The cold start

Here's the first uncomfortable truth: no debt provider will give you a real term sheet before you have loan performance data. There's no way to negotiate around it, because there's nothing to underwrite. Fixed income buys track record, and you don't have one.

So the earliest cohorts get funded with equity — savings, angels, seed money. It will feel expensive, because it is. It's also the price of admission. A typical journey looks something like: tens of thousands of dollars of your own loans → a few hundred thousand from seed equity → your first $1M originated → a first dedicated facility, unlocked in combination with a proper seed round. Getting to that first $1M of originations routinely takes a year. The equity raise and the debt raise go hand in hand — the equity funds both growth and your first-loss position, and debt investors will judge the quality of your equity backers, sometimes calling them directly.

The good news: this compounds. Once you have portfolio performance, options multiply, cost of capital drops, and facility size grows — as long as performance holds.

The data that unlocks a facility

When you finally sit down with debt investors, these are the questions you need real answers to:

• Static pool analysis. For close-ended products: cumulative default (or delinquency) curves by cohort. Healthy curves flatten as they mature — flattening means borrowers who've paid most of the way through keep paying. For open-ended products: delinquency and default ratios as a % of outstanding balance by cohort, plus how you manage line sizes.
• Principal paydown speed. How long does it take to pay down 50% of principal? Risk falls as principal comes back; your paydown should look like comparable products in the market.
• Volume in the credit box. How many loans have you originated in the specific credit box the facility will fund? Lenders know risk scales as you grow — they're checking whether tomorrow's borrower looks like the ones in your data.
• Acquisition channels. Different channels carry radically different credit risk, independent of CAC — paid-search borrowers are not direct-mail borrowers, marketplace borrowers are not walk-ins. You won't have tested everything; you do need a clear story for which channels you'll scale and what that does to risk.
• Cohort stability. No exact threshold, but you know when you have too few loans: it shows up as performance volatility. One noisy cohort is fine. Several is a message — that you don't have a grip on credit risk.
• Servicing and collections. Expect your servicing-and-collections policy to be vetted line by line: payment options, delinquency outreach, collections setup, charge-off process. Weak servicing is a red flag; sometimes investors will like the credit but force a third-party servicer, and if you self-service they'll require a backup servicer regardless.
• Compliance and enforceability. The loan contract is the only backstop the investor has. If enforceability can be challenged — bad e-signature flow, missing disclosures, licensing gaps — or your product sits in regulatory gray space, credit performance is irrelevant; the deal dies. Get legal and compliance tight before the raise, and be ready to answer licensing questions state by state (or market by market) without checking your notes.
• Equity plan and origination plan. A tentative equity roadmap (growth + first loss) and a realistic origination growth plan. Realistic matters: your projections drive facility terms, draw schedules, and covenant levels, which is a gun you can point at your own foot — see Part 7.

The data room, version one

Don't wait to be asked. The first-request list from a serious debt investor is almost always the same seven items, so have them ready the day you start conversations:

1. Pitch deck (the narrative one — see below)
2. Loan product deck (the credit one — see below)
3. Loan tape — loan-level detail for everything you've ever originated: origination date, amount, term, pricing, borrower attributes, status, payment history, charge-off data. Clean it before they see it. Every anomaly you leave in is a diligence question you'll answer later, at a worse moment.
4. Historical financials — monthly, not annual.
5. Cap table — they're underwriting your equity cushion and your backers.
6. Operating model and projections — including origination volume by product and by month.
7. Customer/channel pipeline — where the next cohorts come from.

That's the starter kit. The full institutional list — policy suites, credit-model documentation, flow-of-funds diagrams — comes later (Parts 9 and 10), but these seven get you to a first real conversation. The complete checklist is in Appendix B.

The pitch is two decks

Here's a structural insight that took me too long to learn: a debt raise runs on two decks, not one.

Deck one: the narrative deck. Looks superficially like an equity deck — team, market, product, growth — but every slide is re-aimed:

• The team slide is a lending track record slide. Not "ex-Google, ex-McKinsey" — how much credit have you originated, underwritten, or collected before, and at what loss rates? If your team has scaled a book somewhere else, that's the single most valuable slide in the deck. Lead with it, quantified.
• Equity backing is presented as loss protection. "Backed by [good funds], $XM raised" isn't vanity here — it's the lender's cushion and their assurance you can fund your first-loss. Say it plainly.
• Growth is framed as a throttle you control. The equity version of your deck says "explosive growth." The debt version says "deliberately staged deployment: pilot, controlled ramp, scale — and here are the credit gates between each phase." I know founders who run one deck bragging about how fast they can grow and another bragging about how well they can refuse to grow. Both are true. Know your audience.
• The pipeline answers "can you deploy my capital?" A debt investor with $25M to put to work needs to believe you can originate enough qualifying volume to use it — on schedule. Show the demand math: channels, conversion, realistic volume ramp.

