The Data on Which Lending Models Keep Realtors Loyal

Which Mortgage Company Models Build the Stickiest Realtor Relationships?

New data from Mortgage Market Intelligence, analyzing loan officers across 1,046 mortgage companies, found brokers average an Agent Loyalty score of 5.81 on a 0 to 10 scale, independent mortgage banks average 5.43, and banks and credit unions average 4.55. Brokers were more than six times as likely as banks to produce a loyalty score of 6 or higher, 46.8% versus 7.2%. But the number that should actually change how you evaluate a lending partner is this: among independent mortgage banks alone, company scores ranged from 8.43 down to 2.24. The business model sets the range. What a company actually builds around its loan officers decides where it lands inside it.

A data set landed in my inbox this week that put a real number on something I've been telling team leaders for years without one: not every lending relationship is built the same, and the difference shows up directly in whether your realtors actually stick with the loan officer you handed them to.

The data comes from Mortgage Market Intelligence, a research newsletter that built something called the Agent Loyalty Index, a score that tracks how consistently a real estate agent works with the same loan officer over time. This week they ran that index at the company level. They pulled every loan officer who originated at least 10 loans over the past year, found the one real estate agent each of those loan officers transacted with most, scored the loyalty of that top relationship, then averaged the scores across the whole company. In plain terms, it's a Realtor stickiness score for 1,046 mortgage companies.

The results are worth sitting with.

What the Numbers Actually Say

Broken out by business model, here's how the averages landed:

• Brokers (222 qualifying companies): 5.81 average, 5.94 median. 82.4% reached a loyalty score of 5 or higher, and 46.8% reached 6 or higher.
• Independent mortgage banks and other nonbank lenders (351 qualifying companies): 5.43 average, 5.70 median. 69.8% reached 5 or higher, and 34.8% reached 6 or higher.
• Banks and credit unions (473 qualifying companies): 4.55 average, 4.56 median. Only 32.1% reached 5 or higher, and just 7.2% reached 6 or higher.

That's a gap of roughly 1.26 points between brokers and banks on a 0 to 10 scale. It shows up even more starkly in the threshold numbers. A broker is more than twice as likely as a bank or credit union to build even a moderately loyal Realtor relationship, and more than six times as likely to build a strongly loyal one.

Look at the overall Top 10 companies by Agent Loyalty score and the pattern holds. Five were brokers, four were independent mortgage banks or other nonbank lenders, and one was a bank or credit union. The report also flags a real caveat worth keeping in mind: the companies at the very top tended to have smaller qualifying loan officer populations, so scale is part of the interpretation, not just model type.

Why Model Type Isn't the Real Story

Here's where most people reading this data stop, and where I think the actual lesson starts.

The easy takeaway is "work with a broker, not a bank." That's directionally true, but it isn't the useful part of this data. The useful part is what happens inside each category.

Among qualifying independent mortgage banks, the highest company-level Agent Loyalty score was 8.43. The lowest was 2.24. That's a spread of more than six points, wider than the entire gap between the broker average and the bank average, and it's happening among companies operating under the exact same business model.

If model type explained the whole story, every company inside a category would land close to that category's average. They don't. Some IMBs are building Realtor relationships that are nearly perfect on this scale. Others, structured the exact same way on paper, are barely holding on to any of them.

That tells you company type sets the range you're likely to see, but it doesn't decide where a specific company falls inside that range. Something else is doing that.

What Actually Builds a Sticky Realtor Relationship

The report raises the plausible drivers without proving which ones cause what, and it's worth being honest that this data shows correlation, not causation. But the candidates it points to are the same things I've been building my career around: loan officer autonomy, how directly a company's survival depends on Realtor referrals, marketing support behind the LO, recruiting quality, comp structure, and company culture.

Think about what that actually means structurally. A broker or IMB usually lives or dies on Realtor referral relationships. There's no deposit business, no walk-in branch traffic, no cross-sell from an existing banking relationship to fall back on. The loan officer's entire business is the agent relationships they build and keep. A bank or credit union mortgage desk often has other lead sources built into the institution, which means an individual agent relationship is one channel among several, not the whole game.

That's not a knock on banks. It's a structural difference in what the loan officer is actually incentivized to protect.

I wrote about this exact dynamic when I laid out why the reactive loan officer model is finished, and again when I broke down what an embedded loan officer actually costs a team versus a preferred-lender split. This data is the first real number I've seen that backs up what both of those posts argued from mechanics alone: an LO who owns a relationship, with real autonomy and real marketing support behind them, builds something stickier than an LO who is one interchangeable desk inside a larger institution.

But the spread inside the IMB category is the part team leaders should not skip past. Picking "a broker" over "a bank" is a starting filter, not a finished decision. The specific company you partner with, and what it actually equips its loan officers to do, is what puts you closer to an 8.43 or a 2.24. The category on the license doesn't decide that. Execution does.

What This Means When You're Building or Picking a Lending Partner

If you're evaluating a preferred lender, a sponsor relationship, or building out an embedded LO seat inside your own team, this data gives you a real checklist instead of a guess:

1. Does the loan officer have actual autonomy to run their own book? Or are they routed leads and rules from corporate that treat every referring agent the same way?
2. Is there real marketing support behind that LO, keeping them visible to the same agents consistently, not just showing up at closing?
3. Does the company's business depend on Realtor referrals surviving? Or does it have other lead sources that make any single agent relationship optional to protect?
4. Is the loan officer's comp and career path built around owning a book of referring agents, or around originating whatever volume gets routed to them?

An embedded LO model, built correctly, checks all four of those boxes on purpose. That's not a coincidence. It's built with autonomy, dedicated marketing infrastructure, and a comp structure tied directly to the relationships it produces, sitting inside your own operation instead of a stranger's book at a company you don't control. That's also exactly why a loose, undisclosed sponsorship arrangement is the wrong structure on both the compliance side, which I wrote about after the Rocket steering lawsuit, and now on the plain performance side, based on this data.

Every company's situation is different, and this data can't tell you what a specific lender will do for your team, only what the base rates look like across business models. But if you're deciding who gets your buyers' mortgage business, or whether to build that seat yourself, this is exactly the kind of question worth running before you commit to either. If you're evaluating what actually builds a durable lending relationship for your team's buyers, that's exactly the conversation we have on a partnership call.