To see the innovative models in adult learning, look at the coding bootcamps (e.g., General Assembly, HackReactor, App Academy) and the MOOCs (e.g., Udemy, Coursera, CreativeLive) that are popping up everywhere. They are answering the call to address the skill gaps in the workforce, where there are significant shortfalls between the jobs that are created each year and the number of graduates with the relevant degrees. Many are focused on software development, though an increasing number are focusing on data science, design, cybersecurity, growth hacking, digital marketing, creative tools and so on.
All of them struggle to become mainstream
Coding bootcamps scramble to defend their job placement rates that they boast on their websites, whereas many tech employers are skeptical of hiring their graduates.
“Want a Job in Silicon Valley? Keep Away From Coding Schools”
— Bloomberg, December 6, 2016
“The Dirty Little Secrets About The Worst Coding Bootcamps Out There:
9 out of 10 programs are outright scams”
—Techspiration, June 29, 2015
And MOOCs are riddled by low completion rates and poor performance.
“The Flip Side of Abysmal MOOC Completion Rates? Discovering the Most Tenacious Learners”
— EdSurge, February 22, 2017
“University Suspends Online Classes After More Than Half the Students Fail”
— Slate, July 19, 2013
Because of credentialing.
Usually, higher ed is an expensive purchase (~$200K) that most people who choose it only make the decision once in their lives (four-year college), which makes it a highly considered purchase. Even with the new and shorter programs, the time and financial commitment it takes if one wants to to acquire a new skill that can expand employment possibilities are still significant. Coding bootcamps routinely cost tens of thousands of dollars for a few months of full-time study and most do not guarantee outcomes.
So how do people choose? They choose an institution and a program that they are reasonably certain will give them a good ROI. But it is incredibly hard to evaluate someone’s skills and predict how they will perform on the job. So employers use the crude measure of the brand of the institution that the candidate attended as a signal for quality. As for cheaper offerings, no one — neither employers nor job candidates — expect that taking a $50 course on Udemy that is openly available will be transformative to their employment prospects.
This perpetuates the behavior to seek out the most established education brand and makes it difficult for new concepts to gain a foothold, even if they are of high quality.
So, how do you solve this conundrum?
First, what not to do
Simply creating new credentials — voilà! — does not work. If it were only so easy.
LinkedIn endorsements have been around for years but nobody takes them seriously (sorry, LinkedIn). I get the theory: if you get endorsed in a skill from someone who have received a lot of endorsements in that skill, then, by transitive property, LinkedIn infers that you should also highly regarded in that skill. By mapping peer-to-peer endorsements, LinkedIn is hoping to build the classic network effects that we VCs like a bit too much.
However, people use LinkedIn in such varying ways: some only “connect” with people that they have worked closely with, while others are overly generous. As a result, the data is very inconsistent and inconsistency is not a sound foundation for a credentialing product. Aline Lerner of Interviewing.io backs this up with data. As they say, garbage in, garbage out.
A number of other startups are taking other more promising approaches.
Play the game
Some startups are shrewdly playing the game. They artificially constrain the supply of positions available at their program to signal their elite status. This scarcity in turn induces the herd effect and creates a virtuous cycle around their brand. It is counter-intuitive, but it follows the playbook of every elite brand from education to fashion.
Minerva Schools at KGI established itself as a selective higher ed institution in 2011 and started accepting students in 2013. For its first class, it received 2,464 applications from all over the world but only accepted 2.8%. By being an asset-light model, Minerva is not capacity-constrained like traditional universities. Nevertheless, Minerva signaled that it would only accept the very best, which drew more students to apply the following year. Then, they made the program even more selective. Although Minerva has full discretion over its acceptance rate, by modeling after a metric that the general public is not only familiar with but worships, it is able to propel its brand to be among the echelons of elite universities in the matter of years — not centuries.
Smartly adopted a similar strategy. It offers an MBA curriculum that students can completely learn on their smartphones with fully automated and interactive mini-lessons. The marginal cost to add a student to their app is zero, but Smartly also chooses to be highly selective. Their winter 2016 cohort boasts an average GMAT score of 729, where 86% of the students come from the top 20 schools, and students have worked at a number of brand name firms.
Why, if it’s hard to get in, it must be good!
Education, like luxury goods, is a weird market where the more widely available a product is (e.g., MOOCs), the less it is desirable; the more selective it is (e.g., Chanel), the more everyone compete to get in. Hacking this consumer psychology can allow startups to accelerate establishing their brands — and their credentials.
Democratize through assessments
Another way to short-circuit this problem is to build products based on assessments. If employers can actually measure candidates’ skills in a way that is predictive of their performance on the job, then they are more likely to take a chance with a candidate that has a nontraditional resume — maybe they are self-taught, maybe they have been educated in an unfamiliar school overseas, or maybe they have only worked at lesser-known companies.
