Three risk frameworks I lean on: a one-page red-team memo, a downside math table, and the first-loss question nobody loves.

The one-page red-team memo (write it like you mean it)

I stole this habit from a former chief risk officer who had zero patience for pretty decks. Before any high-stakes transfer, I ask for a one-page red-team memo written by someone who is not emotionally attached to getting the deal approved. Not a slide. Not a model. A page of plain language that tries to kill the decision.

Format matters because people will hide behind formatting if you let them. I want four headings, in this order: What has to be true, How this fails, What we'd see first, and What we'd do. The middle two are where the value is. "How this fails" isn't allowed to be vague. I want mechanisms: customer concentration turning into a covenant breach, a key supplier change that forces a recertification cycle, a regulatory interpretation that makes a product claim unusable, a refinancing window that shuts because rates move and the market goes risk-off.

The "What we'd see first" section is my favorite because it exposes whether the team has ever been surprised in the real world. Good memos include small, early signals: sales cycles slipping by two weeks before pipeline looks ugly, a surge in support tickets about one feature, an auditor asking for a specific schedule twice, a lender pushing for a new definition of EBITDA. If the first indicator is something like if revenue declines, it's not a real indicator. That's just the failure restated in slower motion.

Then I look at the author line. If it's the deal lead, it doesn't count. If it's the person who will have to clean up the mess if it goes sideways, now we're getting somewhere. I don't need negativity. I need a clean attempt to falsify the story we're telling ourselves, before the wire happens and optimism becomes a sunk cost.

The downside math table I keep on one screen

Here's my personal rule: if the only thing on the screen is the base-case model, the meeting is already drifting toward approval. So I build (or demand) a downside math table that sits next to the base case like an annoying sibling. No Monte Carlo theater required. Just a grid that forces adult questions.

The rows are the few variables that can actually break the outcome. Not fifteen lines of noise. Think: revenue timing, gross margin compression, churn, refinancing terms, litigation/settlement exposure, and any single operational constraint that can cap throughput (a plant, a cloud bill, a single licensing agreement). The columns are a small set of scenarios with names you can say out loud without blushing: Base, Soft, Hard, and Stupid. Stupid is the one people resist, which is exactly why it belongs. It's the one where two bad things happen together because reality doesn't schedule its punches.

Inside each cell, I don't want a full P&L. I want the two or three outputs that matter to the decision: cash runway in months, covenant headroom, and the probability we have to do something we hate (raise at a bad time, sell assets under pressure, cut the product roadmap in a way that changes the business). If it's a credit decision, swap in interest coverage and collateral coverage. If it's an acquisition, show the purchase price multiple on that scenario's EBITDA, not the one everyone wishes for.

Then I add one line at the bottom that I call the breathing line. It's simply: how many months can we operate before we have to call the board with a plan B that includes the words layoffs, covenant waiver, or emergency capital. When that number is short, we stop arguing about whether the upside is big enough and start talking about whether we're comfortable living with the clock.

I've watched very smart teams melt down because the downside was technically acknowledged but never made visible. When the hard numbers sit in a grid, arguments get specific. People stop saying we'll manage and start saying who will manage what, by when, and with which lever.

Ask who takes first loss, then sit in the silence

When I'm trying to understand a deal's risk without getting hypnotized by the narrative, I ask one question that sounds simple and usually isn't: Who takes first loss if this goes wrong? Then I stop talking. The silence is part of it. People will fill it with the truth if you give them enough room.

In a clean setup, the answer is immediate. In messy setups, you get a sentence that starts with Well, it depends and then a tour through legal structure, side letters, guarantees, and informal promises that aren't written anywhere. That's the point. First loss tells you where the actual risk sits, not where the PowerPoint says it sits.

I run it through three passes. First, contractual: what do the docs literally say about recourse, collateral, reps and warranties, indemnities, and caps? If someone says we're protected by indemnification, I ask: protected up to what amount, for how long, and what's the process to collect? If the protection relies on a party that could be insolvent at the exact moment you'd need them, it isn't protection. It's comfort.

Second, operational: who eats the pain before the lawyers even show up? Is it customer support handling chargebacks, the finance team funding working capital while payments lag, the sales org discounting to keep logos, or the product team bending the roadmap to satisfy a single partner? Operational first loss is where manageable becomes expensive, quietly.

Third, reputational: whose name is on the press release, whose clients will call, and who has to explain it on the next earnings call. Reputational loss doesn't show up in a spreadsheet on day one, but it can change your cost of capital and your hiring pipeline faster than people want to admit.

If the answer ends up being us, I'm not automatically out. Sometimes that's the right bet. But now we can price it and structure around it: tighten covenants, add triggers, stage capital, require better reporting, or walk away if we can't get transparency. Asking about first loss is less about being cynical and more about refusing to outsource accountability to a story.