Deck two: the loan product deck. This one has no equity-deck equivalent, and skipping it is the most common first-timer mistake. It discloses your entire credit box, explicitly:

• Underwriting criteria, with numbers: minimum credit score, minimum time in business or income thresholds, debt-service coverage, the data sources you pull (bank data, bureau data, financials, platform data — whatever your stack is).
• Line-assignment and pricing logic: how you size a loan, illustrated — e.g., "credit limit = 50% of trailing-average monthly cash balance" (invent your own; the point is you show the formula).
• Collateral and support per product: guarantees, liens/UCC filings, security.
• Terms and fee ranges per product: durations, APR ranges, origination fees.
• Servicing and collections: payment rails, autopay/ACH setup, delinquency cadence, charge-off policy.
• Portfolio monitoring: how you detect deterioration early — balance monitoring, re-underwriting, debt-stacking detection, whatever you actually do.

To an equity founder this feels insane — you're handing over the secret sauce. But the debt investor is buying the output of this machine. The credit policy is the product. Vague answers here don't protect your IP; they kill the deal. And your early performance data, however small, goes in this deck with pride: a clean pilot — even a tiny one — presented alongside a conservative credit box is exactly the seed-stage debt story. Small and spotless beats big and noisy.

One more move for both decks: quantify your data advantage as underwriting lift. "We have proprietary data" is an equity-deck abstraction. The debt version is a measured delta: "with partner data in the model, approval rates rise materially at the same expected loss" or "our early-delinquency detection cuts roll rates by X points" (your numbers, honestly measured). One quantified lift is worth ten adjectives.

What diligence actually feels like

Assume the diligence team will find everything. Not most things. Everything.

They will find the one loan in your tape that deviates from your stated credit policy — the pilot-era exception you forgot about — and they will ask about it. The right answer is the honest one: "early experiment, policy now requires X, here's the date the policy changed." The wrong answer is surprise. Better yet: flag your own exceptions before they do. A one-page "known exceptions and what we changed" memo in the data room converts a gotcha into evidence of control.

They will ask about every gap between projection and actual. If you told them conversion would be 50% and it's running at 30%, have the decomposition ready: which channel, what changed, what you're doing. Lenders can live with misses. They cannot live with unexplained misses.

They will ask exactly who the credit counterparty is, and they'll want the payment-flow diagram. Who is legally obligated to pay? What happens when the end-payor doesn't pay — who eats it, with what recourse? Draw the boxes and arrows before the meeting. If there's any structural ambiguity in your product about who owes whom, resolve it in the product before a lender finds it in diligence.

They will walk your ops stack: origination system and servicing system, built vs bought; collections process step by step; what happens on a failed payment; how you handle disputes. And they'll probe expected losses at different scale points — because they know, even if you don't yet, that losses drift up as you grow past your earliest, most hand-picked customers. Give phase-based answers: "X% at pilot scale, we underwrite to Y% at scale" reads as sophistication, not weakness.

The meta-skill: answer the risk question before it's asked. The best debt materials I've seen include a slide that literally states the lender's scariest question — "what happens to adverse selection if a second lender shows up on this channel?", "what's your exposure if your biggest partner churns?" — and answers it structurally. You know what's in their memo template. Fill it in for them.

There will always be a reason your data isn't enough

Know this going in: there will always be a reason. Not a big enough sample. Big enough, but hasn't seasoned through a full payment cycle. Seasoned, but hasn't seen a macro cycle. Seen a cycle, but not in this exact channel mix.

That's not rejection — that's the genre. Fixed income is a no-by-default business, and "come back in six months" is often literal, sincere advice. Keep originating, keep the cohorts stable, keep the same lenders warm with quarterly updates (Part 10's update skeleton works pre-facility too — sending a clean quarterly update to lenders who haven't funded you yet is the cheapest credibility-building move there is). The second conversation with a lender who watched you hit the plan you showed them six months ago is a completely different conversation.