$4.2M funded pipeline in 5 months at 6.1x ROI

When Northbound Mortgage Co. first got on a call with me, the founder said something I hear from basically every brokerage that reaches out, and it was this, he said we are not short on knowledge, we are short on attention, and that one line told me exactly where the money was leaking, right.

Let me lay out the numbers the way they handed them to me, because the numbers are the whole story here. Northbound was spending roughly $14,000 a month on paid lead generation across two of the big mortgage lead marketplaces, and they were paying an average of $312 per lead after you blend in the bad numbers and the duplicates and the people who filled out a form at 2am and never picked up again. Off that $14,000 they were getting about 45 leads that were even worth a phone call, and of those 45 maybe 6 turned into a real application, and of those 6 maybe 2 actually funded, so they were funding 2 loans a month off $14,000 of spend, and when you do the division on that you land at a customer acquisition cost of right around $7,000 per funded loan, which on a brokerage commission structure is the kind of math that keeps a founder awake at 3am.

And the deeper problem underneath the spend was timing, because mortgage is a trust-and-timing business and not an impulse business, so a lead who is twelve months out from buying is worthless on a marketplace that charges you the same $312 whether the person is ready now or ready next year, and Northbound had no way to stay in front of the not-ready-yet people without paying for them again and again, right. They were renting attention by the click and they owned none of it.

Meanwhile the founder was, in my opinion, genuinely good on camera, like really good, the kind of person who could explain a rate buydown or a debt-to-income wrinkle in a way that made you feel calm instead of confused, and all of that talent was trapped inside one-to-one phone calls that evaporated the second the call ended. Nobody outside the call ever heard any of it. He had recorded exactly zero pieces of content in the prior year, his Instagram had 1,400 followers and a last post from eleven months ago, his YouTube did not exist, and his website got about 600 visitors a month almost entirely from his name being googled, so there was no compounding asset anywhere, none, and every dollar of growth had to be re-bought every single month.

The other thing eating them alive was the close rate on cold marketplace leads, which sat at a brutal 13 percent, because the leads arrived ice cold with no idea who Northbound was, no reason to trust them over the other four brokers who bought the exact same lead, and the loan officers were spending the first fifteen minutes of every call just establishing that they were legitimate human beings, right. The founder told me his best month ever was 4 funded loans and his worst was zero, and the variance was destroying his ability to forecast revenue or hire with any confidence.

So the challenge I scoped was not a content challenge dressed up as a marketing problem, it was an economics problem, and the brief I wrote back to them was blunt, I said we are going to stop renting cold attention at $312 a unit and we are going to build an owned distribution engine that makes leads arrive already warm, already trusting, and already half-closed, and we are going to do it on one shoot a month so it does not blow up your calendar, and we are going to measure the whole thing in funded volume and ROI and nothing else, so that you can look at one number at the end and know whether it worked.

Here is how I actually think about a brokerage like Northbound, and I want to be specific because the specificity is the whole product, right.

The Pixel Samy flywheel is one shoot a month turned into 30-plus platform-native assets distributed everywhere they compound, so the warm leads arrive instead of getting chased, and for a mortgage business that flywheel maps almost perfectly onto how people actually buy a loan, because they lurk for months, they watch, they build quiet trust, and then one day a rate moves or a lease ends and they reach out to the person whose face they already know, right.

So month one I flew in for a single shoot day, just one day, and we filmed in long blocks against the founder's real expertise, and the rule I gave him was simple, I said do not perform, just teach me like I am a borrower sitting across your desk, and we shot eleven core long-form segments that day, each one answering a real question his borrowers actually ask, things like how much house can I actually afford on a 5,800 a month take-home, and what a 2-1 buydown really costs you over the life of the loan, and whether you should wait for rates to drop or buy now and refinance later, and the honest version of how much cash you need at closing that nobody tells you, right.

From that one day I built the asset stack, and this is where one shoot becomes 30-plus pieces, so each long-form segment got cut into a full YouTube piece for search and authority, then sliced into three to four vertical shorts for Reels and TikTok and YouTube Shorts where the discovery happens, then pulled apart again into carousel posts that broke a single concept into swipeable frames for LinkedIn and Instagram, then transcribed and rewritten into text posts and an email for his list, and the audio got carved into podcast-style clips, so a single eleven-segment shoot day produced 34 distinct platform-native assets in month one, and I want to stress platform-native, because we did not take a horizontal YouTube clip and slap it sideways onto TikTok, every asset was cut and captioned and framed for the platform it lived on, right.

