Table of Contents

• Introduction
• Methodology
• Can Bootcamps Scale Effectively?
  • Figure 1: Estimated Bootcamp Enrollment, December 11 2019
• The Unit Economics of Short Term Education
  • Figure 2: Back of the Envelope Unit Economics Calculations
• Bootcamps and the Skills Gap
• Challenges Faced by Scaling Bootcamps
• Coding Bootcamps and Student Outcomes
• Do ISAs Change the Structural Incentives of Running a Bootcamp?
• Conclusion: Investment Opportunities in Short Term Education

Introduction

With all the hype bootcamps received in 2019, it’s important to remember that as recently as 2017 the bootcamp industry was in dire straits. Dev Bootcamp, one of the first bootcamps and a high profile acquisition of the Kaplan Group, had shut down.^[1] Iron Yard followed Dev Bootcamp’s lead in less than a few weeks,^[2] and industry observers predicted that the industry would see consolidation.^{[3] [4]} Rick O’Donnell, the CEO of Skills Fund, a student loan provider for bootcamps and other vocational training programs, told Inside Higher Ed that Skills Fund had “been forecasting a consolidation wave within the boot-camp industry; there are simply too many schools for everyone to be profitable,”^[5] and this sentiment was echoed in other articles about the industry.^[6] Following that prediction, the bootcamp industry saw additional closures (e.g. Learners Guild^[7]) and acquisitions (e.g. Flatiron, General Assembly, Bloc, Hack Reactor, Fullstack Academy, and Thinkful).

However, despite the consolidation of older bootcamps, new bootcamps such as Lambda School and Holberton School are gaining prominence while attracting significant press and policy maker attention. Sen. Chris Murphy named Holberton “innovator of the month”^[8] and Sen. Todd Young toured the campus of Kenzie Academy.^[9] The New York Times, Business Insider, and Bloomberg have all given Lambda School favorable coverage,^{[10] [11] [12]} although Business Insider has subsequently published critical pieces about Lambda School.^[13] It’s striking how quickly the narrative has shifted from one of austerity to one of growth, and the narrative shift raises questions about what, if anything, has changed about the industry since 2017.

Methodology

This report relies on extensive literature reviews, data collected from LinkedIn, and interviews with bootcamp founders, employees, and students. Three bootcamp founders were interviewed, as well as five bootcamp employees, two of whom were c-level executives.

Can Bootcamps Scale Effectively?

Coding bootcamps make up a very small portion of the for-profit education industry. Industry reported figures claim that 15,429 students attended coding bootcamps during 2018, and that 16,190 attended during 2017.^[14] This number is dwarfed by the estimated 842,000 students who enrolled at for-profits in 2017.^[15] Coding bootcamps might be the future of higher education but for now they are a very niche product.

Estimated enrollment data from individual bootcamps (figure 1) show that it’s very difficult for bootcamps to scale effectively. Although many bootcamps don’t publish enrollment data, it’s possible to estimate enrollment by using the alumni search tool available on most bootcamps’ LinkedIn pages.^[16] The search tool was used on December 11th, 2019 to obtain the estimates in figure 1. These estimates are not perfect—not every student has a LinkedIn profile and the figures for 2019 are incomplete—but they are a way of assessing the relative growth and size of each bootcamp. The data shows that most bootcamps have difficulties enrolling more than a thousand students (Lambda School, General Assembly, and Flatiron School are the only bootcamps tracked here to have done so). The data also suggests that year to year fluctuations in enrollment are common.

Figure 1: Estimated Bootcamp Enrollment, December 11 2019.

Note: LinkedIn data was collected on December 11th, 2019. Galvanize was acquired in 2020 after the data was initially collected.

Even for bootcamps that are scaling at a rapid rate, there are indications that growth has required significant organizational investment. Lambda School has expanded to Africa,^[17] the EU,^[18] and the UK,^[19] and is reported to be planning an expansion to India.^[20] One of Lambda School’s selling points is its Income Share Agreement (ISA), and setting aside any other difficulties a school serving multiple countries would have to navigate, the challenge of adapting their ISA to multiple regulatory frameworks makes expanding to multiple countries a significant investment. Lambda School is not the only bootcamp expanding internationally: Holberton School has a presence in the US, Tunisia, Colombia, and Lebanon,^[21] while Kenzie Academy announced an expansion to Brazil.^[22] International expansion suggests that bootcamps feel they need to look abroad to satisfy their goals for growing the business.

