Objective Frameworks Beat Faster Funnels When You Count The True Cost Of A Bad Hire

AI has made the top of the hiring funnel astonishingly fast. Feed a tool 500 resumes and a job description and it returns the top ten in seconds. The speed is real, and it's also beside the point, because a faster way to produce a shortlist does nothing to fix the thing that actually determines who gets hired: the quality of the decisions made around that shortlist. Speed optimizes the funnel. What recruiting is missing is a better decision system, and that's a problem no parser solves.

Studying the challenge closely is Avirup Bhowmick, General Manager Digital Business Partner (HRM) at the India-based infrastructure conglomerate GMR Group, where he works across HR digital transformation, talent acquisition solutions, and the AI tools bolted onto both. He's pursuing a doctorate focused on how recruiters decide whether to accept or override an AI recommendation, which gives his view of the funnel a rare double vantage point: the operator optimizing the process and the researcher studying the judgment inside it.

"Do you accept the decisions coming from the AI tools at face value, or do you add your personal touch to it? Going beyond what's mentioned in the resume, understanding what the candidate is trying to express, that gets missed if you just follow what the tool is recommending," he says. The breakdown, in his view, usually starts earlier, with what the tool was asked to look for in the first place.

A shortlist is only as good as the definition behind it

Avirup's first objection to the fast funnel is that it produces a confident answer to a question no one agreed on. The parser ranks resumes against a job description, but if the job description is vague, or if the recruiter and the hiring manager never aligned on what the role actually requires, the ranking is precise and wrong. "Is the hiring manager thinking in the same wavelength as the recruiter who's scanning the resumes?" he asks.

The common failure, in his experience, is the hiring manager looking at a shortlist and feeling that "what you shortlisted versus what I thought should be shortlisted is something different." The fix is upstream of any tool: an objective, clearly defined role, with the mandatory skills and the good-to-have skills named explicitly, agreed by both sides before the first resume is screened. Without that shared definition, speed just delivers the wrong candidates faster.

The human judgment loop the tool can't replace

Even with a clean job description, Avirup treats the ranking as a starting point rather than a verdict. A strong candidate with weak self-presentation can land outside the top percentile for reasons that have nothing to do with ability, which is why he insists on a human reading past the keywords. "It's possible that the candidate has bad articulation skills and wouldn't have crafted the resume in a way that beats the system," he explains. "So it is very important that there's a human judgment loop that understands the context and interprets beyond the keywords."

This is where his research sharpens the point. A tool that lets a recruiter override its recommendation but never records why has thrown away the most valuable detail in the exchange. "The moment you ensure that the recruiter logs in a rationale which can be audited later," he says, the override stops being a gut reaction and becomes accountable judgment. The captured reasoning can even feed back in to improve the tool, helping it learn. "The profile may not contain these keywords, but when you connect these three pieces of information, that profile stands stronger." Capturing the rationale turns a silent override into a decision someone owns.

One framework, applied the same way to everyone

The same discipline, Avirup says, has to carry into the interview, where he sees the decision system break down most often. Panels frequently lack a shared framework, so the evaluation criteria drift from candidate to candidate and the scores mean nothing. "The evaluation matrix should be common for all candidates applying for the same role, so that you're not changing the rules of the game depending on the candidate," he asserts. A scoring scale only works if its points are defined: a top score should mean something specific, like a candidate who has demonstrably implemented the skill and delivered a measurable return, not a vague "very likely."

He's equally firm that feedback has to be captured immediately. "It's an essential step to normalize this so that you don't have that recency bias or that halo effect causing you to say 'Hey, that was the first candidate, but I made up my mind that this is the best candidate.'" He presses for triangulation, several independent views rather than one person's early conviction, and a fixed turnaround so the decision doesn't stall. A structured interview, scored against a defined rubric, is what makes the evaluation defensible rather than a number assigned at random.

The silence that costs you the next candidate

The part of the process Avirup thinks organizations neglect most is what happens after each stage. The candidate hears nothing, often for days, often forever, and he treats that silence as a failure of the system rather than a courtesy lapse. "There's pin-drop silence," he says of the gap between stages, and it falls hardest on the candidates who don't advance. "Most of the candidates who didn't qualify from round one to round two never receive communication at all, with no logical rationale."

His structural answer is accountability with a single owner. "You can only have one person accountable. You can't have multiple people accountable, then it leads to no accountability." Each stage gets one named owner, a defined turnaround time, and a requirement that every persona in the loop, the candidate included, is informed when a decision is made. Organizations obsess over the experience of the candidate they select, he notes, and ignore everyone else, which is exactly the silence that teaches a rejected-but-capable candidate not to apply again.

The wariness is already measurable: Gartner found that just 26% of job applicants trust AI to evaluate them fairly, and a process that leaves them in silence does nothing to earn that trust back.

The business case for doing it the hard way

Avirup is clear-eyed that a board doesn't fund frameworks for their own sake, so he frames the whole argument in the language of a bad hire's true cost. Subjectivity at the job-description stage, he says, is what produces the new hire who leaves inside six months because "this was not the job description which was described."

That early attrition carries a cost far larger than salary. "When you look at the total cost of recruitment, it is much higher than the CTC alone," he notes, counting the job-portal spend, the AI tooling, the background checks, any headhunter commission, and the opportunity cost of the role sitting vacant while an incumbent's work went uncovered. Set against that, the objectivity he advocates for isn't overhead, but a way to cut the churn that generates it.

His recommended pitch to leadership is empirical: investigate whether early attrition is actually happening, trace the root cause through normalized feedback from the manager, the business leader, and the recruiting team, and if the cause is a mismatch between what the role promised and what the candidate found, the cost of restructuring the process pays for itself against the cost of repeating the bad hire. The structured approach, he notes, is also the better-evidenced one, which is why structured interviews exist in the literature at all.

Where the data ends and the judgment begins

What connects every piece of Avirup's position is a refusal to let the tool be the decision-maker. The parser, the ranking, and the score are all data, and data is useful. But the interpretation, the override, and the final call belong to a person who's accountable for them. "The tools are there, and the data," he says, "but the interpretation and the final judgment lies with human."

The faster funnel isn't the prize the industry thinks it is, because speed without a decision system just reaches a conclusion sooner. The organizations that pull ahead, in Avirup's telling, make their frameworks objective, put an accountable human in the loop over every AI recommendation, and close the communication gaps so the silence between a company and its candidates finally ends.

Disclaimer: The views expressed are personal, drawing on the author's 11+ years of experience across diverse facets of HR and its transformation.