A CRM and referral-partner system for a financing advisory

This is a project for a small advisory firm that helps small and mid-sized businesses find financing. It does not lend. It matches each business to outside lenders and is paid when a deal closes. A large share of its new business comes from bankers who cannot approve their own clients and pass them on. I am not naming the client, and we built all of this with their own team and developer.

The problem

New business depended on bankers, and reaching bankers meant one LinkedIn conversation at a time. That works for a founder with a few dozen contacts. It does not grow.

Behind that, the work lived in inboxes, spreadsheets and a CRM that held the only copy of years of history. Most email was sent from a personal inbox and typed back into the CRM by hand, and the lender database was too thin to match a deal to the right lender.

What we built

A website built for search and for AI answers. Pages that load fast and answer the questions buyers ask in plain words, with structured data, a file for AI crawlers, and a check that stops a release when titles or descriptions are the wrong length. A small chatbot answers from the company's own information.

Lead intake that scores itself. A website form and a qualifying chat. Each lead becomes a deal in the CRM and is scored hot, warm or cold, so the team knows who to call first.

A custom CRM. Deals, contacts and referral partners in one place, with email filed under the right deal, reminders and reporting. We built it with the company's developer. At first it was mostly a record-keeping tool, and the team kept sending mail from their own inboxes.

A call tool. It listens to a qualifying call, picks out the answers to the vetting questions, and shows which lenders fit while the call is still going. We built it, and we do not have usage numbers.

A lender portal. Lenders log in with a one-time link and update their own criteria. An administrator approves each change, so the matching data stays clean.

Outbound to referral partners. LinkedIn first, email second, aimed at commercial bankers. A weekly job finds the cheapest people to reach, a data tool checks who is really a banker and finds a work email, and a message sequence asks for an introduction call. People take over every reply.

What happened

• About 5 banker meetings a week while the campaign was switched on, April to June 2026.
• Of the cold contacts, 12.7% replied and 4.1% agreed to a meeting (42 of 1,017), at 15 September 2026.
• Two referrals traced to the campaign over six months, matched by name.

The campaign was switched on and off, so the number went up and down. It was down to one meeting in the first two weeks of September, and it was paused at the end of September.

A June 2026 audit scored the website 8 of 10 for how well it is built for AI answers, and 3 of 10 for how often it is actually cited. The site is built right and still needs to be talked about elsewhere. We have no measured traffic or lead result from the website work yet.

What this shows

The first version of a referral-partner system found out how many bankers we could reach and how few turned into introductions. That is what a first version is for. The next one starts from the people who already know the founder. See outbound systems.