Now that we've covered a little bit about why 'scalability' is a complex and potentially misunderstood term and how I think certain companies prioritize scalable components of their business, let's return to the original question that started it all:
“What did you mean when you wrote: Education businesses are not globally scalable, but locally scalable…?'
Here's what I mean:
When we examine all of the characteristics of scalability that I've previously talked about, I think you’ll find that they are much more challenging to optimize on a global or national level when compared to a local level (meaning a certain region, city, or even hyper-local, a neighborhood). Your initial reaction might be, “Yes of course - that makes sense. But why is our business or the education business generally more impacted than any other businesses in that sense?"
The 'iron triangle' of business is particularly tough on education. When we talk of the iron triangle, imagine a Trade-off Triangle (like the Play available on the KnowledgeBase) that has three corners: 1) Quality, 2) Speed, and 3) Price. Generally, it's said that you can only have two of the three in a business. For example: high speed and low price but low quality = McDonalds burgers.
Every industry and sub-industry has its own idiosyncrasies, but in education we can see the iron triangle as 1) Access, 2) Cost, and 3) Quality. Access refers to number of learners at a given time and could even more specifically mean teacher-student ratio. Cost is fees parents and students would want to pay. Quality is the level of qualification of the educator and other pupils. So, for example: Khan Academy has amazing educators, is free or nearly free to access, and has incredibly wide access online, but as a student there's no personalization due to the wide access and low cost.
Another example would be Cambridge University: The quality of professors and fellow students is meant to be incredibly high, but due to that and the teacher-student ratio being incredibly low, the cost also turns out to be high. Now, companies, governments, and non-profits seek to 'break the constraints' all the time. The great promise of internet in education was that access to world-class education at a fraction of the cost would be available to all at the touch of a button. While this is nominally true (and you can learn a great deal by reading and watching YouKu and YouTube videos, there is still more to be desired. One company who sought to break these constraints is VIPKid, an online one-on-one tutoring service here in China. The access is massive and popular across that in fact have a one-to-one experience (in this way VIPKid sought to break this constraint), and the cost is middling (with one session costing parents between 120-200 RMB last time I checked). But the quality of education is known to be rather rudimentary and not particularly effective - we see this because of their high churn rate (low retention rate).
If the education was super effective, then more students would stay on the platform. So even with billions of dollars of capital, they have been unable to completely break the Access, Cost, and Quality constraints. In education, which is the industry we can consider LL to be in for the time being, there are many different steps in the value chain where companies can contribute to student learning and growth. There are models companies take, including live classes, online courses, institutions (K-12 or other), consulting or services models, contracting, educational products or curricula, or content (knowledge) production, among many others.
After scouring as much of the data that I can find on publicly-listed companies in these different verticals and thinking of education’s iron triangle, I find the following: in operating schools in K-12 or brick-and-mortar LearningCenters, fixed and variable costs are tied inextricably to geographic distribution, meaning geographic expansion does not inherently increase scale and I believe it actually hinders it. Fixed costs are costs that don't really change, even when the units of output increase. Variable costs MUST increase with the units of output that increase. Imagine you run a dumpling shop. Fixed costs are the rent and the water boiler. Variable costs are the noodles, filling, and the vinegar.
In education, the way in which a business is set up largely defines its fixed and variable costs. The fixed costs as a proportion of the overall costs, the more scalable the business is likely to be. In the education business, MOOCs have highly fixed costs and some variable costs for new course development. Learning Management Systems are similar: mostly fixed costs with some variable costs for localization or new content creation.
