Standard pipeline stages for a software hiring team

A software hiring team needs six pipeline stages: New, Applied, Screen, Technical, Team interview, and Offer. New is the intake safety net for candidates who arrive outside the apply button; the other five map one to one onto the decisions a hiring team actually makes. Each stage gets a single owner, a binary exit criterion, and a target SLA measured in days. If a proposed stage does not change the question being asked or the person answering it, it is a status, not a stage.

Why model the pipeline around decisions?

A pipeline is a queue of decisions, not a diary of activity. Each column on the board is a question, each card is a candidate waiting for an answer, and the column owner is the person on the hook for answering it. That board, plus the audit trail underneath it, is most of what an applicant tracking system is.

Designing around decisions keeps the stage count honest. Software hiring involves a fixed set of distinct questions: is this application worth a conversation, can this person do the job in broad strokes, can they really build things, will they work with this specific team, and will they accept. That is five, plus an intake stage so nothing enters the pipeline unseen. Six.

What does each stage mean, and what gets a candidate out of it?

Stage	The question	Exit criteria	Target SLA	Owner
New	Has a human looked at this person yet?	Triaged into Applied, or rejected with a reason	1 business day	Recruiter
Applied	Is this worth a conversation?	CV checked against must-haves; advance or rejection sent	3 business days	Recruiter
Screen	Do skills, salary and logistics roughly fit?	Screen call held; notes and expectations recorded	5 business days	Recruiter
Technical	Can they actually do the work?	Technical round held; written feedback filed	7 days, feedback within 24 hours	Hiring manager
Team interview	Will this hire work, for them and for us?	Panel held; debrief done; hire or no-hire recorded	5 business days	Hiring manager
Offer	Will they say yes?	Offer signed, declined, or expired	Out within 2 business days of debrief	Recruiter and hiring manager

A few notes on the rows that cause the most arguments.

New exists because candidates do not only arrive through the apply form. Referrals, sourced profiles, agency submissions, and people revived from the talent pool all need somewhere to land where the only promise is that a human will look within a day. Without it, these candidates get dropped into whatever stage someone guesses, or never appear on the board at all.

Screen is a recruiter call, not a technical filter. Thirty minutes to verify motivation, level, salary expectations, notice period, and location or language constraints. Resist the urge to make it technical: that doubles its length, burns engineering time you do not have at this volume, and duplicates the next stage.

Technical is where teams spend their credibility, so its exit criterion is strict: written feedback filed within 24 hours, whatever the format (live exercise, take-home review, system design conversation). Unfiled feedback is the single most common place pipelines silently die. And if the round is not telling you much, the problem is usually the questions, not the stage; we cover that in technical screening questions that verify real skills.

Team interview answers a different question than Technical, which is why it stays separate even though both belong to the hiring manager. It is about collaboration and whether the candidate wants this team as much as the team wants them. The exit is a debrief with a recorded hire or no-hire, held within a day of the panel, not a thread of partial opinions.

Offer has a deceptively simple SLA: out within two business days of the debrief. Most offer losses are momentum losses rather than negotiation losses, and the patterns behind them are predictable; why candidates turn down offers walks through them.

What makes an exit criterion any good?

Three properties. It is binary: held or not held, filed or not filed, sent or not sent. It is recorded in the ATS, not in someone’s head or a chat thread. And it is owned: the stage owner makes the call, even when the input comes from five interviewers.

There is a fourth, softer property: every exit should trigger a message to the candidate. Advancing and hearing nothing feels identical to being rejected. Stage transitions are precisely where automated candidate emails can still feel personal, because they are tied to something real that just happened. A workflow that keeps candidates engaged is mostly a workflow where stage changes and messages travel together.

Why do the SLAs matter more than the stage names?

Candidates never see your column labels; they experience the gaps between them. A pipeline with carefully named stages and ten days of silence between Screen and Technical reads, from the outside, exactly like rejection. Strong engineers are usually in two or three processes at once, and the team that finishes this stage set in about three weeks wins ties against the team that takes six, without being better at anything except scheduling.

Per-stage SLAs beat one end-to-end target because they localise the problem. The weekly pipeline review becomes mechanical: count the cards over SLA in each column and ask that column’s owner what they need. Over-SLA cards in Technical almost always mean unfiled feedback. Over-SLA cards in Offer usually mean an approval chain that needs shortening.

What does the 14-stage anti-pattern look like?

It grows one reasonable column at a time. Someone wants to see which screens are scheduled versus completed, so Screen becomes two stages. Feedback lags, so an Awaiting feedback column appears. A staff-level role needs a second technical round, so Technical 2 lands and stays forever. A year later the board has fourteen columns, three holding a single card and two that nobody can define.

The damage is not the clutter, it is the falsehood. Fourteen columns are expensive to update, so updates get batched on Friday or skipped, and the board drifts away from reality. Once recruiters stop trusting it, the real pipeline moves into heads and spreadsheets, and every report built on stage data (conversion rates, time in stage, source quality) is quietly wrong.

Two rules prevent it. Scheduling and feedback states are attributes of a card, not columns: Technical with a Thursday date is one stage, not two. And if two adjacent stages share an owner and an exit question, merge them. Role-specific extras, like a second deep-dive for senior hires, belong inside the Technical stage as a different interview plan, not as a permanent column every junior candidate skips past.

Where does Recruitifly fit?

Recruitifly’s board is built for a lean stage set like this one, and its assistant, Fly, handles the chores that keep a lean board honest: it drafts candidate follow-ups in the candidate’s language, proposes interview slots and books them once you confirm, builds the pipeline report when you ask what is stuck this week, and runs saved automations for the routine transitions. Every write is propose then confirm: Fly prepares the change, you approve it, then it happens, which matches how a pipeline should work anyway, one accountable owner per decision. We are in private beta; if you want to pressure-test this stage set on a live role, talk to us.