Recruiting decisions, findings, and tips

What the Findings tab is for

Recruiting throws off a lot of small facts: a posting with no applicants in ten days, a strong resume that just landed, an interview that got scheduled and never followed up on, a warm candidate whose old role just reopened. Read one at a time, these are noise. Verinode's recruiting agent watches all of them at once, and when two or more point at the same underlying situation, it clusters them into a single decision with a recommended verdict, a plain-language reason, and what it costs you to sit on it. That clustered view, plus a lighter-weight stream of one-off tips, is what lives on the Findings tab.

Verinode does not make hiring decisions. It reads your requisitions, applicants, interview steps, and talent pool, weighs the pattern, and hands you a recommendation with the evidence attached. You review it and decide.

Where to find it

Open Recruiting from the sidebar at /recruiting. The section is a card slider with seven tabs across the top:

Findings · Open Roles · Applicants · Pipeline · Talent Pool · Hires + Retention · Benchmarks

Findings is the first tab and the one this article covers. Top to bottom, it stacks three things: a Tips strip, a Decisions list (what this article calls decision stacks), and a Decisions grid below that groups every open recruiting decision as clickable tiles. The other six tabs (Open Roles, Applicants, Pipeline, Talent Pool, Hires + Retention, Benchmarks) hold the underlying records and reporting that feed Findings; they get their own coverage elsewhere in the Recruiting section of this help center.

Where the signals come from

Before anything reaches Findings, Verinode's recruiting detector runs across four kinds of records and raises atomic signals on each:

• Per requisition: stalled (posted over 30 days with fewer than three applicants), pipeline thin (fewer than three applicants after two weeks), pipeline strong (eight or more applicants averaging a healthy score, meaning you can afford to be selective), closing urgent (inside two weeks of your target close date with no offer out), and JD stale (posted over 60 days).
• Per applicant: top tier and promising (score thresholds on your screening rubric), cadence stale (over four days since last contact), ghosted (over ten days silent after being advanced or scored), and interview stalled (an interview step still pending 14+ days after it was scheduled).
• Per talent-pool candidate: a role opening that matches their stated interest, and nurture check-ins that have gone overdue.
• Portfolio-level (not tied to one record): a dry pipeline across every open role, offer acceptance sliding below a healthy bar, 90-day retention running low, and a referral channel that has gone quiet.

Each of these is a raw fact. On its own, "12 days since last contact" or "posted 65 days ago" doesn't tell you what to do. That's what the orchestrator is for.

How the decision orchestrator clusters signals into a verdict

The recruiting decision orchestrator runs after the detector pass. It only fires when at least two open atomic signals are on the table together, on the theory that a single signal is a note to self, and two or more compounding signals are a decision. It reads the cluster alongside your current recruiting snapshot (open requisitions, time-to-hire, offer acceptance, talent-pool depth, retention, and where you sit against peer time-to-fill and the coming seasonal hiring window), reasons over it, and returns a verdict.

The possible verdicts are:

• Advance the candidate to the next step
• Decline the candidate
• Offer the role
• Sub out the pipeline (pivot to a different sourcing approach)
• Defer the decision
• Revise JD, refresh the job description and re-post
• Activate talent pool, re-engage a warm or engaged candidate
• Accept risk, proceed while acknowledging a known gap
• Operator review required, a hard stop that hands you the evidence instead of a recommendation

Every decision comes with an action title (what to do), a gain headline (why it matters), and a consequence headline (the cost of leaving it alone). It also names the specialist agent it would route to if you act on it, for example the requisition-diagnoser for a stalled posting or the compensation-benchmarker for a comp-gap applicant, and it links back to the atomic signals it superseded, so nothing disappears, it's folded into the cluster instead.

Verinode never repeats a decision it's already shown you for the same underlying situation: it hashes the open signals and your current recruiting snapshot together, and if that hash matches a decision already sitting open, it skips generating a duplicate.

The legal guardrail

One override sits above everything the model returns. If any applicant behind a cluster carries a flag for criminal-history disclosure (the "ban the box" category) or another protected-class signal, such as a resume mentioning salary history where that's restricted, the verdict is forced to Operator review required, no matter what the model recommended. The action title is rewritten to lead with "Legal review required," the decision is marked with a Legal Review badge in place of the usual verdict pill, and the consequence line carries Verinode's standing disclaimer: Verinode does not make hiring decisions, you review the evidence and decide. This override can't be reasoned around; it's applied in code after the model responds, not as a prompt instruction the model could ignore.

The Tips strip

Above the decision stacks sits a lighter strip labeled Tips, with a running count next to the label. These are one-off, single-signal notes rather than orchestrated clusters, tagged internally as recruiting tip signals. Each tip shows a headline, an optional rationale line underneath explaining why it surfaced, and an optional action line describing what to do about it. Up to six show at once. If there are no open tips, the strip is simply omitted, there's no empty-state message here, the section just doesn't render.

