Anyone can sell a course and print a certificate at the end of it. Only a few of those certificates mean anything to anybody.
Jeffrey Horne is head of learning and development at SCTE, a nonprofit association for the broadband industry: it trains and certifies the technicians who design, install, and maintain the networks, and it writes the technical standards the industry uses. Before that he worked on the consumer side of Coursera and helped universities turn on-campus graduate degrees into hybrid online programs.
Why does the market trust one credential and ignore the next? Jeffrey's answer comes down to a handful of decisions any training business makes when it creates a certification program of its own.
Building credibility without a big name
Look at Coursera and the easy story is borrowed credibility: two Stanford professors, a famous name from day one. Their credibility started years earlier, in their own research and subject matter depth. Getting hired at Stanford was, as Jeffrey puts it, “simply a validation” of work already done.
For an organization without that head start, the path begins small: prove value for early clients, solve their problems, and let the referrals compound. Humans are tribal, and tribes refer and recommend. That is how credibility grows until it hardens into incumbency. Jeffrey points to the PMP. Project managers who hold it got hired at top companies and performed, so the market's default advice to an aspiring project manager is simply “go get your PMP.” Over time the credential reinforces itself. You can have the best marketing in the world, Jeffrey says, and it “will only get you so far,” because eventually “the substance and the applicability of your credential or your content will be sussed out.”
The mistake: starting with topics instead of outcomes
Jeffrey's first point: “You have to start with an outcomes-focused process.” Too often, he sees people open with topics or content and try to build from there. Flip it. Name the tangible skills a person must execute in the field, then backwards-map from those. Does a certified person need to assemble presentation decks and deliver them? Then design the whole path back from that performance.
What you test decides what your certificate means
The same backwards-mapping governs how you test. Does the learner need to know things or do things? That answer decides whether the credential ends in an exam, a portfolio, a hands-on sign-off, or some combination.
Hands-on sign-off brings its own problem: how do you scale it, and how do you document that someone was signed off? That is where teach-the-teacher models come in. Either every candidate comes to you for sign-off, or you certify a group of trainers who deliver the training and sign off for you. Both need rubrics that evaluate whether a person can actually do the skill, and exam questions structured to assess what they will need to know and do.
The failure mode Jeffrey keeps encountering: organizations launching an AI-generated multiple choice assessment that tests the nuances of topics, which is terrible learning anyway. His illustration is Columbus: everybody knows the rhyme about 1492, but the date is trivia. What matters is grasping why exploration mattered in that period – people, technology, and trade were all expanding – because that is what applies to today. Exam questions should probe that level of understanding, not dates.
Why your best practitioner may be your worst teacher
Great practitioners are not automatically poor teachers; Jeffrey notes that some of the brightest and best are also the best teachers. But when a great practitioner does fail at teaching, the cause he names is automation. The mind of an expert has “really automated like 17 substeps between A and B,” so a move registers as a single action: “I do X and Y occurs.” The beginner, meanwhile, trips on exactly those invisible substeps. A good teacher, whether world-class or intermediate, is the one who can parse the substeps back out.
It also explains why people say you have to have a feel for it. Jeffrey traces it to vocabulary: the instructor has not developed the words for the substep or technique they are using. Their body does it on its own. The fix is to break the outcome into subskills, train the mechanics, then bring it back together so the learner automates the whole motion.
“The Inner Game of Tennis” is the book Jeffrey points to here: at the highest level you stop coaching racket position and start asking where the player wants the ball to go.
The economics: demand, then who pays
Before building anything, two market questions need answers. First: is there demand for the skill? A credential is a signal, and signals obey supply and demand. When many organizations need a skill and few people have it, a credential carries value. If everyone can do the thing, nobody pays extra to certify it. And even a strong credential usually works as an early filter; holding it does not mean you automatically get the job.
Second: who pays? In his online-degree work, Jeffrey says, some of the most popular and most lucrative programs were business programs, because working professionals spend their own money when they expect a return on it. In broadband it can run the other way: in some cases contractors and subcontractors pay to credential their own workers, because they need skilled people on infrastructure. Jeffrey puts the question in toy-company terms: “Who's your target audience? Is it the kid? Or is it the parent? Who's got the money? Who's got the purse?”
