Your Best Margin Is Your Worst Customer

Gross margin reports what your ingredients cost. Fully-loaded unit economics reports what the business lost. The two numbers are rarely the same.

Blue Apron reported a gross margin around 31% at its 2017 IPO. A year later the reported number had slipped under 29%. Analysts reading the same filings reported that nearly 70% of its customers churned before they ever broke even. Both were true at once, which is a useful reminder that gross margin is a line on a statement, not a verdict on whether a business works.

(This happens more than anyone writes down. The margin is real. The business loses money. Those aren't contradictions. They're accounting choices.)

The gap between those two numbers is the whole game. It's where most ops problems hide, where forecasts go wrong, and where a company's best-performing product quietly eats the margin the rest of the portfolio earned.

Gross margin is revenue minus cost of goods sold. Ingredients. License fees. The thing the customer pays for. It leaves out the implementation team that spends a month onboarding the account. The success manager assigned because the logo is "strategic." The engineering hours burned on a one-off integration nobody else uses. The refund a rep quietly gave to close the quarter, the kind you catch only when you know who you're negotiating with. The legal review a redlined contract triggered. Those costs land elsewhere on the statement, or nowhere at all.

The result is a product that looks like a high-margin winner and, once you load in everything ops bled to deliver it, runs at negative contribution.

A services business I worked with had a flagship offering, the highest-margin line in the deck, that needed a senior engineer parked on each new account for its first ninety days. That time never touched COGS, because engineering sits under R&D. On paper our flagship earned 62% gross. When we finally built a fully-loaded cost-per-account model, it earned 11%. We had funded two years of growth on the margin from a smaller line nobody in sales wanted, because the quota math on it was thinner.

(The quota math was thinner because commissions were set off reported gross margin. Which, again, was the wrong number.)

The pattern is simple to name and easy to miss. Reported margin is a measurement choice. Fully-loaded margin is the business. The measurement choice serves reporting consistency and investor comparability. It was never a management tool. Run the company off it and you'll make decisions that survive an audit and fail in hindsight.

The symptoms are predictable. Sales leads with the high-margin product because the commission is better, so volume concentrates there. Ops staffs for that volume, and the cost creeps into G&A instead of COGS. Finance models growth off a gross margin trend that looks stable. The board funds a bigger sales team. Eighteen months later cash is tighter than the plan implied, a consultant builds a fully-loaded model, and the flagship is the problem.

Nobody lied. Every number was correct. They weren't the numbers that describe whether the business works.

The correction isn't a new dashboard. It's deciding, once, which number the team runs the company off of, then pointing commissions, forecasts, and hiring plans at that number instead of the one on the P&L. It's the same discipline as competing on belief instead of price: pick the metric that reflects the business you have.

Two companies got there early. Amazon spent twenty years telling investors to watch free cash flow per share and ignore reported margin, a stance that reads as eccentric until you see it as a refusal to be flattered by the wrong metric. Costco's membership economics make the gross margin on the goods themselves close to beside the point. In both cases the executives picked the number that described the business and ran on it. Reported margin became a side effect, not a target.

Most companies do the reverse. They pick the number that's easiest to report, because the accounting system already produces it and the investor deck has a slot for it, then staff and sell and forecast against it. So the best-looking product absorbs the most operational load, the loss shows up in departments sales never looks at, and by the time the real number surfaces they've already hired and committed off the wrong one.

The right question isn't what is our gross margin? It's what does it cost us to keep this customer next quarter? Different numbers. The first is an accounting output. The second is a management input. Confuse them and you grow into a cost structure you can't walk back.

Before your next board pack, ask:

• If I loaded every hour of engineering, CS, implementation, and legal into this product, what does its margin become?
• Which line is sales paid to push, and is it the one ops would choose if they paid the bill?
• What's the smallest gap between reported and fully-loaded margin that would change our hiring plan this year?
• If we stopped selling the flagship tomorrow, would we lose money or make more of it?
• Who in the room can tell sales to stop leading with the product that's eating the company?

Answer those and the flagship stops being the hero of the deck and starts being a line you manage. The margin on the slide was never the business. The cost to keep the customer is.