The Unit Economics of Short Term Education

An important question to answer is whether bootcamps are actually profitable businesses or if they’re propped up by venture capital funding. This report estimates the revenue, expenses, and profit of a hypothetical bootcamp and finds that it’s possible to run a profitable bootcamp that provides a reasonable quality of education. While the unit economics model described here is no substitute for looking at an actual balance sheet, it does provide a general overview of the business pressures bootcamps face.

According to an independent survey of the bootcamp industry by RTI, a typical bootcamp has a median tuition of $11,900 and an average length of 16.5 weeks, although this depends on a wide variety of factors, such as whether the bootcamp is full time or part time.^[23] The model bootcamp budget therefore assumes revenue is $11,900 per student and that each student attends for four months. Bootcamp tuition is significantly more expensive than a year’s worth of 2018 in-state tuition for a 2-year college in the US ($3,243),^[24] which makes sense as bootcamp tuition is not subsidized. Costs can be broken into three categories: customer acquisition costs, instructional costs, and curriculum development costs.

One of the largest drivers of costs is the need to recruit students. On the very low end, affiliate programs such as Career Karma charge a $1,000 referral fee for each student recruited on its platform.^[25] However, as programs scale, it’s likely that they will need to turn to larger platforms such as Facebook and Google, which are more expensive. For example, Lambda School is one of the fastest growing bootcamps and it uses Facebook^[26] ads. This makes Lambda School similar to other large for-profits, which spend heavily on Facebook and Google Ads.^[27] For 2U, an OPM company that runs online programs for accredited universities, marketing and sales make up 19.2% of revenue.^[28] Academic studies of for-profits estimate the customer acquisition costs at $4,000 per student,^[29] and it’s reasonable to assume these costs are similar for bootcamps, so the $4,000 figure is used as an estimate in this report. At the moment, it’s difficult to see a path to reducing these costs. These costs are likely the same for high-quality and low-quality bootcamps (prospective students are uninformed consumers) and these costs likely increase as bootcamps scale and compete for the same students.

The second cost driver is instructional costs. Absent tools like AI and chatbots, bootcamps must employ instructors and budget for their salaries. The way many bootcamps structure their program is by using a combination of experienced instructors and teaching assistants. For example, the model schedule posted on Lambda School’s website advertises a guided project with an instructor on Mondays through Thursdays, as well as daily meetings with a Team Lead.^[30] In Lambda School’s case, Team Leads are Lambda students that take several weeks off from their studies to assist other students.^[31] While the role of Team Lead doesn’t necessarily have to be filled by current students, the example of Lambda School suggests that a good model for delivering instruction is to have a mix of experienced instructors and inexperienced teaching assistants. While the use of inexperienced instructors has received criticism, it can be beneficial to students: there are many repetitive tasks that experienced instructors don’t have time for but which inexperienced instructors can do at an adequate level.

The unit economics model assumes an optimal ratio of 1 experienced instructor per 20 students, and 2 inexperienced instructors per 20 students. Truly experienced instructors are expensive and difficult to recruit because bootcamps will need to compete with tech companies to employ them. Teaching assistants are inexpensive, can be recruited from current and former students, and can do everything from career prep to grading to individualized tutoring. It’s reasonable to assume that experienced instructors come with a salary of $100,000, and inexperienced instructors come with a salary of $40,000.

Separate from instructional cost is the cost of developing and updating curriculum. It would be naive to believe that the curriculum of every bootcamp is bespoke and original, given the number of former bootcamp students and employees who go on to work for or start competitors. There are good reasons to avoid developing curriculum, as developing curriculum is expensive. The best comparison point for curriculum development costs are MOOCs (Massively Open Online Courses); practitioners reported that creating an hour of a video lecture involves 3-10 hours of preparation, depending on whether the curriculum was adopted from prior courses or created from scratch.^[32] The unit economics model assumes that developing one hour of curriculum takes 10 hours of preparation in the first year of running a bootcamp, and that updating an hour of curriculum takes 3 hours every subsequent year. Lambda’s model curriculum advertises 7 hours of instruction on Mondays through Thursdays, which comes out to 28 hours of instruction per week.^[33] Since the average bootcamp is 16.5 weeks long,^[34] there are 463 hours of instruction that need to be developed, presumably by experienced instructors with salaries of $100,000 (i.e., $50 per hour). This comes out to $231,500 to initially develop the curriculum and $69,450 to update the curriculum each year after the first year. This doesn’t take into account video production costs or the costs of creating interactive exercises, but it’s not clear if these are necessary features of a good curriculum or merely optional.

