Hiring Signals

Other tools hand you job listings and role counts. This hands you the conclusion.

Give it a list of companies. It reads their public job boards and tells you what each one is about to do.

Reddit · 188 open roles
  ↳ Exactly one legal role on the entire board, and it is a
    Director of Ads Policy.
    You do not hire a policy director for an ads business you
    are not about to scale.

Duolingo · 64 open roles
  ↳ 5 leadership openings in marketing, plus single roles in
    Japan, the UK and China while 78% of the board sits in the US.
    Three markets opening at once.

Databricks · 809 open roles
  ↳ 19 director-and-above searches in sales alone.
    A revenue org this senior-heavy is being restructured, and
    every incumbent vendor is up for review while it is.

Those are real, from one run. Not examples.

---

Why job boards leak the truth

Marketing copy is what a company wants to be. Headcount is what it already paid for.

A company hires for a market before it enters it, for a product before it ships it, and for a function before it admits it needs one. All of that is public, sitting on a job board, in a format nobody bothers to read at scale.

• First DevOps hire → they outgrew their stack. Cloud, observability and CI/CD budget just appeared.
• Three sales directors at once → new revenue org. Every incumbent vendor is now up for review.
• One role in a country where they have none → market entry, one or two quarters early.
• A board that is suddenly all fresh postings → headcount just unlocked. This is the least constrained their budget will be all year.

---

Eight signals

GTM_PUSH	Revenue hiring outweighs product hiring. Product is ready, they are buying distribution.
GEO_EXPANSION	Beachhead roles outside the home country. The earliest visible sign of market entry.
FIRST_HIRE_IN_FUNCTION	A function being created from nothing.
AI_CAPABILITY_BUILD	Concentrated ML hiring. They are building AI, not buying it.
INFRA_SHIFT	A platform function being stood up. The original stack stopped working.
LEADERSHIP_BUILD	Director-and-above searches concentrated in one function. A layer is being created.
HIRING_SURGE	Most of the board is new. Headcount just unlocked.
NET_NEW_ROLES	Diffed against our own last crawl, not against posting dates.

Every signal ships with a severity, a confidence level, the reason it matters, and the exact roles that triggered it. No black box.

---

The hard part is shutting up

A signal that fires on every company is not a signal.

Three rounds of calibration, each triggered by measuring how often a signal fired rather than by reading the ones it produced.

The first returned at least one signal for every single company tested. AI_CAPABILITY_BUILD was firing on 77% of boards, which in 2026 is a description of the technology industry.

The third ran 153 recognisable companies and found two more of the same disease. LEADERSHIP_BUILD fired on 57%, because counting director-level openings cannot tell this function is acquiring a leadership layer from this function is large and hires seniors like any large function does — five sales directors alongside 107 other sales roles is a big sales org behaving normally. INFRA_SHIFT fired on 34%, because essentially every company with 15–150 open roles has one to three infrastructure roles open at any moment.

Where it lands now, measured across 112 real boards and 11,969 open roles:

HIRING_SURGE               21% of companies
GEO_EXPANSION              16%
GTM_PUSH                   14%
LEADERSHIP_BUILD           13%
FIRST_HIRE_IN_FUNCTION      8%
AI_CAPABILITY_BUILD         4%   (was 77% before calibration)
INFRA_SHIFT                 3%

companies producing any signal   55%
average signals per company      0.87
hard ceiling per company         5

Almost half of all companies return nothing, and that is the product working. A tool that finds something to say about every account has told you nothing about any of them.

The test suite contains a deliberately boring control company required to produce zero signals, and eight other fixtures that pin specific failures the live runs exposed — including one whose whole job is to prove a tightened signal can still fire at all. Fifteen checks. If any of them fails, the build is wrong.

---

You pay for answers, not rows

	Charges
Job-listing tools on this store	per job returned, empty or not — a 808-role company runs $1.60 – $9.70
Per-company tools	a flat fee for every company, whether or not it had anything to say
Hiring Signals	$0.08 per company that produced a signal. Quiet companies are free.

This is not generosity. If silence earned money, we would have an incentive to loosen thresholds and bury you in noise. It does not, so we do not.