Decision stacks

Below the tips, the Decisions list shows every open orchestrator decision as a card. Each card shows:

• The headline (the action title)
• A verdict pill, reading the verdict name, or Legal Review when the guardrail above has been triggered (cards under legal review get a highlighted border in the Analyse signal color so they stand out from ordinary recommendations)
• The gain headline underneath, in muted text
• The consequence headline below that, in italics, spelling out what waiting costs you

If there are no open decisions, this list is also simply omitted, no placeholder copy.

Cost-of-inaction framing

The consequence headline is where Verinode states the cost of inaction, and it's worth understanding what kind of number sits behind it. Recruiting impact is modelled, not measured after the fact, so it's carried as a soft, estimated figure rather than a hard-recoverable dollar amount: a conditional clause ("if the seat stays unfilled through the next demand cycle," "if the requisition closes unfilled") paired with an estimate, not a guarantee.

For requisition-level signals specifically, Verinode also computes a net dollar picture behind the stall. It estimates the gross cost of the hire from the role's posted compensation range (or a conservative default when no range is set), and weighs that against a modelled recovery: the extra billed revenue the crew could realistically take on once the seat is filled, based on your own job history and a conservative assumption of roughly half an additional job per month from the new hire. When both sides of that math are available, the consequence headline reflects the net delta between them. When your job history isn't deep enough to model a cycle-time recovery, the signal falls back to a general clause about the cost of leaving the seat open.

Every decision also carries a numeric urgency in days when the model can estimate one. That urgency drives how urgently a decision reads: inside two weeks it surfaces as critical, inside a month as a warning, and further out as informational. A legal-review decision always reads as a warning regardless of its timeline, since the point is careful review, not an alarm.

Opening a decision

Every decision in the stacks list and the grid below it shares an agent plan: a first step stating the verdict and the gain headline, followed by up to three specialist consultations it's ready to hand off, each with the reason that specialist was routed in. If you act on a recruiting decision, that's the trail IQ follows, so the next step (draft a re-engagement email, revise a JD, size a comp offer) comes from a specialist that already has the context, not a blank slate.

The Decisions grid

Underneath the tips and stacks, the Findings tab renders the full set of open recruiting decisions again, this time as a grid of tiles, one tile per decision, using the same shared decision-tile component every other section's Findings tab uses (Jobs, Vendors, Clients, Team, Equipment, Safety, Compliance, Certifications, Reputation, and Margin all share it). Recruiting's tiles here are the compact, click-to-open kind: no inline Act / Not now / Ignore buttons, just a headline and enough context to recognize the decision. Click a tile to open it full-screen in the decision drill-in view, where arrows step you to the next or previous open decision without closing the view. See the decision workspace for how drill-in and execution works once a decision is open.

Empty state. When there are no open decisions, the grid reads:

No decisions for recruiting right now. As your agent finds patterns, they'll appear here.

The Take Action row

The highest-severity, most time-sensitive recruiting decisions don't only live on the Recruiting page. Verinode's platform-wide Take Action row, the same shared band that surfaces the most urgent decisions from every section on your IQ home screen, pulls from this same recruiting decision data. A decision that clears the urgency bar shows up there as a wider tile with an accent-colored rail down the left edge and a filled "Recommended" label, so you can act on a stalled requisition or a flagged legal-review case straight from Home without opening Recruiting first. Every tile that reaches Take Action is, by definition, something Verinode judged worth interrupting your day for, lower-priority findings stay on the section's own Findings tab until you get to them. Read the decision workspace for the full mechanics of how a decision moves from a tile to an executed plan.

The rest of the Recruiting card slider

The other six tabs hold the records Findings draws its signals from:

• Open Roles lists your requisitions. If you have none yet, it reads: "Open roles will appear here as you post them. Forward your Indeed or IICRC postings to your recruiting inbox to auto-land the requisition." See forwarding documents for how that inbox works.
• Applicants and Pipeline track candidates moving through your funnel.
• Talent Pool holds candidates you've declined with opt-in or added proactively, grouped as Engaged, Warm, and Cold. Empty, it reads: "Your talent pool populates as you decline promising applicants with opt-in, or add proactively-sourced candidates manually. Warm candidates get nurtured monthly."
• Hires + Retention tracks retention checkpoints on completed hires.
• Benchmarks compares your Time To Fill, Applicants Per Requisition, Offer Acceptance, and Retention at 90 and 365 days against peer values, each row showing your number beside the peer figure (a dash where either side has no data yet). Peer values only unlock once enough peer operators have contributed comparable data; see how benchmarks work and reading a benchmark for how that gating and comparison generally works across Verinode. Before you have hires with both a received date and a hire date plus at least one retention checkpoint, the tab reads: "Benchmarks load once you have hires with received_at + hired_at and at least one retention checkpoint. Peer values unlock when the cohort hits cohort-min," meaning the operator cohort needs to reach the platform's minimum size for that comparison to be statistically meaningful; Verinode never states the specific count.

None of your recruiting data is sold to carriers or any outside party. Contributing your (anonymized) hiring outcomes into the peer benchmark is what unlocks the comparison numbers on your own Benchmarks tab, the same trust exchange that runs across every Verinode benchmark.