From term sheet to documents

All numbers illustrative.

The second big mistake, and the one that bites hardest, because the failure mode is invisible at signing.

Here's the asymmetry that makes it dangerous. At term sheet stage you have leverage — competing sheets, or at minimum the credible option to walk. The day you sign, that inverts: you're in exclusivity, you've paid deposits, you're accruing two sets of legal fees, your equity investors expect the close, and your origination plan is queued against this capital. Every week that passes makes walking away more expensive for you and no more expensive for them. Every open item now gets resolved by the party with nothing to lose from delay.

Nobody is behaving badly here — it's simply what negotiating without leverage looks like, for anyone, on any side of any table.

And closing takes months. Everyone gets tired. There will be a strong gravitational pull — from the lender, from counsel, sometimes from your own team — to sign the sheet and defer the "definitional" items to documentation. Eligibility criteria, delinquency triggers, default triggers: "we'll work that out in docs, it's operational."

It is not operational. Refuse.

Those definitions are the facility. What follows is the catalog of what actually happens in the documentation phase — the recurring patterns, in escalating order. Every one of them is standard practice, not scandal. Which is exactly why you handle them structurally — with numbers agreed while both sides are still at the table — rather than leaving them to goodwill under deadline.

The patterns

1. "To be defined" is a blank check. The single line in the sheet reading "eligible receivables (to be defined in definitive documentation)" routinely becomes several pages of the loan agreement: dozens of eligibility criteria (borrower attributes, loan caps, documentation standards, origination requirements) plus a full schedule of concentration limits. Each criterion narrows which loans the facility funds; the intersection of forty criteria can quietly exclude a third of your book. And buried in the list will be terms that constrain your product, not just your collateral — a minimum weighted-average portfolio yield is a floor under your pricing; a cap on modified loans is a constraint on your hardship policy; a "no loan to any borrower who has ever been delinquent" criterion quietly bans repeat lending to your own recovered customers, which might be your best segment. None of this was in the sheet. All of it was in "to be defined."

2. The covenant headline is not the covenant. The sheet says "maximum cumulative loss: 8%." Reasonable — your mature vintages run 5%. The loan agreement delivers that 8% as a month-on-book curve: 1.5% by month three, 3% by month six, 5% by month nine, 8% only at maturity. The headline number is the end of the curve; the binding constraint is the front, where one noisy early cohort can breach a trigger while your book overall is performing beautifully. Same with "delinquency triggers": the sheet's single number becomes monthly tests of three different ratios with three different consequences. Rule: never accept a covenant headline without the full schedule of test levels, test dates, and measurement definitions. How is delinquency measured — by count or balance? At what day bucket? Does a modified loan cure or stay delinquent? Each definition moves the trigger more than the headline number does.

3. Upside is soft; downside is hard. The performance step-up you negotiated — advance rate rising from 80% to 90% if losses stay low — arrives in documentation wrapped in conditions: a defined "increase event," a lender confirmation, sometimes lender discretion. The step-down — advance rate cut on a trigger breach — arrives automatic, immediate, and unconditioned. And when the facility gets amended someday (Part 10), the pattern completes: negotiated upside is the first thing traded away; downside mechanics survive every amendment. Price a facility on its guaranteed floor, not its conditional ceiling — and in docs, fight to make upside as mechanical as downside. Symmetry of automation is a legitimate, winnable ask.

4. Terms appear from nowhere. The documentation phase can add entire obligations that were in no draft of the term sheet. The classic: a condition precedent that you close an equity round — of a specified minimum size, occasionally at lender-influenced terms — before first draw, plus a covenant to raise more by a date certain. Your lender just scheduled your fundraise, with a default attached, in a document your equity investors will never read. Other regulars: mandatory minimum draws (you must borrow on schedule — with unused fees if you don't and obligations you didn't plan if you do), preapproval rights over your first N partners or channels, and insurance requirements with the lender as beneficiary. The defense is procedural: a "no new material terms" understanding in the sheet itself, and an explicit list in the sheet of every condition precedent to closing. Anything that appears later gets met with "that's not in the sheet" — which still works post-signature, if the sheet was thorough. This is why the sheet must be thorough.