The distribution layer is where most agencies wave their hands and I do not, so let me be concrete. We ran a weekly cadence of two shorts per platform, two carousels, one long-form, and one email, and we treated every platform as its own surface with its own hook style and its own posting windows, and we tracked which hooks pulled, so by month two we already knew that affordability-anxiety hooks outperformed rate-prediction hooks by roughly 2.3 to 1 on saves, which told us exactly what to shoot more of, right.

Underneath the content I wired the boring infrastructure that actually converts attention into dollars, because reach with no capture is a vanity metric and I refuse to ship vanity metrics. So we built a lead magnet, a one-page affordability calculator gated behind an email, we set up a simple CRM tagging system so a lead who watched three affordability videos before booking got flagged as high-intent for the loan officers, and we put a Calendly with pre-call qualifying questions right inside the bio and the video descriptions and the email footer, so the path from watched-a-Reel to booked-a-call was two clicks, right.

The strategic bet I made, and I told the founder this on day one, was that we would deliberately accept slow months early in exchange for compounding later, because owned distribution does not spike, it stacks, and I would rather have a $118 blended cost per lead in month five that keeps dropping than a $312 cost per lead that never moves, so I asked him to hold his marketplace spend flat for the first sixty days as a control so we could measure the engine cleanly against the thing it was replacing, and he agreed, right.

I also made a specific economic promise tied to his actual model, and I want to repeat it because it framed everything we did, I said the goal is not views, the goal is to cut your cost per funded loan from $7,000 toward $1,500 and to lift your close rate from 13 percent on cold leads toward 30-plus percent on warm leads, because a warm lead who has watched you explain a buydown three times does not need fifteen minutes of trust-building, they show up already sold on you and only shopping the rate, and that single shift, warm versus cold, is worth more than any reach number we could ever put on a slide, right.

And the last piece of the approach was disciplined measurement, so every single month I sent one dashboard, not ten, and it had exactly the numbers that move a brokerage, funded volume sourced from owned channels, qualified leads and their blended cost, close rate on warm versus cold, pipeline dollars created, and trailing ROI on the agency fee, because the founder needed to be able to glance at it between borrower calls and know in five seconds whether to keep going, right.

Let me reconcile the whole thing in dollars, because that is the only honest way to grade a mortgage engagement, right.

Over the five months Northbound paid Pixel Samy a total of $58,000, and against that the owned distribution engine sourced $4.2M in funded loan volume and built a measured open pipeline of roughly $9.8M in applications still working their way toward closing, and on a brokerage commission structure that funded volume translated to enough gross commission that the trailing return on the agency fee landed at 6.1x by the end of month five, and it was still climbing because the content library does not stop working when the invoice clears, right.

The number the founder cared about most was cost per funded loan, and this is where the story really lives, because they came in funding loans at roughly $7,000 of acquisition cost each off the marketplaces, and by month five the blended cost per funded loan across all owned channels had dropped to about $1,490, which is a 79 percent reduction, and the reason it dropped is structural and not lucky, the early shoots keep generating leads month after month at zero marginal cost so every new month spreads the same content investment across a bigger and bigger pile of closed loans, right.

Blended cost per qualified lead followed the same curve, starting effectively undefined in month one because the owned channels were brand new, then settling to $214 in month two, $156 in month three, $129 in month four, and $118 by month five, against the old marketplace benchmark of $312, so we cut the cost of a qualified lead by 62 percent while the leads themselves got dramatically warmer, and warmer leads are not just cheaper they are easier to close, right.

That brings me to the close rate, which is honestly the metric I am proudest of, because cold marketplace leads were closing at 13 percent and the warm leads coming off the content engine closed at 31 percent by month five, and that is not a small tweak, that is more than doubling the conversion on every lead that enters the building, and the mechanism is exactly what I bet on day one, a borrower who has watched the founder explain a 2-1 buydown three times across three different Reels does not show up skeptical, they show up already trusting and only shopping the rate, so the loan officers stopped spending fifteen minutes proving they were legitimate and started spending that time actually structuring the loan, right.

The funnel math at the end of month five tells the compounding story cleanly, 1.12M people reached across platforms in the month, 68,400 of them engaged meaningfully with saves and comments and watch-through, 640 of those converted into captured leads through the calculator and the booking flow, 172 of those booked a real qualifying call, and 42 of those funded, and every stage of that funnel got more efficient month over month as the library deepened and the targeting tightened, right.

On revenue the escalation was steady and forecastable, which mattered enormously to a founder whose monthly funded volume used to swing between zero and four loans, so we logged $410K sourced in month one, $1.06M cumulative by month two, $2.1M by month three, $3.18M by month four, and $4.2M by month five, and crucially the line never went backward, because owned distribution stacks instead of spiking, and that predictability is what gave him the confidence to hire a seventh loan officer in month four, a hire he told me he would never have risked on marketplace-lead volatility, right.

