Mostly Controllers: Why Gross Margin Is a Decision (Not Just a Calculation)

Is the journal entry dead?

As the primary unit of accounting, maybe.

Today, your accounting team posts JEs to a GL. Then your FP&A team uses a separate stack of tools to rebuild the context your JEs stripped out.

What if both teams worked from one financial data warehouse that kept full context? Accounting logic becomes a locked, auditable layer on top, not the whole system of record.

Debits and credits are still there. They’re one view of the data instead of the whole picture. Enter data-centric accounting. Download Numeric’s Architecture for Agentic Accounting white paper (and start questioning JEs).

Welcome back to our second edition of Mostly Controllers.

I’m good with titles, I know.

And as a reminder, I'm a CFO. What I'm not is a CPA. I think I actually got a B- in my managerial accounting class at BC.

So why is a non accountant writing a series of posts specifically for Controllers? Well, I'm def not here to tell you how to be an accountant.

(I still don’t understand lease accounting. But to be fair… does anyone really understand lease accounting?)

I'm writing this as a translation layer between the work you do and the decisions that are made off of your work. This probably doesn't come as a shock, but there's a lot of derivative “stuff” made from the financials you produce. The remixes come from CFOs, Investors, and Acquirers.

The first time around we explored why CFOs, Investors, and Acquirers get excited about deferred revenue (the largest thing we owe).

Today we’re talking about gross margins, and why there is perhaps more art than science involved.

Before we get into the meat and potatoes… If you’re a Finance person (like me) plz share this with your company’s Controller so they can be our email pen pal. Also, sharing makes you look wicked smaht.

HIT FORWARD MY DAWGS

There’s a war waging inside your cost of goods sold. And the more that gets thrown out and stuffed into OPEX, the better your story is. I’ve had the same convo I’m about to have with you with so many Controllers in my career.

And we must defend that border with logic.

And there has never been a more crucial time to feel rock solid on your cost of goods sold classifications.

AI costs are blurring between:

But first, let’s do a quick refresher on what stuff we stuff “up there”.

Gross margin is “simply” revenue less cost of goods sold.

But when you’re evaluating a cloud hosting bill, a massive anthropic token allocation, the cost of customer support, or the growing payroll of forward deployed engineers, you’re actually making a valuation decision.

And GAAP does give some latitude. There’s a gray zone between “cost to deliver” and “cost to run the company”. And Controllers are the big brains behind that judgment.

Gross margin informs how many dollars are “left” to run the company. COGS are considered a necessary evil in many ways, what you need to spend to just keep customers on your product. It’s a theoretical limit on your profitability. Therefore, Gross Margin gates how “aggressively” CFOs can spend on growing the business.

A dollar of Customer Success looks MUCH better in S&M than in COGS. And a dollar of AI Tokens looks much better in R&D as the cost to produce the product than in COGS to serve the product.

The reasoning is that OPEX is theoretically all fueling future expansion. COGS is serving the sales made in the past.

CFOs also care about picking a consistent definition they can reliably claim each quarter. Nothing looks worse than an investor questioning if the team changed how something was calculated from period to period.

Investors care less about the absolute dollar of margin and more about what the percentage signals for scalability. From their perspective, every incremental percentage of GM drops to the bottom line if you run this business out in perpetuity and it becomes highly OPEX efficient. Yes, this is very theoretical. I get it. But this “operating leverage” is where “software premiums” come from.

True software companies (pre AI) would run at +80% gross margins. Usage and infra companies (which all companies are becoming) run at +50% (usually topping out at 75%). And any services or implementation heavy companies see their margin dragged down accordingly.

Your gross margin very much sorts you into a category, and each category has a range of multiples you may earn. Think of it as a floor and a ceiling.

COGS also drive both CAC Payback and Customer Lifetime Value (CLTV), two metrics investors are obsessed with. Well, at least gross margin should be contemplated. The #1 metrics crime I see is NOT gross margin adjusting these figures. So as a Controller, remember that your gross margin classification goes far beyond the standard P&L, and moves these efficiency metrics up or down.

Another point - gross margin trends matter just as much as gross margin levels. A margin sliding down over time is the exact opposite story investors want to see (even if it was relatively high before). In fact, it’s often a better story to start really low and demonstrate improvements over time (many AI companies are actually negative gross margin, losing money on every transaction… uber and Instacart were like this starting out and eventually got it sorted). Improving gross margin is a big part of showing maturity in your business model. It allows the investor to “imagine” what this thing looks like at scale.

Acquirers are always worried about what they’re absorbing into their own “pristine” gross margin (they always act like their GM is holy, even if it’s worse than the target’s).

A buyer’s quality of earnings (QofE) team re-derives your gross margin from scratch (literally doing a build up at the transaction level). They’ll pull support, success, infra costs, and anything else they believe is true and try to lard the line up as much as possible to argue you are not nearly as efficient as they thought… and should get dinged on valuation.

Part of this is a negotiation tactic. They want to knock off a few dollars. But it’s also hedging against their own value creation plan with their own shareholders. When acquiring, they want to under promise and over deliver for their investors.

During this process they’ll also normalize for what they plan to carry on the books. They probably get a better deal on cloud costs than you because they are committing to more spend. And they probably don’t need all your customer support personnel if they already have a call center in Utah. This is a savings, or a synergy, they’ll realize.

So in many ways an investor will look at your GM trend and say “what could this become” while an acquirer will look at it and say “what can I add in to start, and take out once they’re aboard.” One view is theoretical and the other is extremely surgical.

We’re going to look at one company but classify stuff differently and see what it does for valuation.

Call it a $50M ARR business. Here are the costs that actually exist, whether you like where they sit or not:

The fight is going to center over Support, Tokens, and Solutions Engineers. It’s a chunk of $7M in they grey zone, and can swing your story re: operating leverage either way.

If we draw the line aggressively:

Now draw it conservatively, everything it actually costs to serve the customer:

So that’s a 14 point swing. What’s funny is you spent the exact same amount. And your EBITDA is the same; it’s $6M in both scenarios. But we’ve shifted 10% over the gross margin border. We didn’t make any costs disappear, but we took a lot of wind out of our story.

Let’s revisit the same scenario.

Adding it up, we’ve moved +$3M out of COGS and into other parts of the P&L.

This got us six GM points back, and actually represents how the org works. We haven’t changed a customer contract or anything related to revenue, and our EBITDA is still $6M. But we’re sorting ourselves into a better comp bucket as we sort our COGS.

Wishing you a gross margin that is high and also intellectually defensible,

CJ