We pulled the annual filings of nine public education companies straight from the SEC's XBRL data and lined up one number: sales and marketing as a share of revenue.
The pattern is cleaner than we expected.
| Company | Model | FY | Revenue | Gross margin | S&M | Result |
|---|---|---|---|---|---|---|
| Duolingo | D2C, organic | 2025 | $1,037.6m | 72.2% | 12.1% | +$182.4m pre-tax |
| Chegg | D2C, organic search | 2021 | $776.3m | 67.2% | 13.6% | about break-even |
| PowerSchool | B2G K-12 | 2023 | $697.7m | 59.3% | 30.8%* | −$31.1m |
| Nerdy | D2C tutoring | 2025 | $179.0m | 58.0% | 33.6% | −$39.9m |
| Coursera | mixed | 2025 | $757.5m | 54.6% | 33.8% | −$51.0m |
| Instructure | B2B LMS | 2023 | $530.2m | 64.9% | 37.3% | −$34.1m |
| 2U | B2B2C | 2023 | $946.0m | — | 39.3% | −$317.6m |
| Udemy | marketplace | 2025 | $789.8m | 65.6% | 41.3% | +$3.8m |
* PowerSchool reports combined SG&A rather than standalone sales and marketing, so its figure overstates pure selling cost and is not strictly comparable.
Two companies have ever run below 20%. They are the two that reached profitability. The seven above 30% lost more than $3.3 billion between them.
This is not a story about gross margin
The obvious explanation would be that the losers sell something cheaper to deliver. They do not.
Gross margins across the set cluster between 54% and 72%. Instructure, running a 64.9% gross margin, lost money. Coursera, at 54.6%, lost money. Duolingo, at 72.2%, made money — but the gap between Duolingo's gross margin and Instructure's is eight points, while the gap in sales spend is twenty-five.
Nobody in education earns software margins. The best in this set is 72%, and the median is around 63%. That is a services business wearing a technology multiple, and it means there is very little room above the gross line to absorb an expensive sales motion.
The variable that separates these companies is not what it costs to serve a student. It is what it costs to find one.
What the expensive channels have in common
Look at which models sit above the line. Institutional software: Instructure, PowerSchool. University partnerships: 2U. Consumer tutoring marketplaces: Nerdy. Mixed enterprise and consumer: Coursera, Udemy.
Every one of them acquires customers through a human being.
2U is the clearest case, because the model made the cost structure explicit. Signing a university to an online degree partnership took years, and the company disclosed as much to its own investors. It spent 39.3% of revenue on sales and marketing in 2023 and lost $317.6 million in that single year — more than the other six loss-makers combined.
Below the line, both companies acquire without a salesperson. Duolingo grows through the product and word of mouth. Chegg, at its 2021 peak, grew through organic search: students typed a homework question into Google and Chegg's answer page was the result. Neither was paying for distribution.
Chegg is also the warning. That 13.6% was not a durable advantage — it was a dependency on a channel it did not own. When Google began answering those questions directly, the traffic went away, revenue fell from $776.3 million to $376.9 million, and sales spend rose to 18.2% while the business shrank. Cheap acquisition through someone else's platform is rent, not a moat.
Why this matters more in this region than in the US
A US EdTech company that spends 35% of revenue on sales can raise a growth round and keep going. The public filings above are, in part, a record of companies doing exactly that for years.
That option does not exist here. Across the last 24 months we could find exactly one disclosed MENA EdTech round above $10 million. Everything else came in at $7.4 million or below. There is no growth stage to grow into.
Put those two facts together. A founder in this region who builds a business requiring 35% of revenue in sales spend has built a business that needs a Series B that the market does not currently supply. The distribution model is not a go-to-market detail to be optimised later. It determines whether the company can be financed to profitability at all on the capital actually available.
What we look for
We are not arguing that B2G and institutional sales are uninvestable. Ministries and school groups are where the students are, and in this region they are where the budgets are too.
We are arguing that the cost of the channel has to be underwritten from the start.
Does acquisition require a salesperson? If yes, the plan needs to show how gross margin covers that cost at scale, not how volume will eventually fix it. In this set, volume never fixed it.
Who owns the channel? Chegg's 13.6% belonged to Google. If your low acquisition cost depends on a platform you do not control, model what happens when that platform changes.
Does the product carry any of the sales load? The two companies below the line both have products that recruit. That is a design decision made early, not a growth tactic bolted on.
What does the company look like at $10 million of revenue? Not at $100 million. Ten. On the capital this region actually deploys, that is the point the business has to work.
The line itself
Twenty per cent is a threshold, not a law. Nine companies is a small sample, and two of the figures come from years when the businesses were structurally different from today.
But the direction is consistent, it comes from audited filings rather than a benchmark deck, and it points at the thing EdTech founders most often treat as solvable later. In a market with no Series B, distribution cost is not a line item. It is the plan.