By Nhi (Yiyi) Vu, Senior Operations Specialist at Flowmingo
Ask any engineering manager what has changed about applications in the last 2 years and the answer tends to be the same. The CVs got better. Summaries are fluent, skills sections follow the job description closely, and project write-ups use the right vocabulary.
Polished writing used to say something about a candidate. Today it says very little, because polish is available to anyone with an AI assistant and a job description to paste in.
For CIOs and CTOs in India, this arrives at an awkward moment. The Data Security Council of India’s skilling report, quoted recently in these pages, found that 73% of surveyed enterprises reported limited availability of skilled cybersecurity candidates and 63% cited gaps in practical, hands-on capability.
Technology leaders are short of people who can do the work, and the document they use to find those people has become harder to read for exactly that.
What the numbers show
We wanted to know how much of the strongest talent a CV cut leaves behind, so we set a CV score beside an interview score for each applicant. The data covers nearly 17K candidates in 413 hiring projects run on Flowmingo, a free AI hiring platform, between July and September 2026. Both scores come from AI and run out of 10, and the interview is a first-round video interview.
In nearly 6 in 10 of those projects (57.6%), the candidate who interviewed best had a CV outside the top 20% for that project. In nearly 1 in 4 (23.8%), that person’s CV sat below the project’s median. Only about 1 in 9 (11.2%) also held the highest CV score.
Where the CV of the candidate who interviewed best ranked within their own project, across 349 projects: 57.6% outside the top 20% of CVs, 23.8% below the median and 11.2% with the highest CV score.
A fair reading needs 2 cautions. Both figures are AI scores and neither is a hiring outcome, so the Flowmingo data cannot say who went on to succeed in the job. The projects also cover all kinds of roles rather than technical roles alone. What the numbers do show is that a CV score and an interview score measure different things, often enough that the order of the 2 matters.
Why this matters more in technical hiring
A common shortcut in a busy quarter is to read the top fifth of the pile and stop there. It feels responsible, and with hundreds of applicants for a cloud or data role it can feel like the only option. On the figures above, that shortcut would have missed the strongest first-round performer in more than half of the projects.
The people most exposed to that cut are easy to picture. A support engineer who has spent 4 years fixing production incidents may write a plainer CV than someone who has never been on call.
A self-taught developer from a tier-3 college, or someone returning after a career break, may lack the keywords a screen is looking for. Sumit Sarabhai made a related point in this publication about executive search in India: the most familiar CV is not always the candidate whose judgement fits what the business now needs. The same logic holds further down the organisation.
Diwakar Dayal’s article made the matching point about certifications, which can show foundational knowledge without showing how someone handles a complex incident under pressure. A CV has that limit too, and AI writing help has made it harder to see.
Hearing more people, without adding hours
None of this means interviewing everyone in person. A panel’s time is the scarcest thing in the process, and it should stay reserved for the final stages. The practical change is earlier and smaller: put a short first round in front of more applicants, then use that evidence to decide who reaches the panel.
3 choices make that first round useful for a technical team:
Ask everyone the same questions. When each candidate gets the same 4 or 5 questions, a hiring manager compares answers rather than impressions. Build the questions from real situations on the team, such as a failed deployment, a production bug with incomplete logs or a trade-off made under deadline.
Agree on the rubric first. Write down what a strong answer sounds like before anyone listens to one. That keeps the review honest, and it gives a hiring team a record of why each person went through.
Widen the cut, then measure it. Instead of the top 20% of CVs, try inviting the top 50% for one role this quarter. Compare who reaches the final round with who would have reached it under the old cut. That single comparison will tell a CTO more about their own funnel than any benchmark, including the Flowmingo figures above.
Where technology leaders still decide
Automation is well suited to the first conversation, because it brings more candidates into view and gives each of them the same chance to show how they think. It should not make the decision. A person should review the answers, weigh them against the context of the team, and own the shortlist. That matters for fairness as much as for quality, and candidates are entitled to know how they are being assessed.
The hopeful part of the finding is how little it asks of a hiring team. The talent shortage that technology leaders describe is real, but some of the people they are looking for may already be in the applicant pile, sitting just below a line drawn by a document that everyone can now write well. Hearing a few more of them before the cut is where those candidates turn up, and the CTO still makes the call.
About the data: 16,961 candidates with both an AI CV score and an AI first-round interview score (each out of 10) in 413 hiring projects with at least 10 such candidates each, from 214 organisations (the largest contributed 8.3% of the candidates); applications created 10 July to 30 September 2026. Percentages are of the 349 projects with a single top interview score. Scores are not hiring outcomes.
–Nhi (Yiyi) Vu is a Senior Operations Specialist at Flowmingo. She studies how candidates and hiring teams move through first-round interviews and turns what the data shows into practical guidance for people who hire.