Even further towards the fixed costs side would be education software companies, such as a typing or math software game. Once it's created, students can use them again and again and the company need not invest further, so each unit sold is at high margins. All of these businesses can be globally scalable because at a huge size their margins can maintain and even grow (sales and marketing costs notwithstanding). All of these businesses require larger upfront spending and investment - MOOCs, LMS companies, and Ed-Tech companies frequently receive venture capital investment. This is likely a reflection of the high potential for scalability as well as the upfront development costs. On the other hand, there are other education business models that are much more driven by variable costs. Operating schools, one-on-one lessons / face-to-face tutoring, and after school or academy models are all mostly driven by variable costs. The more lessons that are required to be taught, the greater the costs are associated with delivering that service. To be clear, this doesn't make them bad business at all! Just a different type of business that we need to explore.
If operating schools or one-on-one lessons was a globally scalable business, we would likely see a global school chain or ownership structure operating schools all over the world. The first point that I would note is that after looking at the history of publicly-traded companies, I haven't been able to find K-12 school operator company that has had much staying power at all on any stock market. While there are education companies are worth over 10BB USD (which makes them considered a large-cap stock), there sure aren't any K-12 school operators there.
Looking at the three highest profile international chains of schools for which financial data is somewhat or readily available reveals some compelling findings here about fixed and variable costs. Big thanks to Johnathan Knee for framing the comparisons of the three education companies below:
Nord Anglia Education (where some LL students attend school here in Shanghai) operates over forty schools in fifteen countries. GEMS Education is a Dubai-based international chain and has over seventy schools in fourteen countries. Cognita Schools is also growing in Europe, Latin America, and Asia. They operate 75 schools with the IB program, similar to Nord Anglia.
What's particularly fascinating about all of these 'global' brands is that their margins speak otherwise. For Nord Anglia, the company went public in Hong Kong in 2013. More than half of their profit came from China (from only 40% of their overall revenues). Before listing, two-thirds of their profits actually came from China! For GEMS, it's estimated that the company is actually losing money in every market away from the Middle East, where 50 of its 70 schools are located. For Cognita, 43 of their 75 schools are based in the UK. What's super interesting to me about this is not only the clear concentration of profits, but that the profits are based in geographic regions where the school chains seem to have some sort of local advantage. Nord Anglia was a first-mover in China (Cognita and GEMS don't even operate here), while GEMS leads in the Middle East where it has government connections and Cognita leads in the UK where it has the highest concentration of any of these three (~75% of the schools are in or nearby London!). That hardly strikes me like a truly international chain.
So why is this, that the profits of each of these three international school companies are so largely concentrated in geographic regions (beyond government connections/permits and first mover advantage)? The standing theory now is that they are able to share costs among the different schools within the same region. What are the two major assets that can be shared among different schools? (Because real estate and teaching talent will not be). It's sales and marketing and curriculum production.
Insofar as sales and marketing is concerned, this is actually played out exactly the same way at private for-profit American universities. Publicly available documents cite sales and marketing as the number one cost savings across multiple campuses or sites. It makes total sense -if you want to invest in a snazzy television spot or a YouTube video or even a flyer or logo creation, if that cost is now distributed across multiple sites or schools, you've just realized cost savings competitors cannot claim.
Insofar as curriculum production is concerned, this also makes a ton of sense. If there are a dozen schools that all share similar curriculum, the costs are distributed among them. If you produce the curriculum, then the costs are simply shared and if you license the curriculum, then the fees per pupil will certainly decrease as the package size increases.
What's pretty amazing is that both sales and marketing and curriculum production are fixed cost expenses! So that is a preliminary conclusion I believe we can draw: investing in sales and marketing expenses and curriculum development, provided they are largely fixed costs, is one way to improve the scalability of an education company. The corollary preliminary conclusion is that the more different the needs of sales and marketing and curriculum are depending on sites, the less this benefit accrues.
Let's take this conclusion to its next step. Imagine there are two school campuses: one in Shanghai and one in London. The curriculum should differ because of the government restrictions, local needs, or preparation for different exams. The sales and marketing teams and messaging should differ based on different needs of the local customers. There isn't a great deal of scalable infrastructure to be shared between the campuses, perhaps with the exception of some senior-level executives or finance team members, but even then, the differences between the sites will be so large that there will not be such great overlap. The ownership of schools in these two locations doesn't seem to me at all scalable. There is zero indication to me whatsoever that if we were to open another school in New York that our operating margins at each school would increase. And opening another one in Barcelona doesn't give me much more hope the operating margins would again increase at every campus around the world.
