Our AI screening tool keeps rejecting candidates we would have interviewed
Daniel Okafor · 14 Jun 2026 · Ireland · Software · 201-500 employees
We turned on the ranking module in our ATS three months ago. Twice now a hiring manager has found a strong candidate in the rejected pile, both career changers with non-linear CVs. Has anyone found a workable way to keep the volume benefit without losing people the model does not recognise?
What they have tried
Lowered the score threshold, which just moved the problem. Asked the vendor for their scoring features and got a marketing answer.
Desired outcome
A practical configuration or process check that catches this before rejection.
2 replies
Stop using the score as a rejection gate. Use it only to order the review queue, and require a human decision on every rejection. You keep the triage speed and you no longer have an automated screen-out to defend.
Also run a monthly sample: pull 25 rejected applications at random and have a recruiter review them blind. If more than one or two would have progressed, your configuration is wrong and you now have evidence to take to the vendor.
Join this conversation
New replies appear automatically in this realtime feed.
Sign in to answer, follow this thread and receive reply notifications.
Sign in to participateCommunity answers are general guidance, not legal advice. Confirm local obligations with a qualified adviser.