Finally, there are administrative costs, which include office space and legal fees. Some of these costs can be eliminated by switching from in person instruction to online instruction (unlike office space, tools like Slack are cheap if not free). In many jurisdictions there are fees associated with running a bootcamp; in California, bootcamps and other nonaccredited schools are expected to register with the Bureau of Private Postsecondary Education (BPPE). However, these fees are not relevant to all bootcamps as some bootcamps like Lambda School neglect to work with the relevant authorities.^[35] This report doesn’t explicitly model these costs but it’s important to be aware that they exist.

The unit economics model assumes three cohorts of twenty students each year. The model then calculates the per-student costs, and subtracts the per-student costs from the per-student revenue of $11,900 (figure 2). The model suggests that it’s well within the realm of possibility to operate a small bootcamp of reasonable quality and be profitable (see figure 2). Some bootcamps will save money by forgoing some of the expenses the model deems necessary; for example Holberton School does not have teachers which presumably increases their profits.^[36] Other bootcamps will have additional expenses that the model doesn’t view as being essential to running a bootcamp; for example Flatiron School developed a custom online learning platform.^[37] Overall, it’s likely that the unit economics calculations overestimate the profit a good bootcamp can earn in practice. That said, it’s encouraging to know that running a profitable short term education program is within the realm of possibility.

Figure 2: Back of the Envelope Unit Economics Calculations

Bootcamps and the Skills Gap

Bootcamps are quick to cite skills gap statistics in support of their business models. The latest estimates for the number of unfilled technology jobs put the number at around 918,000.^[38] However, just because there is a skills gap of 918,000 jobs does not mean that there are 918,000 people who are a short bootcamp away from filling those jobs. Right now, the United States is experiencing low unemployment and a tight labor market (in August 2019 there were 0.9 unemployed workers for every job opening^[39]), which means that employers’ hiring difficulties can be attributed to the fact that the workers they are trying to hire already have jobs. Bootcamps will have difficulties recruiting students for the same reasons employers have difficulties recruiting labor; people are already employed and thus have less of a need to reskill for new jobs. Enrollment in for-profit schools has declined by 51% since 2010 (from 1.7 million in 2010 to 842,000 students in 2017).^[40] This decline mirrors the post-2010 recovery, and industry observers attribute declines in enrollment at for-profits like the University of Phoenix to low unemployment.^[41] If bootcamps sold training to employers, it would make sense to use the skills gap to estimate the size of the bootcamp market. But fundamentally, most bootcamps sell training to students, and thus the skills gap is in many ways the opposite of the model needed to understand the bootcamp industry.

There’s evidence that employers are offering less training than they did historically, which presents a business opportunity to sell education to employers rather than students. Employer-paid training declined by 28% between 2001 and 2009.^[42] America’s apprenticeship system is severely underdeveloped compared to countries like Germany.^[43] While 238,000 Americans began apprenticeships in 2018, these apprenticeships were concentrated in the construction industry,^[44] and Germany (a smaller country) outperformed America with 520,332 apprenticeships in 2016.^[45] There is significant room for employers to commit resources to addressing the skills gap, and there are potential business opportunities for education startups to sell training programs to companies struggling to fill high-demand positions.

Challenges Faced by Scaling Bootcamps

The data shows that most bootcamps struggle to break a thousand students (figure 1). There are a number of reasons for this. First and foremost, the number of potential students who are not served by existing institutions is most likely significantly smaller than skills gap statistics suggest (see previous section). This means that bootcamps will need to compete for students with both other bootcamps and with private, public, and for-profit institutions. As startups, bootcamps are at an immediate disadvantage as they do not have the brand recognition of decades old universities or the marketing sophistication of large for-profits. This is why many industry observers describe bootcamps competing for a small number of students, which led to consolidation in 2017.^{[46] [47]}

When competing with other schools for limited students, bootcamps with high quality education are in many ways at a disadvantage. Students are not rational, fully informed consumers, and it’s not clear that prospective students have the information needed to identify high quality bootcamps. Thus, it’s not clear that investments in education lead to a decrease in marketing costs. From a margins perspective, low quality bootcamps can afford to put their resources into marketing, while high quality bootcamps will struggle to balance marketing and instructional costs.

