Margin Dollars Over Margin Percent
Judge unit economics on the right denominator and the dollars left, not the percent.
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 5
- Confidence
- 72%
The rule starts with a correction attributed to Bezos: it is not about margin percent, it is about margin dollars. If the dollars are large, a low percentage can be fine — but a low percentage demands enormous volume, which in turn demands high repeatability. The second half of the rule is about denominators. Measure DoorDash off gross merchandise value and the gross margin looks terrible; net everything else out and measure off the take, and the margins are decent. Nobody measures Airbnb off GMV either — they measure off take rate. Software confuses people here because you pay for the software, not for the transactions running through it, so the denominator question never arises. The reason any of this matters is that eventually the company must be a business and will be valued off profits and a multiple of profits, which is why the payback horizon can stretch with abundant capital but cannot stretch forever.
Origin
Extracted from Acquired — Alfred Lin's answer on whether Don Valentine's insistence on large gross margins still applies, combined with his own CAC-payback experience as CFO-side operator at Zappos.
Core principles
- 01The gross margin percent is meaningless until you agree what the denominator is.
- 02Marketplaces should be measured on take rate, not on gross merchandise value.
- 03A low margin percent is survivable if volume and repeat rates are high.
- 04Eventually every company is valued off profits and a multiple of profits.
- 05Capital availability lengthens acceptable payback periods, but not indefinitely.
How to run it
- 1
Fix the denominator first
Before computing any margin, state whether you are dividing by gross merchandise value, by net revenue, or by take. Most disagreements about margin are actually disagreements about this.
Pro tip Publish the denominator alongside the percentage in every internal deck.
Watch out Comparing a GMV-based margin to a take-rate-based margin produces a conclusion that is simply wrong.
- 2
Convert to margin dollars per transaction
Translate the percentage into absolute dollars left after the direct costs of one unit. Large dollars can carry a small percentage.
- 3
Stress-test volume and repeatability
If the percentage is genuinely low, you need a ton of volume, and volume you have to buy every time is not volume. Check repeat purchase rates before accepting the model.
Pro tip Repeatability is the hidden variable — low margin plus low repeat is unrecoverable.
Watch out Volume assumptions borrowed from a higher-frequency category will flatter the model badly.
- 4
Set the payback horizon deliberately
Decide over what period customer acquisition cost must be repaid — first order, a month, six months, a year — and justify it by market size and capital access rather than habit.
Pro tip Ask what horizon you would accept if no further capital arrived; that is your real floor.
Watch out The stretch is not transitive: extending from one year to four because one year worked is how the model breaks.
- 5
Underwrite to eventual profits
Confirm that the model, at scale, produces a profit pool that supports the valuation you are underwriting, because value ultimately resolves to profits times a multiple.
Watch out Abundant capital can postpone this reckoning for years, which makes it easy to forget it is coming.
In the wild
Look at DoorDash from a gross merchandise value standpoint and the gross margin is really low, which is how the whole category got dismissed as structurally unattractive. Net everything else out and look only at the company's take, and the margins are decent. The same correction applies to Airbnb: nobody measures its margins off GMV, they measure off take rate. The economics did not change between those two views — only the denominator did. Software obscures this because you are paying for the software rather than for all the transactions that happen on top of it.
→ A business that screens as unattractive on GMV margins screens as viable on take-rate margins.
At Zappos, e-commerce discipline required the payback on customer acquisition cost to land on the very first order. Alfred Lin notes that essentially no e-commerce company operates that way now, because with more capital available a company that knows the payback works on the first order can extend it to seven days, a month, six months, or a year. He is explicit that the property is not transitive — at some point the extension breaks, because you will not be able to raise enough capital to stay afloat.
→ Longer paybacks are defensible when the market is large, but only up to the point where financing risk becomes the binding constraint.
Common mistakes
Comparing margins across denominators
A GMV-based gross margin and a take-rate-based gross margin are not comparable numbers. Most sweeping claims about a category's economics are this error in disguise.
Assuming payback extension is transitive
Because a business survived a one-year payback does not mean two, three or four years works. At some point you cannot raise enough capital to stay afloat.
Low margin percent without repeatability
A low-margin-percent business needs a ton of volume, and that only works with high repeatability. One-off low-margin transactions never accumulate into a business.
Is it for you?
Best for
Evaluating marketplaces, delivery, e-commerce and any transaction business where headline margin percent understates the real economics.
Not ideal for
Pure software businesses where you pay only for the software and the margin denominator is already unambiguous.
From the transcript
“I think high gross margin, as Bezos would say, it's not about margin percent, it's about margin dollars.”
“If you're going after a low gross margin percent business, you just need a ton of volume. So you need high repeatability.”
“People don't think about this, but nobody measures Airbnb's margins off of their GMV. They measure it off of their take rate.”
From the episode
Special: Sequoia Capital's Investment Playbook (with Alfred Lin)