5. Cash control creeps. Sheet: "collections account under lender control." Documents: every account of the borrower — sometimes of the parent and every affiliate — under control agreements before the first dollar moves; read-only access to all of them; advance notice before opening any new account; caps on balances outside controlled accounts; sometimes lender approval over your payment processors and rails. Each increment is individually defensible ("we need to see the cash — we're lending against it"). The sum is that routine treasury operations now touch your loan agreement. Negotiate the perimeter (borrower accounts, not the whole family), the carve-outs (operating accounts with reasonable caps), and the mechanics (springing control that activates on default, rather than active control from day one — a standard structure that preserves your day-to-day banking while protecting the lender's downside).

6. Zero-cure defaults and hair triggers. The events-of-default section of a first-draft loan agreement commonly gives no cure period at all to: reporting obligations, financial covenants, collateral and account requirements, insurance, and every negative covenant. Meaning: a borrowing-base report delivered a day late is, technically, an immediate event of default — with the cash sweep, default interest, and acceleration rights that implies. Add the cross-default drafted to trigger on rights "whether or not exercised," and the subjective MAC clause, and a default is available to your lender more or less whenever they want one. Lenders rarely invoke hair triggers — but a facility that runs on waivers runs on goodwill, and goodwill is better spent elsewhere. Negotiate cure periods on everything administrative (3–5 business days on reporting is utterly standard once asked), materiality qualifiers on the technical, and thresholds on the subjective.

7. Remedies read like an operations takeover. On paper, post-default the lender can: collect directly from your borrowers, take exclusive control of the accounts, enter your premises, use your IP and brand royalty-free to run off the book, exercise a power of attorney in your name, and apply all cash in whatever order they like. You will not negotiate most of this away — it's the standard remedies kit, and its very comprehensiveness is why default triggers (patterns 2 and 6) deserve so much of your attention. The leverage story of the whole document is: remedies are absolute, so triggers are everything. Spend your negotiating capital accordingly — narrow the doorways into default rather than trying to soften what's behind the door.

8. Discretion hides in definitions. The last pattern is the quietest. Scattered through the definitions: "as determined by lender in its reasonable discretion," "in form and substance satisfactory to lender," "deemed uncollectible by lender." Each one converts a mechanical test into a judgment call that isn't yours. The two that matter most: anything touching the borrowing base (a lender who can deem loans non-performing at discretion can shrink your availability at discretion — that's a callable loan wearing a formula costume) and anything touching eligibility. Hunt discretion in those two neighborhoods and convert it: objective definitions, "reasonable" standards with deemed-approval timers, dispute mechanics. You won't get it all. Getting half is worth more than most pricing negotiations.

The defense

All eight patterns have the same root cause — open items resolving under inverted leverage — so the defense is one sentence: close the items while you still have the leverage. Operationally:

Every number, in the sheet, before signature. Every eligibility criterion. Every concentration limit. Every covenant level with its full test schedule. Every trigger with its consequence. Every fee. Every cure period that matters. The complete list of conditions precedent. Appendix D is the checklist — it's long, and it's meant to be worked through literally, line by line, with your counsel, before you sign.

Treat "market standard, we'll sort it in docs" as signal. A lender confident in their standard terms can write them down. Most will, when asked plainly: "send the exhibit from your last deal and we'll pre-negotiate it now." A lender who won't show you their standard eligibility criteria before signing is telling you the criteria aren't actually standard yet. Take that as information.

Stress-test the covenant package against your haircut projections — the Part 7 numbers, not the dream case. Walk each trigger against a mediocre-but-survivable quarter: cohort noise, a slow month, a channel wobble, a big partner pausing. A covenant you'd trip in a survivable quarter isn't a covenant, it's a tripwire — and per pattern 7, everything is wired to it. Then check the interaction effects, because triggers compound: the delinquency wobble cuts your advance rate (margin call) while the cash sweep traps your collections (liquidity squeeze) while the minimum-cash covenant still binds (default risk) — three separate clauses, one bad month, simultaneous. Model the month, not the clause.

Pay your lawyers to argue now, not later. Legal spend before signature buys terms; legal spend after signature buys explanations of terms. Same hourly rate, wildly different ROI. The cheapest facility I know of is the one whose founders spent an extra three weeks pre-signature grinding through exhibits — and the most expensive ones all signed fast and "worked it out in docs."

They did work it out in docs. That's the problem.