---

Boring on purpose

Reads the public, documented job-board APIs that Greenhouse, Lever, Ashby, Workable, Workday, SmartRecruiters and Recruitee publish so companies can embed listings on their own careers pages. Third-party use is the intended use.

No browser automation. No fingerprint evasion. No proxy rotation. No anti-bot arms race to eventually lose.

Zero third-party dependencies. Every line is Python standard library. Nothing in the tree can break this on somebody else's release schedule.

Unglamorous decisions, all of them. They are the difference between a tool that works today and one that still works in six months.

Three known-hard cases are handled explicitly rather than papered over:

Workday serves newest-first. Read the first thousand roles of a two-thousand-role board and you have read the fresh half — then reported "100% of this board opened in the last month". Truncated boards report the true total, how much was actually read, and return null for freshness rather than a confident falsehood.

An ATS migration re-dates an entire board at once. One company showed "103 roles opened in the last 30 days, 100% of the board" — from 84 postings published within minutes of each other on a single morning. That is an export, not a hiring plan, and a burst on one calendar day now suppresses the claim.

SmartRecruiters returns 200 OK with zero results for a company that does not exist, where every other platform returns 404. Counting that as a hit would attach an empty board to any string at all.

---

Input

{
  "companies": [
    { "board": "stripe" },
    { "board": "figma", "name": "Figma" },
    { "board": "https://nvidia.wd5.myworkdayjobs.com/NVIDIAExternalCareerSite", "name": "NVIDIA" }
  ],
  "minSeverity": 50,
  "signalTypes": [],
  "includeJobs": false
}

Field
board	The company's slug on its ATS — usually just the company name. Seven platforms are covered: Greenhouse, Lever, Ashby, Workable, SmartRecruiters, Recruitee and Workday. For Workday, paste the careers URL instead (https://nvidia.wd5.myworkdayjobs.com/NVIDIAExternalCareerSite); it is recognised on sight.
ats	Optional. Leave it out and the platform is detected, then remembered for next time.
name	Optional label for the output.
minSeverity	Floor for what gets returned. Default 50. Raise it for only the loudest calls.
signalTypes	Restrict to specific types. Empty means all.
includeJobs	Attach every classified role alongside the signals.

Auto-detection resolved 112 of 153 company names with no configuration at all — and the resolved platform is cached, so a repeat run costs one request per company.

---

Output

One record per company.

{
  "company": "Databricks",
  "total_open_roles": 808,
  "roles_posted_last_30d": 159,
  "hiring_velocity_30d": 0.197,
  "leadership_openings": 53,
  "function_mix": { "engineering": 300, "sales": 121, "ai_ml": 78 },
  "country_mix": { "US": 500, "GB": 60, "DE": 40 },
  "signals": [
    {
      "type": "LEADERSHIP_BUILD",
      "headline": "19 leadership openings in sales",
      "why_it_matters": "Several director-and-above searches in one function means an org is being built or rebuilt. Both are moments when incumbent vendors get re-evaluated.",
      "severity": 90,
      "confidence": "high",
      "evidence": [
        { "title": "Director, Enterprise Sales", "location": "New York, United States", "url": "https://..." }
      ]
    }
  ]
}

---

Who buys this

Sales teams who would rather call an account the week its budget unlocked than work down a static list. Investors watching portfolio and target companies for expansion, pivots and org build-outs. Competitive intelligence teams who want a rival's market entry a quarter before the press release. Recruiters looking for teams being built, before the roles are on every board.

---

Working examples

Two runnable scripts, standard library only, nothing to install: github.com/SignalVane/hiring-signals-examples

• run_and_export.py — run over a company list, write one CSV row per signal. The shape you want for a spreadsheet or a CRM import.
• watch_target_accounts.py — run on a schedule and print only what changed. You hear about an account the week it crosses a line it had not crossed before.

The same repo has FINDINGS.md: a full pass over 112 company boards and 11,969 open roles, including the loudest signal of that run — which turned out to be wrong, and why.

---

Run it weekly

NET_NEW_ROLES and the history-confirmed signals sharpen every run, because the Actor diffs against its own previous crawl instead of trusting posting dates. One run is a snapshot. A schedule is change detection.