The resume was never that good anyway
In 2021, the average job posting received 28 applications. In 2025, it received 95. That 3x jump happened with no change in the actual number of qualified candidates — it happened because AI tools made applying frictionless.
A third of recruiters now say AI-polished resumes are their single biggest source of stress. They're not wrong to be stressed. When every application arrives with the same clean keywords, identical action verbs, and polished formatting, the resume has stopped being a signal. It's noise dressed up as signal.
This is the interview crisis nobody is naming clearly: the top of your funnel is now synthetic, and if you haven't changed how you evaluate people below it, you're flying blind.
Here's the uncomfortable truth: resumes were always a weak predictor of actual job performance. Decades of HR research consistently put unstructured resume screening in the bottom tier of hiring accuracy — roughly on par with astrology, the joke goes.
What AI-polished applications have done is simply make the weakness impossible to ignore. When 95 CVs land in your inbox and 70 of them are technically flawless, the screening layer has collapsed. You can't skim your way to a shortlist anymore.
Founders and early-stage teams feel this acutely. A bad hire at a 12-person company doesn't just slow a project — it can set back the whole trajectory. And yet the default response to resume inflation is more resume screening: more ATS filters, more keyword matching, more ways to rank a document that's been optimized to pass your filters.
More filters on broken signal is still broken.
What structured interviews actually fix
Structured interviews are not the same thing as lengthy interviews or formal interviews. They're specific: every candidate gets the same questions, in the same order, evaluated against the same criteria before the call ends.
The research here is unusually strong. Structured interviews predict job performance roughly twice as well as unstructured conversations. The gain comes from two places:
Consistency eliminates bias anchoring. When every interviewer follows the same script, you remove the effect of first impressions and the natural drift that happens when you let a candidate steer the conversation. You stop rewarding people who are good at seeming competent and start rewarding people who demonstrate competence.
Pre-defined criteria force the real question. Before you write the interview questions, you have to write down what "good" actually looks like for the role. That exercise alone — what does success look like in 90 days, what behaviours predict it, how do you verify them — is one of the highest-leverage things a founder can do before a single application lands.
The skills-based pivot is the same idea applied upstream
Skills-based hiring asks the same forcing function even earlier: instead of specifying credentials, specify capabilities. Instead of "5 years experience in B2B SaaS," write "can map a complex buyer journey and identify where deals stall."
This matters in 2026 because the credential signals are deteriorating fast too. An AI-augmented workforce means someone with 2 years of focused AI-assisted work can genuinely outperform someone with 8 years of unassisted work. Experience as a proxy for skill is weakening at the same pace as the resume as a proxy for fit.
The companies getting this right are building hiring processes around demonstrated capability — short paid projects, structured work samples, scenario-based questions — and using the resume as a rough pre-filter rather than a decision tool.
For founders specifically, there's another dimension: the people you hire in the first 20 shape everything that comes after. You can't afford to pick from a synthetic pool with an unstructured gut-feel conversation. The stakes are too high for a process that good.
What to do this week
If you're hiring right now, three practical moves:
- Write your evaluation rubric before you write your job description. Decide what great looks like, write 5 questions that would surface it, score them 1–3. Then write the JD. This takes 90 minutes and will save you 10 bad conversations.
- Treat the resume as a flag, not a score. Read it for red flags only. Don't rank candidates by resume quality — you're ranking their AI tool, not them.
- Use the same questions for everyone in the same role. Even imperfect consistency beats none. If one interviewer covers culture fit and another covers technical depth, you can't compare candidates — you've run different processes on the same role.
The AI resume wave is not a temporary problem that will resolve itself when companies get better screening tools. It's a permanent shift in the signal-to-noise ratio at the top of every hiring funnel. The response isn't better filters — it's better questions.
Umamy helps founders build structured, skills-based hiring processes — learn how it works.
