Deals rarely die at the moment everyone remembers. They die earlier, quietly, at one of three pressure points—and by the time the failure is visible, the language used to explain it ("the credit environment", "timing", "chemistry") usually points somewhere else.
Three words, each carrying one test. Not labels for people. Not stages to tick off.
Relationship
Test: when something goes wrong, does the information move—and how fast?
The common definition of a strong relationship measures warmth: frequent contact, mutual liking, long history. Warmth is pleasant and almost useless as a credit input. A relationship that only carries good news is a broadcast channel.
The businesses that are genuinely well-banked ring the bank before the covenant test is missed, before the customer failure shows up in the debtor ageing, before the shipment problem becomes a cash problem. And this is testable long before anything goes wrong: watch what happens to small unflattering facts early in the process. A client who volunteers the awkward detail—the concentration, the dispute, the delayed project—is showing you what the difficult quarter will look like. A client who lets you find each one yourself is also showing you.
The test runs in both directions. A banker who cannot deliver an early, plain "this will be difficult, and here is why" is storing bad news up for the moment it does the most damage.
Structure
Test: can the terms be written down without fiction?
Every proposed deal contains sentences that sound reasonable in conversation. "The shareholders will support the business if needed." "The inventory turns quickly." "The pipeline covers it." The structural test is whether those sentences survive being written into documents that people must actually perform against.
For each load-bearing claim, ask: who signs this, and what happens to them if it turns out to be untrue? A claim nobody will sign is not necessarily false—but it cannot hold weight in the structure, and a structure that leans on it anyway has fiction built in. This is the same story-versus-evidence discipline that defines bankability, applied at the drafting table.
Structural failure is rarely dramatic. It is the working capital need packaged as term debt. The facility limit that quietly ignores the seasonal peak. The covenant calendar set against a generic borrower instead of the actual cash cycle—common enough that it has its own note. The damage surfaces later, disguised as a performance problem.
Execution
Test: is the first difficult quarter survivable?
Everything performs in the base case. Assume the base case fails once—one soft quarter, one delayed receipt, one covenant test that lands awkwardly—and ask what happens next.
Mechanics: what is the sequence? Who calls whom, what gets waived or reset, what does the documentation actually require to happen? Headroom: is there enough distance between ordinary volatility and formal default that ordinary bad luck does not trigger extraordinary process? People: are the individuals who will handle the difficult quarter the same ones who negotiated the good one, and do they have the standing on both sides to manage it without escalation?
A structure that only works while everything works is not a structure. It is a bet with paperwork.
The order of failure
The reason to hold the three together is that they fail in sequence, each leaving fingerprints readable early.
The relationship cracks first: bad news starts arriving late, or via third parties. Structure follows, because once information stops moving, the terms drift out of alignment with reality and the covenants begin measuring the wrong things. Execution failure arrives last and loudest—the missed test, the difficult meeting, the workout—but by then it is mostly the visible result of the two quieter failures that preceded it.
Read backwards, this is a diagnostic: an execution problem invites the question of what the structure missed; a structural problem invites the question of what the relationship failed to carry. Read forwards, it is a maintenance schedule. The cheapest place to fix a deal is at the point where bad news still moves freely.
This lens covers the process risk of a deal. For the full diagnostic across an opportunity—including the business itself and the state of its information—see five dimensions of bankability, where these three appear as dimensions three through five.