Category
Lead Generation & Ads
Publish Date
8 July 2026
Open most mortgage ad accounts and you find the same pattern: a dozen or more ads, a few dollars a day spread across each one, and not a single one with enough data to say anything with confidence. The broker running that account is not testing. They are guessing twenty times instead of once, and paying for the privilege.
Twenty half-built ads is not a test, it is a distraction
A local mortgage budget is finite. Split it across twenty ads and each one gets a trickle of spend, which means each one takes weeks to reach a sample size large enough to mean anything. Meta’s own guidance points at that problem directly: an ad set generally needs somewhere around 50 conversions in a week for the platform’s delivery system to optimize well. Spread thin enough, no single ad in a twenty-ad account ever gets there.
The instinct behind running twenty ads is usually reasonable. A broker does not know in advance which hook, offer, or angle will resonate in their zip codes, so they hedge by launching everything they can think of. The problem is not the hedge, it is what happens next. Instead of watching a handful of genuinely different ideas long enough to see a winner emerge, the account stays a scattershot of half-funded tests indefinitely, and every dollar goes toward maintaining the spread instead of scaling anything.
The fix is not fewer ideas up front. It is fewer ideas running at once, funded enough that each one actually gets tested, followed by real concentration on whatever wins. Three to five distinct concepts, each getting real spend, will tell a broker more in ten days than twenty ideas will tell them in ten weeks.
An angle is a different idea, a variation is a different outfit
The single most common mistake in a scattered ad account is testing the wrong thing. A broker changes the headline font, swaps a stock photo, tightens the copy by a sentence, and calls that a new test. It is not. That is a variation, a different outfit on the same idea, and variations rarely move the needle enough to matter.
An angle is different. A rate-drop urgency hook, a first-time buyer education piece, a cash-out renovation pitch, a refi savings calculator, a local market update. Each one appeals to a different mindset and a different stage of readiness. Testing angles against each other is how a broker learns what their local audience actually responds to. Testing fonts against each other mostly tells you nothing, because the underlying idea was never the variable.
The order matters too. Find the winning angle first, then build variations of it. A broker who starts by polishing the visuals on an idea nobody wants is optimizing something that was never going to work. Once an angle proves itself, that is exactly when small variations start paying off, because now the broker is refining a proven concept instead of guessing at an unproven one.
A campaign has not been tested until it has spent enough to prove it
Killing an ad after a day or two of lukewarm results is one of the most expensive habits in a mortgage ad account, because it throws away data before there was enough of it to read. Industry practice generally treats a test as unreliable before it clears two benchmarks: enough spend and enough time. A common rule of thumb is to hold off judging a campaign until it has spent roughly twice its target cost per result, and to give it close to a full week regardless, since the first several days often reflect Meta’s algorithm still exploring rather than a true read on performance.
That framework matters more in mortgage than in most verticals, because local mortgage cost per lead already swings widely by market. Suburban and rural campaigns can run leads in the teens to low double digits, while dense metro accounts commonly land well into the double or even triple digits per lead. A broker judging a campaign against a number pulled from a national benchmark, instead of their own market’s real cost per result, will kill winners and keep losers based on the wrong yardstick entirely.
The practical version of this rule: pick a target cost per result before launch, let each angle spend several multiples of that target over the better part of a week, and only then decide what stays and what goes. Anything judged sooner is a guess dressed up as a decision.
Cost per lead is the easy number to watch, and the wrong one
Cost per lead updates in real time and feels like a scoreboard, which is exactly why it is dangerous to optimize against on its own. A campaign can post a low cost per lead and still be a bad campaign, because a form fill is not revenue. It is a person who may or may not be a real buyer, may or may not answer a follow-up call, and may or may not ever book.
The number that actually predicts revenue is cost per booked call, and it does not always move in the same direction as cost per lead. An angle that produces expensive leads can still produce cheap booked calls, if the people it attracts are unusually qualified and ready to talk. An angle with a rock-bottom cost per lead can produce almost no booked calls at all, if it is pulling in curious clickers instead of real buyers. Purchased digital mortgage leads close at low single-digit percentages industry-wide, which means the gap between a lead and a funded loan is where most of a campaign’s real cost lives, whether a broker is watching that gap or not.
This is where the Empire OS methodology draws a hard line: judge every angle on cost per booked call, not on the number that happens to update first. A broker who only watches cost per lead will cut the angle that was actually working and keep the one that was quietly wasting the budget.
Scale the winner without breaking it
Once an angle proves itself on real spend and real booked calls, the temptation is to raise its daily budget and move on. That is usually the fastest way to break a campaign that was finally working. Meta’s delivery system treats a large budget jump as a fresh restart, and a winning campaign can lose its footing exactly when a broker is trying to lean into it.
The more reliable path is to duplicate the winner instead of editing it. Copy the winning campaign and raise the budget on the copy, leaving the original untouched, to see if the account can absorb more spend at the same cost per result. Separately, take the winning creative into a new but related audience, or the winning audience into a related new angle, to find more demand without starting from zero. Save the biggest swings, an entirely new concept for a different market segment like refi alongside purchase, for once the easier scaling moves are already working. Each of these protects the proven baseline while the broker tests how far the winner can go.
This is the part most brokers skip, because it takes more discipline than creativity. Finding one winning ad is a real accomplishment. Turning that one ad into a repeatable source of booked calls, without accidentally resetting the very campaign that got you there, is the part that actually grows a brokerage. It is also the part that is hard to run alone while also originating loans, which is exactly the gap a done-for-you ad build closes: someone dedicated to funding each angle properly, judging it on the number that matters, and scaling the winner the right way instead of the fast way.
If your ad account has twenty ideas and no clear winner, that is worth a conversation, not another ad. Book a call and we will look at what your account is actually telling you.
