Know Which Broker to Call Before You Send the First Email
A vendor tier rating said who was in good standing. It could not say who places this lender's kind of loan, so we scored for the second thing from public data.
- 14x
- More concentrated outreach
- 4x the replies
- Against a conventional campaign
- 2 deals
- In pipeline from 88 sends
Consortium Equity is a private lender. Their growth depends on mortgage brokers sending them deals that fit a narrow underwriting box, which means the whole job is knowing which brokers place that kind of loan and getting in front of them first.
They had a list. What they did not have was a way to tell which names on it were worth an hour of anybody’s week.
The list said one thing and the market said another
The seed data was a Brokers Advantage export: 372 active California brokers, each carrying a tier rating from Tier 1 down to Tier 4. The obvious move is to work the list top down.
That move is wrong, and the reason is the central finding of the engagement:
A vendor’s tier rating describes the broker’s standing with that vendor’s program. It says nothing about whether the broker places loans that fit Consortium’s box.
Working a tier-sorted list means spending your best hours on brokers who are excellent at something you do not fund.
What we built
A Synapse app running inside their own NimbleBrain workspace. It takes each broker on the list and does four things without anyone touching it.
Pulls the regulatory record. Full NMLS Consumer Access profile: licence status, every state they are licensed in, regulatory disclosures, and how long they have been in the business.
Checks the complaint history. The CFPB consumer complaint database, queried per broker.
Reads what they sell. The broker’s own website, classified by Claude into what they place: DSCR, hard money, fix-and-flip, owner-occupied, FHA. This is the step the tier rating cannot do, because a tier is assigned by a program and a website is written by the broker.
Scores and drafts. All of it weighed against Consortium’s box, then a personalised cold-outreach email written for each broker in their voice, leading with the edge that matters to that broker.
What the scoring found
Ranking by fit rather than by tier moves the list around hard.
One Tier 1 broker had 61 active state licences and eleven years of tenure, which reads as a top-of-list name on any spreadsheet. Their website showed an owner-occupied, FHA and VA conventional focus, and their CFPB record carried dozens of consumer complaints. They ranked nineteenth of twenty-five.
The broker that ranked first was also Tier 1, but with a quarter of the licence footprint. What separated them was a website showing non-QM and DSCR specialty, a clean regulatory record, and an office in the same building as Consortium’s.
Neither of those conclusions is available from the tier rating. Both are available from public data nobody had assembled.
Everything is checkable
There is no proprietary data set in this and no model output anyone has to take on faith. NMLS Consumer Access is public. The CFPB complaint database is public. Every broker’s website is public.
Any score in the app can be traced back to a link the client can open, on the official source, and confirm for themselves. That verifiability is the product. A score somebody has to trust is a score somebody will overrule the first time it disagrees with them.
What the outreach did
The scored list feeds Precision Outbound, which drafts in the firm’s own voice off their closed deals rather than from a template. Across a five-day window it sent 88 emails, drew four positive replies, and put two deals into pipeline. A conventional cold campaign run against the same market needed 5,500 sends over those days to earn a single reply.
The volume figure matters as much as the reply count for a firm this size. Eighty-eight sends does not exhaust a broker market that everyone in private lending is working, and it does not put the principal’s name in front of people who were never going to place their kind of loan.
What we fixed on the way
The scoring surfaced a second problem with nothing to do with ranking. Outbound from the primary domain was bouncing at 37 to 43 percent, driven by stale addresses in the source export rather than by spam filtering. Mailbox providers do not ask why a sender bounces; sustained rates above about five percent read as a spam signal, and forty percent is the kind of number that gets a domain quietly graylisted at Google and Microsoft within two to three weeks.
Fixing it meant authentication records on the domain, address verification before send, and moving cold outreach off the principal’s own sender identity. Good targeting into a burnt domain is still nobody reading it.
Why NimbleBrain
The finding that reordered the list came from asking what the tier rating measures, rather than accepting that a number sorted by a vendor is a number worth sorting by. That question costs nothing and nobody had asked it.
Everything downstream follows from it. The app runs inside their own workspace, the score traces to a record they can open, and the underwriting box the scoring weighs against is theirs rather than ours. A ranking somebody has to trust is a ranking they will overrule the moment it disagrees with them, so the whole thing is built to be argued with.