In addition, structural forces make it difficult for bootcamps to maintain educational quality while scaling. A key method of ensuring quality instruction is to maintain a ratio of instructors to students. Bootcamps teach programming, so instructors who are truly qualified have the option to work for other tech companies, which means it’s expensive to employ good instructors and difficult to retain them. Rapid increases in enrollment present challenges in recruiting enough qualified instructors to maintain a specific instructor-student ratio. However, enrollment declines also present risks. Bootcamps will either have to fire instructors, which is a difficult pill to swallow given the effort spent recruiting them, or accept rising per student instruction costs (if the number of students halfs but the number of instructors stays constant, the cost per student of instruction doubles). The enrollment of many bootcamps fluctuates on a year-to-year basis (figure 1), which increases the difficulty of maintaining a specific instructor-student ratio.

Admissions policies are a tool that many bootcamps use to ensure the quality of their education, and admissions policies become difficult to implement as bootcamps scale. Acceptance rates are considered by some observers to be the best measure of the quality of a bootcamp,^[48] and there are reasons why good bootcamps will have stricter admissions policies. First, having a robust admissions policy helps maintain a stable number of students, which is essential to maintaining a specific instructor-student ratio. Bootcamps that accept a fraction of students who apply can increase their acceptance rate in off-years when applications are down, thereby ensuring a stable number of students every year. Second, to ensure that graduates are prepared at the end of the bootcamp to secure jobs, bootcamps need to ensure at the start of the program that their students have the necessary educational foundation to succeed. Ultimately, it becomes harder and harder for bootcamps to maintain admissions standards as they scale. First, the easiest and cheapest way to reduce customer acquisition costs is to stop turning away students. Second, relaxing admissions standards increases the size of the potential market a bootcamp can serve. It takes extraordinarily discipline to maintain admissions standards while growing a bootcamp.

It’s concerning that some bootcamps appear to be increasing the expenditure per student while rapidly scaling. For example, Lambda School increased the length of its program from 7 months (30 weeks) in 2018^[49] to 9 months in 2019,^[50] and added additional services such as mental health counseling.^[51] Increasing the course length while dramatically scaling the size of the bootcamp is a sign that a bootcamp is not getting the results with a larger body of students that it got with a smaller body of students. It’s possible that Lambda has not been able to maintain its instructor-student ratio or its admissions standards as it scales.

Coding Bootcamps and Student Outcomes

Since bootcamps are significantly more expensive than community colleges (compare average bootcamp tuition of $11,900 for a four month course^[52] to the $3,243 price tag of a year’s worth of 2018 in-state tuition at a 2-year university in the US^[53]), it’s reasonable to ask whether bootcamps consistently place their students in relevant jobs. However, most bootcamps are not transparent about their outcomes data, making it difficult for prospective students to assess their effectiveness. While in some cases there have been legal consequences for this, such as a $375k fine brought against Flatiron School by the New York Attorney General,^[54] overall there doesn’t appear to be consequences for publishing misleading outcomes data. It’s unsurprising that the bootcamp world is awash with outcomes reports that haven’t been independently verified, such as Holberton School’s claim that 100% of April graduates find “employment opportunities” in three months^[55] or Lambda School’s claim that “every single Lambda School graduate who has been on the job market for six months is either employed in a full-time role as a software engineer or has joined an early startup working for equity.”^[56] There is a clear need for accurate, verifiable data.

