Stocks I own: Coursera (COUR)
Best-in-class public EdTech. Reliable, moderate grower (and *true* AI company, as I wrote in 2001) has turned the profitability corner. Key negatives: ESO dilution, NRR drop & hypercompetitive sector.
My Coursera (COUR) investment is deeply underwater. I started accumulating1 in 2021 and—unlike some of my other tech holdings—I never trimmed during the interest rate increases2 `but instead added to the position because each quarter fundamentally did support my long-term, positive thesis. It’s a good investing self-lesson about timing and the importance of a valuation discipline. I’ve never genuinely doubted my conviction with respect to Coursera’s long term potential. But it hurts to overpay on multiple entry points. In any case, if I had waited rather than buying way too early, today (after the Q2 2024 print) would have been the perfect time to add a new position in COUR3. Why? Because its reliable growth profile has turned the corner on profitability and—from the market’s perspective—Coursera is likely to finally shed the pessimistic, associative baggage of EdTech sector participation and generative AI threats.
EdTech has been treacherous terrain for investors. Last week, 2U filed for Chapter 11 bankruptcy. 2U and Coursera are like night and day. Unlike Coursera which is an effective, flywheeling platform that connects learners, schools and organizations; 2U only aspires to be a platform. Unlike Coursera which is well managed, 2U self-inflicted multiple crises on itself including a balance sheet crisis. In contrast, Coursera has a strong balance sheet. In the last quarter, Coursera’s free cash flow added +$17.0 million to a cash balance of $709.0 million, while repurchasing ~$31.0 million in common stock. Coursera has a market capitalization of only $1.67 billion, but no debt and over $700.0 million cash on hand; that is, the current stock price of $10.72 embeds fully $4.72 net cash per share!
As usual, I don’t want to use this substack to riddle the reader with too much detail. I’m evolving as an investor who now understands that too many participants miss the forest for the trees. I’ve made some of my biggest mistakes by paying too much attention to the wrong details; e.g., I once sold out of a fine semiconductor stock by listening to an expert who just knew too many details.
The truly sublime talent is to discern the few things that really matter. If you want lots of numbers, here is Stock Rover’s COUR Research Report4. In 2021, I first explained why Coursera is an AI company (and why the skills graph is its hidden asset). Later in 2021, I said COUR was suffering undue guilt by association (with Chegg and other EdTech). In 2022, I said the thesis was unchanged but employee stock option (ESO) dilution had become a yellow flag.
The few things that matter: moderate but reliable growth; turned the corner on profitability; generative AI is a positive catalyst; and edtech product leadership
Moderate but reliable growth. Coursera has three segments: Consumer (~57% of revenue), Enterprise (~34%) and Degrees (~9%). Overall growth has slowed to +11% which (YoY) represents a blend of +12% in Consumers, +8% in Enterprise, and +14% in Degrees where the gross margin is 100%. Growth is entirely organic5. Among all three segments, the salient negative is a drop in net retention rate (NRR) for paid Enterprise customers. When I bought my first tranche four years ago, my dashboard highlighted their NRR of 113%. It dropped to 108% in 2022, then to 98% in 2023; and down to 93% in the last quarter. The CEO implies NRR has stabilized: “we continue to work through some government partnerships with more transitory budgets … the regions where we've seen Coursera for Business [note: Enterprise = Business + Campus + Government] hit the most has been in Europe and, to some degree, in North America. And that's kind of where we're seeing some signs of stabilization. This is the case across many different variables: it's retention of contracts, it's NRR, it's bookings that are all looking a bit healthier.”
Turned the corner on profitability. From negatives in 2023, adjusted EBITDA in the last three quarters has switched to positive (3.4% in Q4 2023, 4.9% in Q1 2024, and 6.1% in latest Q2 2024).
Generative AI is actually a long-term positive catalyst not a negative: My 2021 thesis explained why Coursera is an AI beneficiary. Markets don’t seem to understand this yet. I’m patient. During the last Q2 quarter Coursera “surpassed more the 2 million enrollments in [their] generative AI catalog of courses, credentials, and hands-on projects”. In the tradition of Coursera’s AI roots, their CEO (Jeff Maggioncalda) is a legit early adopter who mobilized the company’s Project Genesis to capitalize on generative AI6. One of the outcomes is Coursera Coach, who I noticed yesterday while I was attending Advanced Prompt Engineering for Everyone by Dr. Jules White.
