Your Hiring Pipeline Is a Greenhouse Inbox

It is Tuesday at 11:47 PM. Your founder opens Greenhouse on her phone, scrolls past 84 new applicants for the senior backend role, taps through six profiles, replies to two recruiter forwards, and closes the app. The req has been open for 162 days. The 84 applicants joined the queue while she was in a Series B board meeting that afternoon. Six of them have GitHub commits matching the stack you ship on. None will hear back this week.

The head of engineering pulls the Greenhouse report Friday morning. The senior backend req shows 1,840 candidates in the funnel, 47 in screen stage, 12 in technical interview, 3 in onsite, and zero in offer. The pass-through rate from screen to technical sits at 26 percent. The pass-through rate from technical to onsite sits at 25 percent. The pass-through rate from onsite to offer reads zero, four months running. The req's age line on the dashboard reads 162 days.

Three weeks later, two of the six GitHub-commit matches from Tuesday night accept offers at a competitor. One was the candidate your CTO would have hired in the first 20 minutes of a call. The recruiter you outsourced the role to in March is still sending you the same eight profiles your engineering manager rejected in April. The req stays open. The product roadmap slips a quarter on the backend feature that needed the hire to land.

Recruiting operations is a function. Most Series A and B teams have not staffed it. The function lives in the gap between the founder triaging Greenhouse on her phone at midnight, the head of engineering reading the Friday report, the contract recruiter sending the same eight profiles, and the hiring manager who only sees a candidate after week three. On the org chart, it sits inside People Ops or talent. In the calendar, it eats 18 to 30 hours of founder and hiring-manager time per req and produces one hire every five to seven months.

The Greenhouse inbox nobody triages

Pull every open requisition in your Greenhouse or Ashby instance. Log the days-open, the candidates-in-funnel count, the source mix, the pass-through rate at each stage, and the last activity timestamp on the top 20 candidates in screen stage. Most teams find that reqs older than 90 days are sitting on 600 to 2,000 candidates in the funnel, with 40 to 80 in screen stage and a last activity timestamp older than 14 days on more than half. The hire the team was supposed to land in Q1 is now a Q4 conversation.

Walk the file. The senior backend candidate who applied on day 41 had three commits to a relevant open-source library and a four-year run at a company on your wishlist. Her application sits at "new" on day 162. The candidate who got promoted to screen on day 88 had a recruiter on the call who asked four generic questions and wrote a four-line summary. The hiring manager read it on day 91 and moved on. The candidate took a competitor offer on day 96.

The team that should own this knows it is broken. The founder watches the req age line climb and books another time slot to triage Greenhouse on Sunday night. The head of engineering watches the technical interview funnel stay at 12 candidates and asks the recruiter for more profiles. The recruiter pulls another LinkedIn search and forwards the same eight profiles. The function sits unstaffed while the queue keeps growing.

The cost shows up as a product roadmap that slips a quarter on every senior IC role, a founder who spends six hours a week on candidate triage instead of customer calls, and an offer-accept rate that lands below 60 percent because the best candidates went cold at day six. The CFO writes the open-req count into the board pack. The CTO writes the slip into the next planning cycle. The real read is that recruiting ops is the function that sits on top of People, engineering, and the founder at the same time, and no single role owns the candidate at day six.

Hiring a recruiter is the slow answer

The textbook fix is an in-house technical recruiter or a head of talent. Loaded comp on a senior tech recruiter in the US runs $140K to $190K a year. A head of talent runs $200K to $260K. Months one through three go to onboarding, calibrating the ICP with each hiring manager, learning the stack, and building the screening rubric. Months four through six are when the first two hires land and the funnel math starts to read differently on three of nine open reqs.

The output is good on the reqs the recruiter gets to. The other six reqs still sit in the founder's Sunday-night triage queue. The senior tech recruiter becomes the bottleneck on every screen call because each one runs 45 minutes plus 20 minutes of writeup. On 240 inbound applicants a week across nine reqs, the recruiter clears 18 screens. The 222 other candidates wait two to four weeks for a first touch.

The fractional version is faster to start and stops at the same wall. Twelve to twenty thousand a month buys a contract recruiter who runs 30 to 40 screens a week and ships profiles into your Greenhouse. The profiles match the JD on paper. Half of them get rejected by the hiring manager in week two because the JD on paper does not match the bar the team set out to hire against. The recruiter rebuilds the search query and ships the same shape of profile a week later.

Both versions assume the work is human bottleneck work. Read every new application within 30 minutes of submission. Score it against the JD, the top performers on the team, the closed-won candidate profiles from the last 24 months, the GitHub commit history, the conference talks, the open-source contributions, and the company tenure pattern. Draft a personal outreach note in the hiring manager's voice and send it. Schedule the screen and run the structured call.

Code the answers against the rubric. Brief the hiring manager the same afternoon. Loop in the technical interviewer with a tailored question set tied to the gaps the screen surfaced. On 240 applicants a week across nine reqs, that is 120 to 180 hours of senior work. No recruiter clears that pile and also sources passive candidates on LinkedIn.

What a fractional AI recruiting function does

Hand the open reqs, the Greenhouse instance, the JDs, the team's top-performer profiles, the closed-won and closed-lost candidate history, the hiring manager's calendar, the LinkedIn Recruiter seat, and the GitHub and conference-talk feeds to a fractional AI agent that runs the recruiting cycle on a per-candidate basis. The agent does the work an in-house recruiter, a sourcing analyst, and a scheduling coordinator would do together. The cadence is per-applicant on screen and brief, daily on passive sourcing, weekly on funnel health, monthly on JD calibration. The founder stops triaging Greenhouse on Sunday night.