Some in the bootcamp industry attempted to clean up misleading outcomes reports, which resulted in the creation of the Council on Integrity in Results Reporting (CIRR), a self-regulatory body in the industry. Sheree Speakman, CIRR CEO, said in a statement that many bootcamps “report outcomes to both their state regulatory agencies and to CIRR,” and that while bootcamps aren’t required to submit data to CIRR, many voluntarily submit data “to increase transparency and accessibility for prospective students who review outcomes in order to select a school and program.” That said, CIRR is best understood as a failure to self-regulate. While the CIRR standards are themselves reasonably robust, it’s questionable whether these standards are actually followed by CIRR members. For instance, CIRR standards mandate that bootcamps submit data twice a year and that these reports must be audited by a third party,^[57] but in practice many bootcamps only submit a report for a single cohort and then never return to submit subsequent reports or audit the submitted report. In fact, as of September 13, 2019, of the 25 bootcamps that have submitted outcomes reports to CIRR, 10 of those bootcamps submitted reports for only one reporting period.^[58] Bootcamps’ ability to cherry pick outcomes reports calls into question the effectiveness of CIRR’s reporting.

The official data that is available indicates that leading bootcamps have poor outcomes. General Assembly, for instance, is obligated by California law to submit outcomes data to BPPE, which indicates that General Assembly fails to place a sizable portion of their students. For instance, for the 2017 Web Development Immersive, only 165 graduates out of an initial 310 students enrolled found full-time employment, meaning only 53% had successful outcomes.^[59] It’s a similar story for the 2017 User Experience Design Immersive, where only 192 of 291 students enrolled found full-time employment, which is a 66% success rate.^[60] BPPE data is thus highly relevant to prospective students and should be available for all bootcamps. Unfortunately, some bootcamps such as Lambda School operate without BPPE approval (as a result, BPPE subsequently fined Lambda School and ordered it to cease operations^[61]).

Do ISAs Change the Structural Incentives of Running a Bootcamp?

Poor employment outcomes help explain why bootcamps are so quick to adopt Income Share Agreements (ISAs). An ISA is a financial product where a borrower gives up a percentage of their future (typically pre-tax^[62]) income in exchange for upfront funding. ISAs are frequently portrayed as a way of “eliminating risk” for students,^[63] and it’s true that if only 53% of students get jobs then there is substantial risk in attending a program like the 2017 Web Development Immersive course at General Assembly (General Assembly, like many other bootcamps, recently announced an Income Share Agreement^[64]). However, the notion of eliminating risk raises questions about the quality of education offered by many bootcamps as well as the appropriateness of ISAs as a financial product. Simply put, if the quality of education was better, then there would be less of a need to de-risk it.

The notion of de-risking education is at odds with another claim about ISAs: that they “align incentives” between students and school.^[65] What bootcamps which make this claim are often not transparent about is that ISAs are financed, which means that the ISA provider will sell the contract to third party investors.^[66] The lack of transparency can be very upsetting to students. For example, Pursuit (a coding bootcamp with an ISA) used to call their ISA program the “Pay It Forward commitment,”^[67] which was upsetting to some students^[68] because their ISA payments actually went to investors,^[69] not to future students as was implied. Another example is Lambda School, which used to claim on its website that they “only get paid when you do,”^[70] which would contradict Wired reporting on how Lambda School’s ISA is financed.^[71] Truly aligned incentives as advertised would mean keeping the ISAs on the books, so that bootcamps only earn money when their graduates earn money and so that the amount of money bootcamps earn is directly proportional to the amount of money graduates earn. It’s possible that financed ISAs have some sort of financial incentive for bootcamps with good outcomes; such incentives are not unique to ISAs and are offered by some private loans, like Climb Credit, that are available to bootcamp students.^[72] However, students deserve to know the details of how exactly their ISA financing works and how exactly the incentives are aligned between student and school.

The biggest problem with ISAs from a business perspective would have to be their uncertain legal status. The main legal debate over ISAs is whether ISAs are a new type of financial product that requires a new legal framework, or are they a new type of loan, in which case they would be subject to the Equal Credit Opportunity Act, the Truth in Lending act, and state usury laws.^{[73] [74]} These legal debates are often framed in terms of concern about potential harm to consumers, which makes sense as these laws are designed to protect consumers from abuse. However, a difficult question to answer is what would happen if ISA borrowers decided to simply refuse to pay the monthly payment? If ISAs were loans, then there would be an established legal framework that would provide lenders with some amount of recourse. But since ISAs claim to not be loans (Lambda School’s template ISA claims in all caps that “this agreement does not constitute a loan”^[75]) it’s not clear what recourse is available. Lambda School’s template ISA contract mentions potentially garnishing wages (“including any rights available to Company to garnish wages”),^[76] but it’s not clear if this is something Lambda School actually has the ability to do.

