At the bottom of this post you’ll get a quick round up on who’s taking the CFO seat at PE-backed companies. Scroll down for the latest from The Knot, Paramount, and AuditBoard.

This scene reminds me of the forecasting process at early-stage companies. The number of cycles I've wasted over-engineering a six-tab Excel forecast just to be 40% off—when I could have been 10% off using simple multiplication on a yellow legal pad—is more than one.

I recently talked to Kevin Drost, CFO at New Era ADR, a PE-backed company modernizing dispute resolution in the legal space. Before that, he helped scale and sell Reverb, a used instrument marketplace, to Etsy. He's gone from literally having no data to forecast the business, to having a well-oiled BI team tasked with forecasting to Wall Street within 50 bps.

He's also seen what happens when you try to force big-company rigor into a business that's not ready for it.

In the earliest days, Kevin argues, forecasting serves a completely different purpose than it does later in the lifecycle. It's about setting a shared hypothesis for where you're going—which in many cases, is directional.

The business doesn't have the historical inputs or pattern recognition to justify anything close to certainty. Trying to model cash flows or LTV:CAC with three decimal points of precision becomes a wasted exercise.

I remember joining a sub-$10M company and trying to predict what our first quarter's sales in Asia would be. We had never sold a single thing in Asia, and the rep we were relying on was calling from a landline in Europe. Talk about trying to nail jello to a tree. Yet… did that stop me from building a four-tab model? Nope.

Those first forecasts are really about internal alignment. What are we aiming for? How will we know if we're making progress? When do we need to hire, adjust pricing, or raise capital? These are directional bets, not engineered outcomes.

As Kevin put it,

One of the hardest parts of early-stage forecasting is the absence of data. I asked Kevin how he approached market sizing when Reverb was just getting off the ground.

So they found a proxy:

It wasn't elegant. But it was functional.

The goal isn't to be right. That's impossible. It's to be right enough to get the car out of the driveway so you can allocate resources, test your roadmap, and course correct along the way.

This is where a lot of teams fall into trouble. They build highly complex models with dozens of inputs, all formatted cleanly and backed by "data" that, in reality, isn't meaningful. Kevin calls this the trap of false rigor.

This is a common mistake, especially for operators coming out of big companies or consulting (guilty as charged with the big braining it). They think forecasting is about building a beautiful model. It can make you breakfast and predict sunset in any geography, but can't tell you how many software licenses you'll sell next month.

Early-stage forecasts are about decision enablement:

Counterintuitively, sometimes you need to "do less" in order to see the picture more clearly.

Once Reverb had enough transaction data, they refined their approach. They weren't just looking at return on ad spend—they were tracking the marginal return on each new dollar. That allowed them to push marketing spend up to the point where returns started to flatten, and no further.

That level of control only comes when you pair analytical rigor with solid data. And also understanding where and when your customers want to transact.

Speaking of that… you earn a series of secrets on your customer profile if when you embed yourself into the customer experience.

At Reverb, every new hire was given $1,000 each quarter with one directive: buy and sell as much gear as you can on the platform. Whoever had the most money left over at the end (and the most transactions) won.

This wasn't a quirky startup perk. It was a systems-level onboarding strategy.

It worked. Decisions were better because the team had firsthand experience. You can't build an accurate model of behavior you don't understand.

That scrappy mindset doesn't last forever. When Reverb was acquired by Etsy, Kevin had to completely change how he thought about modeling. Forecasts weren't just for internal use anymore. They were shared with a corporate parent, embedded into investor decks, and had to be bulletproof.

Having an entire BI team at your disposal, with a decade of historical data cleanly segmented by customer type, is a much different scenario than where he started—with no data at all.

Forecasting is a muscle. And like any training program, the intensity should match your current level. Too little rigor, and you're making blind bets. Too much, too early, and you're wasting time on manufactured certainty.

Some of the best finance leaders I know are ruthless simplifiers. They love a ten-tab model that takes up 124 GBs. But they really love a plan on one page of a yellow legal pad that tells them how many reps they should hire to produce $10M in incremental revenue.

Because yes, every model is wrong. But the decisions it supports? Those need to be right.

Kevin explains the art of Cowboy Forecasting, and how to evaluate the level of effort you put into your forecast across various stages.

Thanks for reading, and make sure to check out our sponsor, Abacum.

Download the 2026 Finance Guide: Lessons from the Trenches to get the tactical frameworks used by top PE-backed leader for FREE.