Inbound triaged in 30 minutes, not 30 days. The agent reads every new application within half an hour of submission, scores it against the JD and the top-performer profile, pulls the candidate's public footprint across GitHub, LinkedIn, conference talks, and the company's last three roles, and routes the top 5 percent to the hiring manager with a one-page brief. The 84 applicants from Tuesday night get a real read by Wednesday morning. The six GitHub-commit matches get a personal outreach note from the hiring manager's inbox by Wednesday afternoon.

Outreach drafted in the hiring manager's voice, not in a recruiter template. The agent reads the last six months of the hiring manager's outbound replies and learns the voice. The note references the candidate's commit, the conference talk, or the open-source contribution that maps to your stack. The reply rate on cold outreach climbs from 8 percent to 24 to 32 percent on candidates who match the ICP and have a public signal worth referencing.

Screen call run on a structured rubric, coded against the funnel. The agent runs a 25-minute structured screen on the candidates who match the ICP and book the slot, codes the answers against the rubric the hiring manager calibrated in week one, and writes a one-page brief that lands in the hiring manager's inbox the same afternoon. The brief reads against the same signal set the discovery brief ships on the sales side. The hiring manager walks into the technical loop already knowing what to probe.

Passive sourcing run daily, not on a Friday afternoon. The agent pulls the LinkedIn Recruiter, GitHub, and conference-talk feeds against the ICP profile and surfaces 30 to 50 passive candidates a week per req. The hiring manager reviews the queue between meetings, picks the top 5, and the agent drafts the outreach. Sourcing stops being the work the contract recruiter does once a quarter and starts being a daily motion.

Funnel health reported weekly, not on a Greenhouse dashboard. The agent rolls up the funnel math across reqs by source, by stage, by hiring manager, and by drop-off reason. The req that lost three onsite candidates in four weeks on the same take-home assignment gets flagged. The JD that pulled 1,840 applicants and 12 technical interviews gets rewritten against the patterns from the closed-won candidate profiles. The Friday Greenhouse report stops reading the same way for the fourth month running.

The unit economics of a six-month open req

A company running nine open reqs across engineering, product, and GTM at $180K average loaded comp is sitting on $1.6M of unfilled output a year, before counting the roadmap slip on every product feature the role was supposed to ship. Add the contract recruiter at $15K a month for two of the reqs, the LinkedIn Recruiter seat at $14K a year, the Greenhouse seat at $9K a year, and the founder and hiring-manager time at 22 hours a week against a $400 fully-loaded hour. The run rate on the recruiting function lands at $720K to $880K a year against four hires in twelve months. The CFO sees the math on the cost-per-hire slide and the head of engineering writes the slip into next quarter.

Layer in the offer-accept rate line. Candidates who get a first touch on day six accept at 78 percent. Candidates who get a first touch on day 22 accept at 41 percent. The team running a 22-day average time-to-first-touch loses one in three offers. The CFO sees the math on the offer-accept slide. The head of talent writes the comp band wider. The accept rate moves four points and the hiring math gets worse on the cost-per-hire line.

A 14-day sprint to stand up the agent runs in the low to mid five figures. Ongoing cost lands closer to one senior contractor than an in-house talent team. Time-to-first-touch on inbound drops from 22 days to 8 hours. Pass-through from screen to technical climbs from 26 percent to 38 to 44 percent because the candidates getting through screen clear the bar. The founder's Sunday-night Greenhouse hour goes back into customer calls. Function, not headcount.

The harder number to price is the compounding line. A company that hires the right senior backend engineer in 90 days instead of 162 days ships the backend feature one quarter earlier and books the deal that needed it in Q3 instead of Q1 next year. A company that hires the wrong one and rolls them out in month four pays the loaded comp twice and slips the roadmap another quarter. The agent reads the structured screen against the closed-won candidate profile and flags the gap before the offer goes out.

What changes after the sprint

Picture the same Tuesday 11:47 PM moment, fourteen days after the sprint ships. The 84 new applicants on the senior backend role land in Greenhouse. The agent reads each one within 30 minutes, scores against the JD and the top-performer profile, pulls the GitHub footprint, and routes the six matches to the head of engineering's inbox with one-page briefs by midnight. Outreach drafts in the head of engineering's voice queue up against the four candidates who have not yet applied through the funnel.

By Wednesday at 9 AM, the head of engineering reviews the six briefs in 12 minutes, approves outreach on four, and the agent sends the personal notes through her inbox. Three reply by Thursday. Two book a 25-minute structured screen for Monday. The screen briefs land in the head of engineering's queue by Monday afternoon. The technical interview goes on the calendar for Wednesday with a rubric tied to the role. The offer goes out two weeks after the candidate first opened the application.

If your last quarter's recruiting motion lived in a Greenhouse inbox the founder triaged on Sunday night and a contract recruiter who shipped the same eight profiles every week, the version where every inbound applicant gets a real read in 30 minutes and the top 5 percent get a personal note from the hiring manager's inbox is fourteen days away. Recruiting is a function. You can hire against it, you can buy another contract recruiter for it, or you can scope a sprint and have it running this month. The work is the same. The math is not.