Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Most of the applications I see have little to do with the job description. In my experience, somewhere between 90 and 95% of them are not aligned with the role they were sent to. That is my own observation, not a published statistic.
The problem is not lazy candidates or overworked recruiters. It is the front door: we ask for a CV and a click, then act surprised when so many people click.
My argument is simple. Move the evidence to the start. Ask for a portfolio, a short case study or a short recorded video at the point of application, and let the work do the first round of screening. For ed tech, a sector that builds things for learners, that is both practical and a natural fit.
The cost of applying has collapsed, while the cost of reading an application has not moved at all. That gap explains most of what recruiters are seeing.
Greenhouse data, reported by TheHireHub.AI in August 2026 and covering more than 6,000 companies, shows how much changed between 2022 and 2025:
AI tools are a large part of the story. LinkedIn now processes about 11,000 applications a minute, a 45% rise on last year, and 77% of hiring teams say they regularly see AI-generated or AI-assisted applications, up from 53% in early 2024. Greenhouse research reported by Recruiting News Network found that 34% of recruiters spend up to half their working week filtering out spam or low-quality applications, and 91% have met candidate deception of some kind.
The deeper issue is signal, not volume. When many applications are written by the same few tools against the same job description, they start to look alike, so reading more of them tells you less. Greenhouse’s chief executive has described the result, as reported by TheHireHub.AI, as an AI doom loop: candidates apply more with AI, employers filter more with AI, and rejected candidates apply even more.
No published study measures the exact share of applications that miss the brief. The wider data points the same way as what I see: far more applicants, and fewer of them genuinely aligned with the role.
A CV describes work; a portfolio, a case study or a short recording shows it. Putting that evidence at the point of application changes who bothers to apply.
Recruiting glossaries from X0PA and Pitch N Hire both rank work samples among the strongest predictors of job performance. Pitch N Hire notes they are usually used after an initial screen, for shortlisted candidates, while X0PA recommends using them early. My argument is to bring a light version to the front door when volume is high.
One example shows how that works as a filter. X0PA reports that when Fair HQ hired a business intern, 900 people started the application but only 90 completed the work-sample tasks, leaving the most committed tenth of the pool. X0PA also describes tasks of 5 to 15 minutes as a filter for high-volume roles, set before candidates submit an application.
Video needs the most care. Vervoe reports a significant drop-off at the fourth recorded answer and recommends two or three questions at most. It also suggests asking candidates to respond to a realistic scenario rather than describe a past one, and opening with a short video from the hiring manager.
Ed tech is growing in revenue while its venture funding shrinks. That makes every hire count and every claim worth checking.
The global market is worth about $189 billion in 2025 and may reach $215 billion in 2026, according to Fortune Business Insights as cited by TechnBrains. Estimates vary widely with how each firm defines the sector.
Venture funding is moving the other way. HolonIQ data reported by 36Kr shows global edtech venture capital falling from a peak of $20.8 billion in 2021 to $2.6 billion in 2025, a drop of 87.5%. The first half of 2026 came in at $1 billion, down 25.7% year on year, even though the number of deals was roughly unchanged.
Investors are placing smaller, more selective bets. Vocational training and K-12 were the two most active tracks in the first half of 2026, together making up about 75% of deals, and the report says capital favours companies that create measurable value for schools, teachers and learners. TechnBrains makes the same point from the buyer side: only 45% of the most-used tools have published research meeting a recognised evidence tier, and the products that last solve a nameable problem and can prove it.
A sector that sells on evidence should hire on evidence. Three features of ed tech hiring strengthen the case.
Ed tech also practises this idea in its own products. HolonIQ notes continued investor attention on alternative credentials, professional certifications and work-integrated learning, all of which prove skill rather than claim it. Hiring on demonstrated work turns the same idea inward.
Front-loading evidence only works if candidates trust the process and can afford the time. The data suggests they welcome the idea, but not every way of doing it.
The encouraging signal comes from Criteria’s 2026 candidate-experience report: 68% of job seekers said they would happily drop the traditional resume for skills assessments and structured interviews. The warning comes from Greenhouse’s 2026 survey, where 38% of candidates said they had abandoned a hiring process because it used an AI interview. Both findings are reported by HeroHunt, which adds that the most common trigger was a pre-recorded video scored by AI with no human present.
That points to a handful of design rules.
One caution on live exercises. X0PA notes that tasks done under observation, such as live coding, can create gendered performance gaps, and recommends take-home work instead.
Rules on automated assessment also differ by country and are still moving. For example, HeroHunt reports that the EU has pushed some high-risk hiring-AI obligations back to December 2027, so check the position wherever you hire.
Start small: one exercise per role, scored against a rubric and reviewed by a person. Then adjust using the results.
The examples below are starting points for common ed tech roles, not tested templates.
The engineering prompt reflects how seriously ed tech takes student data. TechnBrains argues that privacy in education has to be designed in from the start, not added as a setting.
The flood of poorly aligned applications is a front-door problem, so the fix belongs at the front door. Ask for evidence, whether a portfolio, a mini case or a short recording, and let the work do the first round of screening.
Ed tech is a sector built on proving that learning works, which makes it well placed to lead. Keep the ask small, keep a human in the loop, and tell candidates what you are doing and why.
Try it on your next vacancy: pick one role, add one 10-minute exercise, and compare the shortlist with your last one. I would be interested to hear what changes.
Figures are as reported by these pages on 22 October 2026. Several come from vendor blogs and glossaries, so treat them as indicative rather than definitive.
