They didn't need more leads. $50M of them were already in the building.
Efficient Capital Labs, a pre-Series-A fintech lender, had nine people in go-to-market and no revenue operations at all. Marketing, sales and underwriting each ran their own systems, and none of them talked, so when a borrower started a loan application, sales never saw it. I connected the application system to the CRM, scored and routed every application as a live lead, and built a signal-based outbound engine on top of it. Inbound conversion went from 12% to 42%. Roughly $50M of loan pipeline that sales had never been able to see became active. The outbound engine drove $15M of net-new loan disbursal in a single quarter.
The problem
Borrowers were starting loan applications (real companies, real intent, real money) filling in a form that fed straight into the underwriting system. Underwriting owned that system. Sales owned the CRM. Nobody had ever connected the two.
So an application would start, and one of two things happened. It completed, and underwriting processed it. Or it stalled, and nothing happened at all: no follow-up, no nurture, no rep call. Only 12% of started applications ever completed, which is another way of saying 88% of the people who raised their hand walked away, and nobody noticed.
This wasn't a broken process. It was $50M of demand, already raised, sitting in a system sales wasn't allowed to look at.
Applications went dark. Only 12% of started loan applications ever completed. The other 88% raised their hand, then walked away (no follow-up, no nurture, no rep call) because the application system fed underwriting, and sales was never wired to it.
No function, ever. Nine people in go-to-market (a CRO, two in marketing, six in sales across three continents) ran the pipeline out of spreadsheets. No routing, no scoring, no attribution, no shared view of a deal.
Outbound was a hobby. Reps built their own lists and wrote their own emails when they had time between deals. That is not an outbound motion.
Spend nobody could measure. Events and partnerships both drove disbursal nobody could attribute, so the company could not decide whether to spend more on either.
The diagnosis
I was hired to build an outbound engine. That was the second thing they needed, not the first.
Every pre-Series-A company believes its problem is the top of the funnel. It's the diagnosis that feels most like growth. ECL genuinely did need an outbound motion; it had nothing. But it also had $50M of people actively filling in applications, 88% of whom were being ignored. Before you go buy demand, look at the demand already raising its hand.
The silo wasn't marketing-to-sales. It was sales-to-underwriting
Every RevOps engagement starts by looking for the marketing-to-sales handoff. At a lender, that's the wrong seam. The application system (the thing underwriting owns, the thing that is the product) was where the buying intent lived, and it had no wire to the CRM. A borrower starting an application isn't an operational event; it's the highest-intent signal in the business.
You cannot run an allbound engine on a spreadsheet
Nothing downstream (scoring, routing, nurture, attribution) works without a data model underneath it. We built the CRM, the data model and the customer journey first, and only then the engine on top. Automating a broken process just breaks it faster.
Response rate is a signal problem, not a volume problem
The instinct when outbound is failing is to send more. We went the other way: define the ICP tightly, segment by intent, research every account with agents before writing to it, and send fewer, better messages. That's why the cold response rate came out near 16% rather than the 1-3% a volume motion produces.
What I built
The wire between the application system and the CRM
Every started loan application now creates a scored, auto-routed lead in HubSpot, with parallel nurture and a rep check-in task attached. Intent and fitment scoring is derived from the application's own parameters: the borrower tells you how serious they are by what they fill in. This is the build that surfaced the ~$50M.
A signal-based outbound engine, from nothing
ICP and intent signals defined; an 18,000-account, 50,000-contact ABM database assembled with Clay, Apollo and custom AI scraping; accounts segmented high, medium and low intent; per-rep account books; omnichannel sequences across email, social and phone through Instantly and HubSpot, with AI agents researching each account and writing the message for it. It scaled from ad hoc to 1,500 accounts a week.
The deliverability plumbing underneath it
20 sending domains with four active at a time, 20 rotating inboxes, health maintained automatically. Outbound at volume is a deliverability problem before it is a copy problem, and most engines die here.
A pipeline the company could actually run
Off spreadsheets and into a 6-stage HubSpot pipeline with the loan-application sub-steps embedded, stage automation and exit checklists. Daily pipeline reviews and weekly quota reviews replaced a manual multi-sheet pull. A closed-lost re-engagement programme reactivated roughly $2M of pipeline that had been written off.
Events and partnerships turned into governed channels
A quarter-ahead events calendar with pre-event targeting and nurture, in-event meeting booking, airtight capture into HubSpot, and post-event distribution. The same discipline applied to partners. Both had been driving disbursal nobody could attribute; both became measurable.
Leadership visibility, and a team to keep it running
Looker dashboards for the CRO, CEO and COO: pipeline, channel performance, origination and disbursement trends, re-application ratio. And ECL's first two RevOps hires, recruited and ramped, who still run the function on the architecture and SOPs left behind.
The outcome
Four-month engagement, plus advisory through the following quarter. All metrics measured in that window.
We thought we had a pipeline problem. What we actually had was a pipeline we couldn't see: companies applying for loans our sales team never knew about. Four months, from spreadsheets to a revenue engine.
What this engagement didn't solve
Contact-level de-anonymisation did not work. We identified anonymous accounts visiting the site through Apollo and RB2B, but resolving those to individual contacts (the thing everyone wants from this category of tool) did not land. The tooling isn't good enough at this price point yet, and I wouldn't sell it as a deliverable today.
Clari went in but never proved itself. Revenue intelligence was implemented, but four months is not enough time for an AI forecasting tool to have anything to forecast from. It's a capability ECL now has, not a result I can claim.
And I left after four months. The systems were handed to two people I hired and ramped, and they still run them, which is the outcome I wanted. But a build engagement ends. What it can't give you is someone in the room the next time the market moves and the whole motion needs rethinking.
Why this matters for you
This is your problem if:
You have just raised, or are about to, and everyone in the building agrees the problem is 'not enough pipeline'
You run an application, trial, quote, or onboarding system that your sales team cannot see into
Your outbound is a volume exercise producing a 1-3% reply rate, and the answer keeps being 'send more'
Marketing, sales, and whichever team owns your product's front door each run their own system, none wired together
You are spending money on events or partnerships and cannot tell anyone what either returns
Before you buy demand, look at the demand already raising its hand
The highest-intent signal in most businesses isn't a form fill. It's someone halfway through your onboarding, trial, quote, or application, who then stops. Find out how many of those you have. It's almost always more than you think, and almost nobody is calling them.
The silo that matters isn't always marketing-to-sales
Everybody looks there first because that's where the playbooks point. Ask instead: which team owns the system where a customer first shows real intent, and does sales have a wire into it?
Reply rate is a signal problem, not a volume problem
If your outbound is at 2%, sending twice as much gets you 2% of twice as much and burns your domains doing it. Research the account before you write to it, and send a tenth as many.