Playing the example a little bit more locally: let's imagine a school in Shanghai and a School in Hong Kong. We may see the same effects played out, though to a less extreme scale. Overall language requirements between the two schools would be more similar, though one is still Cantonese speaking and the other Mandarin or Shanghainese. Again, the sales and marketing teams need to do different work. Again, the curriculum needs to be different. So, there may be more overlaps than between Shanghai and London, but I still don't ultimately believe that if you opened a third school in Shenzhen that you would automatically get increases in operating margin across all three schools.
Let's take it one step further still, to a more LearningLeaders-specific example. What if we were to open an LL Beijing and LL Hangzhou? Certainly, we would see increased margins in our business because all of them are in mainland China, correct? As you might guess by this point, I don't think the increase would be existent, or if it was, would be large. But what I can say is that I think it would have a more positive effect on margins than opening in Hong Kong. And Hong Kong would have more positive effects on margins than opening in Singapore. And opening in Singapore would have more positive effects on margins than London and Barcelona and New York. Simply put, the closer to school campuses are (in the K-12 business) and provided that they are not cannibalizing demand at another campus/center, then I think you should locate them as close together as possible. This is because any potential shared costs, such as sales and marketing or management or curriculum, will be as similar as possible. Even the people in Beijing speak and act differently than those in Shanghai. Those in Los Angeles act differently than in New York. So, when it comes to an education business where complex solutions are critical, I believe scalability (and thus increasing operating margins) comes from doubling down where the company's structural advantage lies. In our case, it is reputation in Shanghai. Mind you, I don't think this argument trumps all arguments. I don't think this is a reason to never build a center in another city or country. What I do believe is that simply by looking at the operating margins and scalability of LL right now, since our advantage is local, we should continue to exploit that advantage for as long as possible. All things being equal, if we have 1MM RMB to invest in a LearningCenter, we’re going to get the greatest financial return if it’s closer, rather than further away from our current operations.
There may be a day 5-10 years in the future when we view our physical LearningCenters as fixed costs in and of themselves and the true operating margins for the company are derived from technology products (VR or otherwise). This is indeed the transition that TAL has been making in plain sight.
In summary here, scalability is greatest in business models where fixed costs play the majority of the costs. Average total cost of a student falls only when central costs do not increase. When costs are variable, profitability won't grow as the company grows in size. Scale is a relative position to others in a market or geographic segment. It's possible to grow a business but become less scalable. A business that dominated one city or region and then expands nationally may actually soon find they are operating at a scale disadvantage compared to incumbents who are already present in those national markets. In education like the work that we do, the majority of the costs are locally based. Thus, we will compete most effectively and generate outsized operating margins and return on invested capital if we stress a locally driven expansion strategy, rather than a national or global one. It's my belief, therefore, that education businesses (like LL's that require face-to-face or personal and synchronous service delivery) are locally scalable, but not globally scalable.
This examination into the scalability of education businesses has left me with so many more questions that I want to dig into more in 2020:
- Which parts of the LL experience can we shift from variable costs to fixed costs? Any assessments? Any constructive alignment work? Setting Learning Goals? Certain Learning Activities? Declarative or Procedural Knowledge? Peer interaction? Student-coach interaction? More of student-content interaction?
- How can we potentially leverage asynchronous communication and feedback in group classes? How much more valuable is feedback if received immediately versus within 24 or 48 hours? Within the week?
- What roles can peer networks play and what elements of the learning experience could be best driven by peers as opposed to professional coaches?
Thanks for keeping along these last three weeks and I hope it's given you a bit of an insight into why we aren't opening LL centers all over the country and the world in 2020!