Sometimes it is both, and Jeffrey recommends a good market analysis before choosing: who is your target market, how do you reach them, where is your niche. He also raises public funding, which exists even in broadband. The budget holder is not always the learner.
New credentials win where the incumbent has not moved
If you are new, do not attack the established standard head-on. “The opportunity is not in the established aspect,” Jeffrey says, “but the new component that's affecting change within the industry.”
Take agile. Hyper-focused upstarts built agile certifications while the PMP was the incumbent, and the PMP's owners responded by folding agile content into their own learning paths. Broadband had its own version: fiber arrived as a serious evolution over old coax, and SCTE had to develop fiber credentials and hands-on training because the demand was there. The pattern for an upstart: find the innovation the incumbent has not standardized yet, credential around it, then scale through contracts.
Incumbents have to keep up, too. Most credentialing bodies revise the credential and its standards at least every four years by Jeffrey's estimate, and more often where the industry moves quickly. Stop maintaining it and competitors step in, because employers still need the current skills and will find other places to get them.
Three things AI changes about credentialing
The first is validation: anyone with access can potentially cheat on any exam, which forces credentialing bodies to keep evolving proctoring and to ask whether a simple exam is still sufficient proof of skill. The second is production: AI generates content quickly, at a similar production quality. Jeffrey chooses that phrase carefully, because “there is a difference between production quality and actual learning quality.”
A mentor of Jeffrey's worried that AI would “scale the bad learning that's out there” – which, Jeffrey says, it very much can. His counter is a human in the loop. When SCTE migrated course content from one platform to another, AI sped the project up significantly – and subject matter experts still went back through the courses to confirm the migration had not changed what was being taught. For him, the standard depends on the stakes: “90% or below is fine for putting content on Instagram. It's not great when you're a credentialing body.”
The third is opportunity. AI is being infused into everything, which opens a market for tailored tools that go beyond chatbots: agents that engage the learner, and tools that evaluate whether a person can perform a skill. Jeffrey adds that this takes a lot of due diligence on the rubrics underneath.
For learning professionals, Jeffrey expects AI to take over a lot of general design work, and the advantage to go to those who understand assessment, rubric development, and learning data.
What it takes to create a certification program that lasts
SCTE is a standards body, not a membership site, but a credential behaves the same way on WordPress. It is a claim about what its holder can do, and employers test that claim every time they hire one. Everything above is about surviving that test.
Jeffrey Horne is head of learning and development at SCTE. Find him on LinkedIn.
Frequently asked questions
How often should a certification be updated?
Jeffrey Horne, who runs learning at SCTE, estimates that most credentialing bodies revise their credential and its standards at least every four years; in faster-moving industries, they may update every year or two. An institution that stops updating invites competitors, because employers still need current skills and start looking elsewhere to develop them.
Is a multiple-choice exam enough to certify someone?
It depends on whether the skill is something to know or something to do. Knowledge can be checked with a well-built exam. Skills that must be performed need portfolios, hands-on sign-off, or rubric-based evaluation. An auto-generated quiz that tests topic trivia is weak evidence for either, so the assessment strategy has to be designed before any exam questions get written.
How do you scale hands-on skills testing?
Scaling pushes you toward the teach-the-teacher model: rather than every candidate coming to you for sign-off, you certify trainers who deliver your training and sign off on skills on your behalf. Those trainers need rubrics that evaluate whether a candidate can do the skill, and any exam questions have to assess what the person will need to know and do.
Can AI write the content for a certification course?
AI can speed up production, but production quality is not learning quality. SCTE used AI to migrate course content between platforms and still invested in subject matter experts to verify nothing changed in what was being taught. In Jeffrey's words, “90% or below is fine for putting content on Instagram. It's not great when you're a credentialing body.”
Running a Membership or eLearning Platform on WordPress?
MemberFix provides managed WordPress operations for membership, eLearning, and community platforms — so your team never has to worry about the tech breaking during a launch or losing members to site issues. We handle development, support, updates, and monitoring with SLA-backed guarantees. Trusted by industry leaders like Echelon Front, TechServe Alliance, and hundreds of others since 2014.
About MemberFix Radio:
We interview experts and share insights for membership site owners and subscription business leaders building sustainable, profitable recurring revenue models.
