Ask any recruiter where they lose candidates and they will point outward. A competing offer, a counter from the current employer, a change of heart. Sometimes that is true. More often the candidate did not leave for a better offer, they leaked out of a hole in your own process. They waited nine days for a reply, could not tell what stage they were in, or got passed from recruiter to hiring manager with no handoff and quietly went cold. A leaky pipeline does not feel like a crisis because each loss looks like bad luck. Add them up and it is the biggest tax on your hiring.
Where candidates actually drop out
The leaks are boringly predictable. Once you know the four big ones you can design around them.
- Slow response. The single biggest killer. A strong candidate who applies or replies on Monday and hears nothing by Friday assumes you are not serious, and by then they are three interviews deep somewhere else. Speed is a signal of how much you want them.
- Unclear stages. When a candidate does not know whether they are being considered, waiting on a decision, or already rejected, the silence reads as rejection. Ambiguity makes people disengage.
- No follow-up. Someone goes quiet after a good first call and nobody nudges them. Not because they lost interest, but because the recruiter is juggling ten roles and the thread fell off the screen.
- Bad handoffs. The recruiter builds rapport, then hands the candidate to a hiring manager who has not read the notes, asks questions the candidate already answered, and the momentum dies in the gap between the two.
Notice that none of these are about the candidate's quality or your comp band. They are all process failures, which means they are all fixable without spending a rupee more on salaries.
Design your stages before you design anything else
A pipeline without named stages is just a pile of resumes with vibes attached. You need a small, ordered set of stages that everyone on the team uses the same way. Small is the operative word. Fifteen stages feel thorough and are impossible to keep clean. A workable default for most India hiring looks like this.
- Sourced to Invited to In progress to Review to Shortlisted to Offer to Hired, with Rejected as the exit any candidate can move to.
Each stage has to mean one thing, and moving a candidate has to mean a real change in status, not a mood. The test is simple: could a new team member look at the board and know exactly what needs to happen next for every person on it? If a candidate can sit in "In progress" for two weeks and nobody can say what "in progress" is waiting on, the stage is not doing its job. Clear stages are also what make it possible to measure your funnel and see where the leak is, which is the first step to any real effort to reduce time to hire.
Put an SLA on every stage
Stages tell you where a candidate is. SLAs tell you when they have been there too long. This is the mechanism that turns a static board into something that actively prevents leaks. Assign each stage a maximum time, and treat a breach as an alarm, not a nice-to-have.
- New application or reply to first contact: within 24 hours, ideally same day.
- First screen to decision on next round: 2 to 3 working days.
- Interview to feedback logged: 48 hours, before memory fades and the candidate cools.
- Final round to offer or reject: within a week, with the candidate told the timeline up front.
The exact numbers matter less than the discipline of having them and acting when they break. The most common failure is not that a stage is too slow, it is that nobody notices it is slow until the candidate is gone. An SLA you do not monitor is a wish. An SLA that pings you the moment it breaches is a safety net.
A simple rule that plugs most leaks
No candidate goes more than 48 hours without a next step or a status update. Even "we are still reviewing, expect to hear by Wednesday" keeps someone warm. Silence is what kills. If you enforce only one thing on this list, enforce this one, and watch your drop-off fall.
Keep candidates warm without living in your inbox
The reason follow-ups get skipped is not laziness, it is load. A recruiter running a dozen roles simply cannot hold every thread in their head. This is exactly the kind of work that should be systematised, because the cost of forgetting is so high and the fix is so mechanical. Two things solve most of it: a pipeline that shows you the whole board at a glance, and automated follow-ups that fire when a candidate has been quiet too long.
Automated does not mean impersonal. A short, well-timed nudge that references the last conversation keeps momentum without feeling like spam. The same principles that make cold outreach work also make follow-up work, and if you have not read it, the piece on outreach that gets replies applies just as much to a candidate mid-pipeline as to a first touch. The goal is that no good candidate ever goes cold because a human was busy.
How TalentGPT keeps the pipeline tight
This is precisely what TalentGPT's pipeline and task inbox are built for. The pipeline is an ATS-style board where every candidacy sits in a named stage, so you always know who is waiting on what. When something needs a human, it does not depend on you remembering. A candidate reply, a completed AI interview that needs review, a LinkedIn task, an outreach issue, all of it lands as an item in a single task inbox with a deep link to the exact place you need to act. You work the inbox to zero instead of hunting across threads and tabs.
Talia, the AI recruiter, runs the mechanical parts of keeping people warm. It sends the follow-ups you approve, moves candidates through stages as they progress, files a review task when an interview scorecard comes back, and flags the enrollments that stalled. The handoff problem shrinks too, because every note, activity, and interview result lives on the candidacy, so whoever picks it up next has the full history in front of them. You are not rebuilding context, you are continuing it. Getting this right compounds: fewer leaks means you convert more of the pipeline you already paid to source, which is a large part of why some teams manage startup hiring on a budget without a big team.
Measure the leak, then close it
Once your stages are clean and your SLAs are set, you can finally see the funnel. Look at where the biggest drop between two stages happens and treat that gap as your project for the month. Maybe eighty percent of screened candidates never reach the hiring manager because feedback logging is slow. Maybe offers are dropping because the final decision drags. You cannot fix what you cannot see, and a well-designed pipeline is mostly a tool for seeing. That visibility is also what underpins any serious attempt at outcome-based hiring, because you can tie process changes to real hiring results rather than gut feel.
Frequently asked questions
How many stages should a hiring pipeline have?
Five to eight for most roles. Enough to represent the real steps a candidate moves through, few enough that the board stays clean and everyone uses the stages the same way. If you find yourself adding a stage to capture a mood rather than a status, resist it. Complexity is where pipelines rot.
What is a reasonable response time to a candidate?
Within 24 hours for a first reply, and never more than 48 hours of silence at any point after that, even if the update is just "still in progress, expect to hear by X." Speed early is read as genuine interest, and it is often the difference between winning a candidate and losing them to a faster competitor.
Do automated follow-ups annoy candidates?
Not if they are timely, short, and reference the actual conversation. What annoys candidates is silence, or generic mass mails that ignore where they are in the process. A relevant nudge at the right moment reads as a company that has its act together, which is a point in your favour.
How do I stop candidates going cold during handoffs?
Keep every note, activity, and interview result on the candidacy itself, so anyone who picks the candidate up has the full context. When the hiring manager can see the recruiter's notes and the interview scorecard before they speak, the candidate never has to repeat themselves and the momentum carries across the handoff instead of dying in it.
A pipeline that does not leak is not about working harder, it is about never letting a good candidate wait in silence. Run your hiring on TalentGPT's pipeline and task inbox and stop losing people to your own process instead of the market.