Issue #4: Tech bootcamps
A look at how bootcamps work, how they're evolving, and where they go from here.
Tech bootcamps turned 10 last year. The idea of training a junior software developer in a matter of weeks was still groundbreaking when this niche post appeared on Hacker News in 2011. Fast forward a decade, and it's fair to say the idea has taken hold. Today it's a $418 million industry projected to be worth over $1 billion by 2026.
I’ve been a part of this industry for the same amount of time. Which I guess makes me a veteran. I recently turned 35, which qualifies me to play in the “veterans” 5-a-side league near where I live in South London. There’s a nice symmetry to being a veteran in both my professional and amateur lives.
Thanks for reading The View! Subscribe for free to receive new posts and support my work.
Anyway, as I was saying, I joined General Assembly in 2012 and worked on the first bootcamps, initially as in-person classes and then entirely online. Today, I’m going to summarize a few of the trends I’ve seen and offer a view of where this industry is moving.
The last ten years have seen three clear phases for bootcamps, which I’ll characterize with the pithy titles: early, middle and late.
This early period started with the launch of Dev Bootcamp in 2012 and lasted for the next 24 months. Other bootcamps soon followed, including General Assembly, Flatiron School and Hack Reactor, as the next two years saw an explosion in new program launches. In 2013, 2000 students graduated from coding bootcamps; by 2014, that number had tripled to 6000.
It was during this early period that some of the standard features of the bootcamp space were solidified. These are still the defining features today:
Learning takes place in an immersive, practical environment.
4—6 weeks of pre-work before attending a course.
A median tuition of $11,900 and an average length of 16.5 weeks.
A typical student being a 31-year-old (usually male) with a bachelor's degree and looking to make a career change.
From 2012—2014, many alumni of the early bootcamps left to start their own bootcamps. Flatiron School, App Academy and Hack Reactor all emerged from graduates of the early bootcamps.
The barrier to entry for starting a coding bootcamp wasn't very high. All you needed was an instructor, classroom space and 25 willing students. For example, when I worked on General Assembly's coding bootcamp, 5 out of the first 6 instructors we hired all left to set up their own bootcamps.
But none of them lasted more than a year. It turns out that education, at any level, is not the easy-money scheme many thought it was.
Many early bootcamps struggled with high acquisition costs. Class Central estimates that the cost to acquire each new customer for a coding bootcamp was $4000. The transition from serving early-adopter students to mainstream students who need more structure was difficult. After the initial rush of early adopters had gone through their programs, many bootcamps struggled to find these new students without venture funding.
A lot of the growth we saw in this early period was capitalizing on a short-term market opportunity without a broader vision. Getting into the market was easy. Sustaining growth, course quality and student and employer satisfaction was more challenging. Bootcamps without scale, expertise and established best practices found it challenging to produce excellent results time and again.
As 2014 turned into 2015, we entered the middle period for bootcamp growth, characterized by razor thin profit margins and pressure to find new growth areas.
As we can see from Fig. 1, the margins on each bootcamp student going through a bootcamp aren’t huge: 9% in the first year, 31% in the second.
Figure 1. Estimated unit costs to run a coding bootcamp. Source: Class Central
Firms like General Assembly and Flatiron struggled to sustain 30% YoY growth as the typical student persona shifted from highly motivated early adopters to more mainstream learners. The initial rush of early adopters had given way to a steadier trickle of enrollments.
This middle period is characterized by bootcamps pursuing new and alternative revenue streams. Some built online versions of their courses. Hack Reactor and General Assembly launched remote-first versions of their software engineering bootcamps in 2016. Others, like Flatiron School and Iron Yard pursued national and international expansion.
While others looked to university partners to take advantage of institutional reputation, pedagogical expertise and infrastructure. This contributed to a new wave of bootcamps, led by Trilogy, which contributed to a peak in enrollments again in North America.
Throughout this middle period, we can see evidence of how difficult it is to scale a technical bootcamp business. Only a handful of providers broke a thousand students consistently each year: Flatiron School, Lambda School and General Assembly.
Figure 2. Estimated Bootcamp Enrollment, 2016-20
Emma Rindlisbacher at Class Central identified 3 main reasons bootcamps failed to scale in this period:
They fail to reduce customer acquisition (CAC) costs
Instructor recruitment fails to keep pace with growth
The bar for course admissions is lowered to achieve scale
Why is a thousand students a year a critical benchmark? One reason is scale. Above a thousand students a year, you can centralize the three functions identified above (marketing, recruitment and admissions) and reduce the costs of running each market.
I'll look at this in more detail in a future newsletter. But each of these 3 main reasons can account for why some bootcamps achieved scale and others didn't during this period.
Starting in 2017, we entered the late period for this particular bootcamp model. We started to see consolidation across the industry. In 2017, WeWork acquired Flatiron School, Galvanize acquired Hack Reactor, and Kaplan shut down Dev Bootcamp. In 2018, The Adecco Group acquired General Assembly, and in 2019, 2U bought Trilogy.
The future of bootcamps
From 2019 onwards, the focus for bootcamps has shifted to who pays for education. Many bootcamps have experienced volatility through the COVID-19 pandemic because consumers were their primary payers. Those who were insulated and grew had a significant portion of revenue from employers or the government.
This period has seen the emergence of a new type of bootcamp provider, like Guild, Multiverse and OpenClassrooms. While all three of these companies have actively positioned themselves as not a bootcamp, there still isn’t a good term to describe these types of accelerated learning providers.
