All Four Sides

I say a version of this line in the first two minutes of most podcasts: I've stood on all four sides of the consumer-direct lead.

It's true. It's also, by itself, a résumé — and nobody cites a résumé. The reason it's worth writing down is something I only understood after I'd sat in the fourth seat: each of those positions is confidently wrong about something, and the error is invisible from inside the seat. You need one of the other three to see it.

Here they are, in the order I occupied them.

2005: I thought the problem was routing and speed

I founded Kaleidico as a software company. Not an agency — that came after 2008. The product was icoSales, lead management for mortgage call centers that were buying thousands of internet leads a month and drowning in them.

I assumed the problem was routing and speed. And I was right. That's the trap. Being right about the visible constraint is exactly what lets you miss the one underneath it.

The thing underneath was that choice itself was the problem. Give a loan officer a queue and a free hand, and sales instinct takes over — instinct that is paralyzing and, more often than anyone wanted to admit, wrong. So we built prioritization on conversion patterns instead. Intelligent, for 2005.

What came back from the data reorganized how I've thought about leads ever since. The desirability of a lead was not correlated to its convertibility. High loan amounts, high credit scores — big commissions, smooth processing, the files every LO wants — did not predict revenue. The conversions were sitting in the messy middle.

Almost nobody else was seeing this, because almost nobody else was prioritizing at all. What passed for lead management in that era was mostly fancy spreadsheets and lists. And where firms did prioritize, they prioritized on what the loan officers asked for, which is to say on desirability. They were optimizing for the preference of the person working the lead rather than the behavior of the person on the other end of it. We were the first to see the pattern and build for it.

Twenty years on, I'll make the harder version of that claim: most of this industry still isn't doing it. Salesforce, GoHighLevel, Velocify, Shape — the platforms that own this category are, functionally, expensive spreadsheets. Getting conversion-pattern prioritization to work inside any of them takes a serious pile of professional services, which is why the teams that genuinely have it generally bought a third-party layer to get it. I am still building this pattern by hand today, for my own properties and for the companies I work with.

So: right about routing, right about speed, and building for the wrong party.

2008–now: the agency's blind spot is the number it's judged on

Kaleidico became a demand generation agency after the mortgage meltdown took our call-center client base with it. I've been in that seat ever since, and it has a structural problem that has nothing to do with anyone's competence.

Without a real partnership — shared data, shared visibility into what actually funded — an agency has no way to model convertibility. It can only see what it can measure, and what it can measure is Cost Per Lead. So it competes on that. Everyone competes on that. It becomes a race to the bottom.

Here's why that backfires, and it's the part that took me a lender's P&L to feel properly. Lead generation is a marketplace, much like the stock market. In an efficient market, the better asset costs more. That's not a vendor extracting margin, that's the market working. When you push relentlessly for the lowest possible cost per lead, you are not winning a negotiation — you're making a quality trade, and someone downstream becomes the filter for what you let through.

That someone is the sales operation.

I can prove the trade with my own numbers. At Velocity Lending I got aged leads and real-time leads to convert at the same CAC. Same cost per funded loan, both channels. But to hit the same LO production goals, the aged channel meant sales-processing thousands of records where real-time meant hundreds. Identical acquisition cost, an order of magnitude difference in operational load.

An agency looking only at CPL cannot see that difference. The client feels every bit of it.

2016–2018: I knew leads cold, and still got the hard part wrong

I bought Velocity Lending and ran it as a DTC mortgage lender alongside Kaleidico for two years, buying leads against my own P&L.

I want to be honest about what I did not get wrong, because the flattering version of this story would have me humbled on price and I wasn't. I had a good read on the economics going in — I'd spent my career in consumer-direct lending. We managed LO compensation, lead buying, and first-party acquisition to a consistent CAC of about $1,500 per funded loan, and we held it.

What I underestimated was everything the lead lands in.

I had spent my whole career on the generation side. I knew the digital lead channel intimately, including its weaknesses — it skews to the lower end of the bell curve on quality and qualification, lower intent, more shoppers and tire-kickers. None of that surprised me. What surprised me was how hard it is to build a sales operation that is simultaneously fast, automated, responsive, and still feels like a human being talking to another human being.

I had that budgeted at about thirty days. It took closer to six months to really dial in.

Along the way my own metrics reordered themselves. Speed to lead, which had meant speed to first dial, became speed to first contact — text and email. Contact rate turned out to matter far more than I'd credited, because it's the honest read on whether your sales operation actually works. App-to-loan turned out to matter far more too, because it's the read on product and operational fit: do you have the right products and lending partners, and are you processing efficiently? And cost per lead quietly demoted itself. CAC is the appropriate measure, because CAC is the one adjacent to profit.

The mechanism that made the whole thing work wasn't dial count. It was contact speed, and then whether the LO had the training and the products to take a clean app and a clean file on the first call. When that happened our pull-through cleared 50%, because the path from application to closing was smooth — nothing floating away while a customer got frustrated.

And then the part that cost me something to learn. Sitting on the buy side, I discovered what lead buyers get wrong about agencies. The answer is patience and genuine collaboration — real data exchange, treating your marketing partner as an outsourced marketing department with all the grace and give of that relationship, rather than a vendor to be used and abused. I had run an agency for eleven years. I had to go sit on the other side of the table to see it.

Velocity worked. It was also brutally hard and a real distraction from what I'm actually best at. I'd do it again, because there is no other way to buy that experience.

And it clarified something I'd never had to say out loud. I have always been better at leading a marketing operation than a sales operation. They are very different animals. I came out of those two years with a genuine, collaborative respect for both — and no illusions about which one is my native ground.

2025: my own book places last

The Lead Buyer's Playbook is the fourth seat, and I'll rank it honestly: last.

It will teach you to buy leads well. It cannot hand you a sales operation, and it cannot hand you the temperament to be a good partner to the people generating your demand. Those are the two things that actually decided whether Velocity made money, and the book covers neither.

That gap is why I'm writing the next one. It's called Sales Team of One, and I'm writing it in the open at SalesTeamofOne.com — drafts, arguments, and the parts I'm still working out.

The pattern

Look at the four seats together and they're all describing the same blind spot from different angles.

The software seat found that the loan officer's instinct about which lead to work is wrong. The agency seat externalizes filtering cost onto a sales floor it can't see. The lender seat budgeted thirty days for the operation and needed six months. The author seat can't hand you one.

Every one of them is the same finding: the lead is not the constraint. The operation that receives it is. And not one of those four seats can see that from the inside. That's not a failure of intelligence. It's a property of the seat.

What to do Monday

If you take one thing from this, take this one.

Make speed to lead a text, then email, then phone sequence, in that order. Take initial contact completely out of the hands of your loan officers.

Systems work leads. People work people.

Then change what you report on. Your number one metric is scheduled appointments and application calls — not dials, not cost per lead. Everything I learned in twenty years of building the front end, and two years of living with the back end, comes down to protecting the moment a human being finally talks to a human being.

I write The Lead Brief on how this plays out week to week — what's actually working in consumer-direct acquisition, and what only looks like it is. Subscribe here.