keeping those suppressions current is a data job that runs alongside automated follow-up while the borrower is still comparing.

What This Channel Should Be Judged On

Retargeting reports flatter themselves more than any other line in a media plan. The audience is composed of people who already know the brand and were already fairly likely to come back, so a naive read of last-click attribution will credit the retargeting campaign with conversions it merely stood next to. Mortgage retargeting is worth funding, but it has to be measured with more suspicion than it usually gets.

The measurement that survives scrutiny is a holdout: withhold the campaign from a random slice of the eligible audience and compare funded loans, not clicks, across the two groups over a full ninety-day cycle. That is a slow test and most brokerages will not run it, in which case the fallback is to judge the channel on whether application starts rise among returning visitors while the campaign is on and fall when it is off. What should never happen is the channel being scaled because its cost per conversion looks best in the dashboard, which it always will, since it is buying back people who were already coming. Insurance agencies working the same slow, quote-driven comparison problem have arrived at similar defences, which are worth reading in the way insurance agencies keep quoting prospects in view.