There’s an argument they are all “accelerated learning providers”, but you try saying that phrase a hundred times and see how you like it. The term just isn’t as pithy as “bootcamp”. Which is why that name stuck.
Until a new term emerges, I tend to refer to all outcomes-focused, accelerated programs as bootcamps. i.e. is it training that gets someone a job.
This can be problematic because what people tend to refer to as the "bootcamp industry" really involves a massive amount of variability, both in terms of the visions and strategies of different providers.
But whatever we choose to call these companies one thing is true. The growth of coding bootcampsmight have slowed, but the business of bridging the gap between education and employment is exploding. Bridging that gap is one of the most significant challenges facing today's developed world.
What that looks like is still emerging, but the major evolution is in who pays for education.
Bootcamps in the workplace
The significant learning from the last few years is that the digital skills gap is not as clear-cut as initially presented. Over the years, we’ve seen a lot of bootcamp providers use skills gap data to support their business models.
For example, a Wall Street Journal article estimated there are 918,000 unfilled technology jobs. A lot of bootcamp providers have used data like this to suggest that there’s a vast pool of students who are just a short bootcamp away from finding a job.
But, as Emma Rindlisbacher at Class Central identified, "the number of potential students who are not served by existing institutions is significantly smaller than data on the skills gap suggests". With a tight labor market and low unemployment, employers have difficulty hiring because people are already employed and thus have less of a need to reskill for new jobs.
Therefore, Rindlisbacher suggests, the skills gap exists inside organizations. Work-based learning is the natural progression for many of these accelerated learning providers: a way to link classroom skills to actual on-the-job performance.
Look at the 150+ bootcamp providers worldwide, and you'll see that being a B2C consumer brand in 2022 isn't innovative. With curriculum content increasingly viewed as a commodity, and job placement data impossible to trust, the funding mechanism of these programs becomes a key differentiator.
The providers who will thrive are those shifting the burden of paying for training away from consumers and on to employers or governments.
We can see companies like Multiverse, Emeritus, and OpenClassrooms successfully scaling education programs in this environment, where the buyer is the employer looking to commit resources to fill these high-demand positions rather than individuals funding their own education. These providers have merged the cohort based model bootcamps popularized with clear labor market focused outcomes for the learner and employer.
With funding ring-fenced by the government, these organizations have lower customer acquisition costs and benefit from a land-and-expand strategy inside organizations: much different to the high-churn consumer bootcamp market.
Rather than expensively marketing to individual students in order to constantly fill new cohorts, bootcamps in the employer market can fill cohorts with entire teams of employees in accordance with internal corporate learning initiatives.
Once a provider has sold into one large organization, the costs for that organization to find another provider are significant. Whereas for B2C bootcamps, filling a new cohort every 3 months with new students leads to high customer acquisition.
The last three years have seen bootcamps move upstream in the funding they seek: away from individual students and towards employers and the government.
As I wrote about in a previous issue, bootcamps who are able to build their model around sources of government funding will have a significant advantage. The pain point for companies looking to upskill or reskill their employees is in accessing these different funding routes. Bootcamps are well served to solve this problem.
One thing I’m concerned about is what this means for jobs data. The good thing about the bootcamp era was it forced companies to be transparent about job outcomes. For all the controversy around the accuracy of the numbers, at least there was data available. As funding is moved from consumers to enterprise and government funding, it’s likely we’ll lose access to this data. Is there a benefit in organizations being transparent about whether their own employees find new jobs? Probably not.
Alternative & Independent: the universe of technology-related “bootcamps”, RTI Press, 2019.
Understanding the EdTech Content Landscape, Brighteye Ventures, 2022.
As General Assembly investor and Maveron partner Jason Stoffer wrote, “it turns out it’s not that hard to hang a shingle and launch a ‘coding class.’ GA’s programs were higher quality but local competitors just kept nipping at our heels and driving up the cost of customer acquisition.
“It turned out there were no moats [competitive advantages] at all in the coding school market — that made driving significant sustainable economic profit in the coding boot camp market difficult… When General Assembly started focusing on selling next-gen skills training to enterprises, it was clear none of its smaller competitors could compete with General Assembly’s brand and global footprint across N America, Asia and Europe. Big corporate clients like L’Oreal and Booz Allen wanted GA, as the market leader, to train their employees. So, ironically, the consumer business turned out to be hard economically and competitively intensive, but opened the door to build an incredibly lucrative and defensible enterprise business.”
Jason Stoffer, “How I’d Approach General Assembly Again as a VC: 9 Key Lessons,” Medium, April 18, 2018
CAC = Customer Acquisition Cost. How much does it cost to get one person to enroll in a bootcamp.
I initially found this data on Class Central, but I discovered that the General Assembly data was inaccurate. I think this is because the original source was LinkedIn. It looks like Class Central pulled graduates of all General Assembly courses (part-time and immersive), not just graduates of their immersive courses. I have corrected this in the table.
It's worth noting that I couldn't find any published graduation rates for bootcamps in 2021. Several bootcamps only made their 2020 numbers available in January this year, so the impact of COVID-19 on bootcamp enrollments is still unclear. CIRR contains some 2021 data but notably none of the big three bootcamps are currently members of this organization, which seeks to provide transparency on bootcamp outcomes data.
That said, both Flatiron School and General Assembly have their outcomes data audited by independent, Big Four accounting firms.
It shows just how much this term has become embedded in tech that when you search for “bootcamp” on UnSplash the first image that comes up is for a coding bootcamp and the second is for a fitness bootcamp. You have to scroll to the second page before there’s any sort of military image.