On the earnings call, Jeff highlighted Coach:
“Earlier this month, we were pleased to announce the general availability of Coach for our paid consumer learners and all Enterprise Customers. We also unveiled an updated visual identity for Coach, along with a redesigned, more integrated user experience. As a reminder, Coursera Coach is our interactive AI-powered guide that is tailored to learners' unique goals and anchored in the expert content on Coursera. To date, Coach has been used by more than 700,000 learners and concurrently respond in 21 languages, supported by the underlying large language models. Today, Coach enables learners to ask questions to clarify material and stay on track, summarize key takeaways for better note-taking, practice for quizzes and tests to solidify knowledge and gaps, and explore how they're learning aligns with current or future career goals. In the future, we expect Coach will play an increasingly prominent role throughout Coursera's platform, from personalized learning and discovery to career counseling and guidance.” — Jeffrey Maggioncalda, Q2 2024 Earnings Call
Product innovation and leadership continues: You have to peer beyond the pandemic surge/retreat and observe the broader disruption happening in EdTech. While a premium degree at a prestigious brand like Stanford will always meet demand, the annual retail cost (including room and board) to attend an elite US college exceeds $80,0007. Further, in the traditional, broken model, exclusivity is reinforced by an arduous admissions process. There are big markets at lower price points, especially if the angle is job-friendly skill acquisition. If lower price points are complemented with convenient admissions and via solutions to remote delivery, the markets are massive and global. Coursera is meeting this with the so-called Pathway Degree: an “open enrollment model” enabled by credits that flexibly stack online and/or offline courses.
An important signal of their product leadership is quality and growth of their Professional Certificates, where (during the Q2 quarter) they “added 15 new certificates from leading technology brands like Google Cloud, IBM Meta and Microsoft … We now offer more than 60 entry-level Professional Certificates, with a strong pipeline of additional titles coming later this year.”8
Employee stock option (ESO) dilution is a flag
Amazingly, over the last five earnings calls, no analyst has asked about ESO dilution. I am dumbfounded by Wall Street’s apparent lack of attention to this. I wonder if investor relations read my analysis because on the last call, the CEO offered (emphasis mine):
“But we also believe it is important to address future dilution concerns of our investors. So management has recommended, with the Compensation Committee approval, the introduction of a price floor of $10 per share for all future employee grants, effectively eliminating the number of shares issued for dollar-denominated stock-based compensation.” — Kenneth R. Han, CFO, Q2 2024 Earnings Call
I am slightly confused by his language of his last clause (I assume he means that the floor eliminated ESO grants while the stock languished below $10 since May, and now has again breached today!), and the DEF 14A Proxy does not appear to reference this ESO price floor directly. However, the Proxy does explain a shift to performance-based restricted stock units, PSUs, associated with an investor outreach following a bleak 52% approval rate on their Say-on-Pay vote.
Nevertheless, the yellow flag of ESO dilution is flying at full staff with a reddish tint: Coursera’s stock-based compensation (SBC) expense was $110.8 million in FY (CY) 2022 and $109.6 million in FY 2023. I still do not like this expense line but no investment is perfect.
To summarize
Now that Coursera has quite turned the corner on reported and cash flow profitability with a very strong balance sheet, its long-term stock price prospects are double (really, triple, I think) current levels. Yes, public EdTech is a sour sector, but this is the best-in-class public trade in that sector. At the same time, I want to be responsible by itemizing my view of the negatives:
It’s a very competitive space
The ESO dilution is bordering on imprudent (at least they mitigated with repurchases, necessary given their bulging cash account)
The drop in Enterprise net retention rate (NRR) is a valid concern.
Thanks for reading. This is not investment advice, as you know.
I’ve purchased COUR in 10 tranches. My approach to buying is typically the following: I first add ticker to my watch list (I try never to impulsively buy). When I decide to buy, I have a typical starter size (initial tranche) for an initial purchase. Even if I’m really enthusiastic, I use the same initial size and add over time. Then I re-evaluate after each quarterly report.
The big lesson (to me) was taught around Sept/Oct/Nov 2021 when we were debating inflation in my investing group (at the time). It’s maybe hard to recall, but many professionals at the time believed the Fed’s (and Treasury’s) claim of “transitory” or “transient” inflation. At the time, I noticed that many in my group basically had no life memory, and therefore no expectation, for a high rate regime. Many of us in the group held significant positions in high-growth but unprofitable technology companies (not to mention some SPACs). Leaders in that group used DCF models (with growth assumption) to value these companies; those valuations were worse than useless. The big lesson, for me, was a reminder that an unprofitable company has no immediate intrinsic value: it might be worth zero. You have to make an assumption it will grow into profitability. I think many investors are secretly technical investors in this sense: if a big-revenue company’s stock price is down, they sort of assume it has some natural floor and should revert. But the price floor of an unprofitable company is zero.
After reporting Friday, COUR bounced ~ +50% from $7.41 to $11.44 but it’s down to $9.72. Despite the jump, I still believe it will eventually double past $20.00. The market cap is only $1.7 billion.
Stock Rover’s COUR Research Report. Shared with permission. I love Stock Rover, it’s my favorite stock research tool and it dynamically links to my Merrill Edge portfolios.
Coursera made on acquisition in 2019, and a private investment in 2023. See Dhawal’s excellent analysis of the 10K at https://www.classcentral.com/report/coursera-annual-report-2023/
Coursera’s Foray into GenAI, Harvard Business School Case Study, April 16, 2024. See https://store.hbr.org/product/coursera-s-foray-into-genai/124089
Perplexity: What is the annual cost of an elite college education in the United States? https://www.perplexity.ai/search/what-is-the-annual-cost-of-an-7M5DxzS3Rw2TRqVS8hPxqw
Jeffrey Maggioncalda, Q2 2024 Earnings Call