This is why a number of assessment companies are gaining traction.
HackerRank and Codefights have developed automated assessments to evaluate candidates on their programming ability, under a gamified veneer. For employers, they see HackerRank and Codefights as sources for providing candidates that are pre-vetted on their technical chops. Unlike the traditional referral and recruiter routes, HackerRank and Codefights are not encumbered by the brand signaling bias and therefore expand the pool of candidates in a meritocratic way.
Interviewing.io views the problem from a different perspective. In this lopsided market where the best engineers hold all the power, companies need to make them feel respected and not funneled en masse into an impersonal test. Interviewing.io targets these highest quality candidates by providing them value from the start — offering them practice technical interviews and direct access to engineers at sought after tech startups, skipping recruiters altogether.
A reader wrote to me about my last article asking why I give so many examples in coding. An astute observation! Coding is a hotbed for education innovation because it has the largest skill gap today. Employers are motivated to try something new and are willing to pay because they cannot find enough qualified candidates and traditional means for sourcing are becoming very expensive. As certain models prove themselves in coding, I am hopeful that they can scale to other subjects and at a lower cost.
Nevertheless, I will highlight examples outside of coding.
Interviewed tackles a completely different set of jobs: accounting, administrative, customer service, design & creative, IT & networking, sales, and so on. It built a variety of assessments including cognitive and psychometric tests, english proficiency tests, integrity and honesty tests, personality tests, and job skill tests. (Although the company does offer programming tests, I expect the target roles to be different from what the aforementioned companies are focusing on.) Streamlining assessments to only surface the best candidates for your job opening is a lot better than being swamped with applications of varying quality from job boards. Then, building a feedback loop between the outcomes for the assessments and the employees’ performance on the job, once hired, would allow Interviewed and the employer to iteratively improve the evaluation rubric — which makes for a sticky product.
Imbellus, an Upfront portfolio company, and Scoutible take a different approach. They carefully design games that are analogous to real jobs. Then, they observe candidates playing the game and through the candidates’ clicks, pauses, and actions, they infer the cognitive and non-cognitive abilities of these candidates. By enabling the candidates to show, not tell, they are able to study, without bias, how candidates would behave in analogous situations in the span of an hour or less.
I cannot confirm yet if we got it right on any of these models, but more experimentation on assessments is certainly good. If we build assessments that are predictive of success on the job and are widely accepted by industry, these assessments will become the de facto standard. They will make credentials in many cases moot.
Sidestep the question
A third way to address the credentialing problem is to sidestep the necessity of traditional credentials through networking, getting a friendly introduction, and putting a foot in the door through trial positions. In real life, credentials are mostly important in selecting between strangers. For a friendly face, of course they will be given a chance.
Purple Squirrel is facilitating friendly referrals. It is building a two-sided marketplace of employees at different companies who are willing to take informational interviews and job candidates who are interested in working at these companies, for a reasonable fee. The reality is that many people find their jobs through referrals. If the employee liked the candidate, then they can give them a referral, which moves the candidate to the top of the stack of resumes for HR.
A different form but to the same end are startups setting up mentoring and alumni networks for higher ed (e.g., PeopleGrove, AlumniFire, CampusTap, Switchboard). At the same time, they are also addressing an additional pain point for higher ed institutions: to improve the job placement metrics for their graduating seniors and thereby demonstrate their ROI.
Coding schools are also getting creative. Udacity launched the Blitz program in November 2016 where they set up grads from its signature Nanodegree programs to work in teams on real projects for companies. Companies pay Udacity a predetermined fee and Udacity is responsible for any overages. For the Nanodegree graduates, this part-internship, part-contract development program offers them a chance to get to know the companies they work with and, with it, a pathway to full employment.
General Assembly also has an Outcomes Team that focuses on helping students connect with employers. They set up career fairs and work with employers to make available apprenticeships or internship positions to try out new grads.
Using technology to facilitate connections between people is applied here as it has been applied in many other contexts to wild success. These people happen to be jobseekers and employers. Whether incepted as a standalone product or as add-ons to education programs, enabling human connections can divert from the question of credentials.
Today, any upstart educational program needs to contend with the question, “are you another University of Phoenix?” Unfortunately, these for-profit private universities have smeared the public perception of online education. Hopefully not for long.
With a number of startups taking different approaches to crack the credentialing challenge, this space is certainly bustling with innovation. What excites me the most is that if these startups succeed, then they will open the floodgates to innovation in education.
✌ I am an early stage investor at Upfront Ventures. I am passionate about the intersection of improving human potential and big, transformative businesses. If you care about the same things, drop me a line at firstname.lastname@example.org.
✌ If you like what you read, please ❤ it below — as a writer it means the world!