We recently helped Ethos, Klaviyo, Superhuman, and Restaurant 365 fill key finance roles.

We specialize in Strategic Finance and FP&A.

Oh, and if you’re a candidate passively open to kick ass roles (even if you’re in a great role now) and you haven’t entered our warm talent pool, you can do so here. Takes 3 min.

We are currently working on:

Duration risk is a funny thing.

I remember learning about it in macroeconomics freshman year in college. The class was taught by an 85-year-old guy who I'm pretty sure was also an economist during the Great Depression (bad time for America, GREAT time to be an up-and-coming economist).

I also had a boxing coach in college who often remarked on the standard three-minute round, "Three minutes is a long time, no matter what she says."

Which is a different type of duration risk.

Anyway, I vaguely remember discussing duration risk in economics class. At the time, I basically understood it as not writing a check your ass can't cash. Which is a double negative, and technically not the definition, but we're going to run with it for a sec.

My friend Brian Morrissey recently interviewed me about the Anthropic S-1 on his podcast about media stuff (very meta) The Rebooting. We got into duration risk, and more specifically, trying to time your market opportunity.

Listen / Watch

He had a quote about the current AI buildout I liked:

Life on the rails!

The people investing were directionally correct. Railroads fundamentally changed how goods and people moved around the country, opened up new markets, and created an enormous amount of economic value. Chicago, Atlanta, and Omaha are all products of the railroad boom (you can thank me if you ever get on Jeopardy).

Unfortunately, a lot of the people financing that expansion didn't make it to the “returns” part of the cycle. In fact, most of them went broke.

I think there's a legitimate risk of something similar happening with AI.

The No. 1 question in my mind is whether the use cases for AI can keep growing at the same pace that the cost of tokens is falling. And whether all the infrastructure being built today will be economically competitive by the time the latent demand arrives.

I doubt that he’s alive anymore (due to his duration) but I think this is what my college economic professor would call duration risk.

Most readers aren't running an LLM company. I sure as hell ain’t. So let’s talk about the concept more broadly.

Technically, duration risk is how much a bond's value moves when interest rates do, and the longer the duration, the more sensitive it is. Longer life of bond = more surface area for interest rates to change during its life and change what it’s worth.

The business equivalent is how far into the future you've committed yourself to an economic assumption that may no longer be true.

Think about signing a 10-year office lease based on your expectations for headcount and office utilization. How many SoHo ghost towns did COVID + ZIRP produce? Or signing a five-year customer contract at a fixed price, with no idea what your cost to serve will look like in year four (“oh shit, we have to build AI into this thing to stay competitive now?!”)

Both can be fantastic decisions if your assumptions hold. The trouble starts when they don't… particularly when the revenue you're counting on and the expenses you've committed to have very different durations.

Imagine two companies, each doing $100 million in revenue and $20 million in EBITDA.

Company A has customers on three-year contracts with annual price escalators, relatively flexible vendor agreements, and a cost structure where a meaningful chunk of expenses can flex with demand.

Company B has customers who can cancel every 30 days, plus a five-year office lease, long-term vendor minimums, and a pile of other fixed commitments negotiated under the assumption that today's revenue won’t churn.

They’ve got identical 20% margins today, but I'd pay considerably more for Company A, all else being equal. If revenue falls 20%, Company A has contracts slowing the decline and costs it can pull back. Company B has neither, so the same revenue hit does a lot more damage to its margins and, more importantly, its cash balance. It would experience a nonlinear unwinding.

Some of the assumptions you make are about your own company. Maybe you hire 50 salespeople because you think you're going to double revenue, only to discover that your product just isn’t that good.

Macro shocks are obviously harder to control for… the whole market grows slower than expected, or rates stay higher for longer and your customers stop spending, or a competitor launches something substantially cheaper and the pricing behind your business model suddenly looks ridiculous, or Anthropic now allows you to remove the backgrounds in photos (RIP to Canva).

I've often said that many markets only have an 18 to 24 month window of opportunity. Webvan and Instacart are v. similar, except for their timing.

Again and again we’ve seen you can be completely right about the market opportunity and still run out of money while waiting for it to arrive.

I'd be more willing to throw the ball deep on the revenue side than the expense side. Perhaps that's the CFO in me. I'd be much happier locking in five years of revenue than five years of expenses (caveat: if the product burns a lot of compute and the customer wants fixed pricing for all five years, I'd want a good view on how much cost to serve could move in that window… not that anyone actually knows with certainty… hence why so many of us have moved to usage based models).

If I were to add up all the instances I’ve told sales to take that deal off the street, I’ve been happier more times than I’ve been sad. Book that ish.

Because if I know a customer is committed to paying us for the next three years, I'm more comfortable hiring people, building products, and making other investments to propel the business forward. I have a better baseline to plan off.

Now, of course, you can't wait until every customer has signed before building a product, hiring a sales team, or opening a new location.

I'm not suggesting you match the duration of every customer contract to every vendor agreement. That would be psychotic.

Some amount of investment has to precede demand. No risk it, no biscuit, as Bruce Ariens used to say.

But it doesn’t mean you can’t be acutely aware of the delicate balance between your commitments.

There are a few back-of-the-envelope exercises I run to see how much of our future we've already committed to.

For each of the next three years, calculate what percentage of your revenue forecast is backed by signed, non-cancellable commitments, and what percentage of your expense forecast is already contractually committed.

Take your expected net new revenue and push it out 18 months, while leaving committed expenses exactly where they are.

Take your ten largest expense commitments and assume revenue falls 20% tomorrow.

And finally, at times like these we recite the three holy rules of being a cfo:

(Oh, and if you’re still reading I've got some railroad bonds I'd like to sell you.)

Wishing you trade at a high revenue and EBITDA multiple,

CJ