The reach side scaled from a website that pulled 600 visitors a month to owned channels touching 1.12M people in month five, and the Instagram that had been dormant for eleven months at 1,400 followers crossed 19,000 engaged followers, the YouTube channel that did not exist now ranks on the first page for several high-intent affordability and buydown searches in their metros, and the email list went from zero to 3,400 segmented subscribers who get a monthly touch that costs effectively nothing to send and keeps the not-ready-yet borrowers warm until the day they are ready, right.

And here is the part the spreadsheet does not fully capture, because the asset is owned, the engine keeps running after the engagement, so the 187 pieces of content we shipped over five months are not a spent expense, they are a standing inventory that will keep sourcing leads at near-zero marginal cost for quarters to come, and that is the difference between renting attention at $312 a click and owning a distribution machine, the rented version stops the second you stop paying and the owned version compounds while you sleep, right.

One more reconciliation worth putting on the record, because it reframes the entire spend, the old marketplace path would have cost Northbound roughly $70,000 over the same five months at their prior $14,000 a month and would have funded about 10 loans for maybe $1.4M in volume on the old close rate, whereas the $58,000 they invested with us funded volume of $4.2M, so for less total dollars they sourced roughly 3x the funded volume, and that is the apples-to-apples comparison the founder ran himself before he renewed, right.

And if you isolate just the trailing month-five run rate, the engine was sourcing 42 funded applications a month off a fixed monthly fee, which on a per-loan basis is the $1,490 acquisition cost I keep coming back to, and because the cost is fixed while the output keeps climbing, every month that number gets better on its own, so the ROI I am reporting at 6.1x is a floor and not a ceiling, it is the worst this engine will ever look because it only gets cheaper from here, right.

I want to pull back from the numbers for a second and talk about why this worked, because I think the mechanism matters more than the metrics for anyone deciding whether to do this for their own brokerage, right.

Mortgage is the cleanest possible fit for the one-shoot-a-month flywheel, and the reason is the buying cycle, because nobody wakes up and impulse-buys a thirty-year loan, they marinate, they worry about whether they can afford it, they google the same questions over and over, and they slowly decide who they trust, and that long quiet consideration window is exactly the window that owned content owns and paid leads cannot touch, right. A marketplace lead is a snapshot of one moment, the moment someone filled a form, and you pay full price for that single moment whether the person is ready or a year out, whereas a content library is present for the whole window, it shows up in the search at month one of their consideration and it is still showing up in their feed at month eleven when they finally pick up the phone, and by then they do not feel like they are calling a stranger, they feel like they are calling the person from the videos, right.

The founder said the thing that stuck with me in our month-three review, he said the calls feel completely different now, people get on already calling me by my first name and asking about my dog from the Reel, and they are not interviewing me anymore they are just ready to move, and that is the entire game, because trust is the expensive part of a mortgage sale and we manufactured it at scale once instead of rebuilding it by hand on every cold call, right.

The other reason it worked is that I refused to let it become a content project, it was always an economics project that happened to use content, so every asset had a job and every job pointed at a dollar, and we never once celebrated a view that did not eventually trace to a captured lead or a booked call or a funded loan, and I think that discipline is what separated this from the kind of agency engagement where you get a pretty feed and a flat bank account, right.

And the one-shoot cadence is what made it sustainable, because the founder is a mortgage broker not a content creator, and if I had asked him to film weekly he would have quit by week three, so we compressed his entire monthly output into a single focused day, I did the slicing and the distributing and the measuring, and he got to stay in his zone of genius which is talking to borrowers, and that division of labor is the only version of this that survives contact with a busy founder's calendar, right.

If you run a brokerage and you are funding loans at $7,000 of acquisition cost off leads that go cold in your hands, the move is not to buy more leads, the move is to stop renting attention and start owning it, one shoot a month, distributed everywhere it compounds, measured in funded volume and nothing softer, and that is the entire thing, right.

The last thing I will say is about durability, because a lot of founders ask me what happens when they stop, and my honest answer is that the engine has a tail, the 187 assets keep ranking and keep circulating and keep capturing leads for months after the cameras stop rolling, so even in a hypothetical month where Northbound shot nothing, the library would still be feeding the funnel, and that is the whole reason I push owned distribution over rented clicks, you are building an appreciating asset on your own balance sheet instead of pouring fuel into someone else's marketplace, and for a business where one funded loan can be worth thousands in commission, an owned engine that produces forty-two of them a month at fifteen hundred dollars of cost each is not a marketing line item, it is the single most profitable thing on the books, right.