Another challenge ISAs present to both students and bootcamps is it appears that in many cases bootcamps earn significantly less from ISAs than they would if students paid upfront. According to a Wired profile of Lambda School, Lambda School sells their ISAs at a deep discount relative to tuition.^[77] The financing of ISAs has the potential to be a money losing scheme for bootcamps, which raises questions about whether it makes financial sense for bootcamps to offer ISAs. The low value for which ISAs are sold also raises consumer protection questions about whether ISAs enable price gouging. Students who pay full upfront tuition have a legitimate argument that they are overpaying relative to students with ISAs. And students with ISAs may very well wonder how much of their payments are directly going to their bootcamp, and how much of their payments go to the middlemen financing the ISA.

Lambda School appears to be responding to these challenges by developing alternative mechanisms to collect ISA payments. Lambda School’s template ISA contract requires borrowers to maintain an “approved bank account” that the ISA servicer has access to,^[78] and could potentially require students to set up this bank account prior to beginning the course. These tactics place severe restrictions on students’ financial autonomy and should be concerning to any consumer protection advocate. Lambda School has also experimented with allowing employers to hire Lambda School students through Lambda School with a program called Lambda Studios.^[79] While Lambda Studios appears to be discontinued—the relevant page on Lambda School’s website no longer exists—such employment relationships have the potential to allow bootcamps to take money from students’ paychecks directly, or to fire students who miss ISA payments. Indeed, absent a new legal framework for ISAs, employing students directly might be the only way for ISA providers to guarantee that students will make payments.

In general, ISAs raise significant consumer protection and affordability issues. Analysis of ISAs offered by accredited universities has found that students pay significantly more with an ISA than they would pay with a conventional student loan.^{[80] [81]} ISAs also create significant potential for discrimination: income is correlated with demographic factors like race and gender, which gives ISA providers an incentive to offer more expensive ISA terms to marginalized groups, and as a result there needs to be oversight and protections to ensure that discrimination does not occur.^[82] Finally, the consequences laid out in Lambda School’s template ISA contract for missed payments have the potential to be punitive, especially given that the most likely reason for missed payments would be not having the money on hand. According to Lambda School’s template ISA, students who miss payments are responsible for paying the entire $30,000 payment cap.^[83] Given that the payment cap is ostensibly for the purpose of protecting high earners from paying too much, it isn’t appropriate for the payment cap to also be the amount students with missed payments are responsible for. ISAs deserve scrutiny from consumer protection groups, which would be beneficial for both the students themselves and also for schools who want confidence in the appropriateness of ISAs for their students.

Conclusion: Investment Opportunities in Short Term Education

Acquisitions of bootcamps are sometimes portrayed as evidence that bootcamps are valuable investment opportunities. The reality suggests that bootcamps are overvalued by investors and buyers alike. The recent example of K12’s acquisition of Galvanize is instructive in this regard. Prior to the acquisition, Galvanize raised $167.5M from investors.^[84] Galvanize was then bought for $165M, meaning that investors would be lucky if they broke even. The acquisition was not a success for K12 either. Prior to announcing the acquisition of Galvanize during the January 27, 2019 earnings call,^[85] K12’s stock traded at $19.38. One day later, the stock declined by 15% to $16.62, making the acquisition a failed gambit to boost K12’s stock, which was declining prior to the Galvanize acquisition. It’s hard to avoid the conclusion that Galvanize was overvalued by investors, who put too much cash into the business relative to what they earned in the acquisition.

The structural obstacles to scaling a bootcamp identified by this report will make it difficult for any bootcamp to achieve venture scale returns. This does not mean that short-term education is not a worthwhile endeavor or a viable business idea. Affordable alternatives to four year degrees are good and needed, and it’s theoretically possible for bootcamps to turn a profit while providing high quality education. However, all the available information suggests it’s difficult for bootcamps to be both profitable and to provide high quality education while growing at a rapid pace. It’s difficult to imagine a scenario where a coding bootcamp is a good venture capital investment.

 

---

February 13, 2020: Report updated to include a statement from CIRR CEO Sheree Speakman.

---