✓
Action completed!

Greenhouse Open Jobs & Metadata Health

WezOps review of 87 open roles on the Dataiku Greenhouse board (board token dataiku, snapshot 2025-11-21).
Talent Acquisition Operations
Global Greenhouse Config
Metadata ≈ system reliability
What this page is
A live picture of your open demand – and where the data model quietly works against you.
Everything here is derived from your public Greenhouse job feed only. No stages, no candidates – just the job catalog and its metadata.
Open roles
0
Remote conflicts
0%
Evergreen / pipeline
0%
Goal: make recruiter performance, time-to-fill and funnel analytics trustworthy by treating job metadata like a product.

Snapshot of open demand Where the heat is, globally

We are only looking at open jobs from the public careers site – no closed reqs, no historical stages. The picture is GTM-heavy, global, and already shows where the metadata model will block clean TA reporting.

Open jobs
0
Avg age ≈ 122 days. 72% are < 90 days; a handful have been live > 1 year.
Region mix (by location)
EMEA · Americas · APAC
~42% EMEA, 33% Americas, 22% APAC, 2% other.
Evergreen / pipeline
0
Reqs with PIPE / EVG in the ID. Avg age ≈ 402 days.
Metadata health
Strong story, noisy schema
Demand is clear – GTM + AI – but remote flags conflict with location, and "N/A" placeholders exist.

Where you are hiring By department & region

GTM (Sales, Solution Engineering, Revenue Operations) drives about two-thirds of demand. R&D hiring is selective – primarily senior AI/ML roles.

Department breakdown (top 5)
Chart A
Two-thirds of headcount sits in Sales, Solution Engineering, and RevOps. R&D hiring is selective—focused on senior AI/ML roles.
  • 14 distinct GH departments sit behind 87 open roles – just enough that "Other" bucketing matters.
  • 8 departments have ≤2 roles; they're often grouped under Strategy & Ops or G&A for external facing.
Region breakdown (by location string)
Chart B
The majority of your open demand is EMEA-based, followed by the Americas and a moderate APAC footprint.
  • 40+ unique location strings in the feed – many multi-city blobs that make filtering tricky.
  • We'd normalise these into Country → City → Hub for cleaner dashboards and recruiter workloads.

Evergreen & pipeline reqs Always-open vs. active headcount

Evergreen and pipeline roles (identified by PIPE / EVG in the req ID) are great for talent pooling—but if dashboards mix them with real headcount, every metric drifts.

Evergreen share of the catalog
Chart C
About 18% of live roles are evergreen / pipeline. That's high enough to seriously skew time-to-fill and load-balancing reports.
  • Average age of an evergreen req is ~402 days—vs ~59 days for standard headcount.
  • Solution Engineering and Sales account for most of these; 2 have been open > 3.5 years.
Impact on funnel metrics
Chart D
If an unsegmented dashboard asks "what is our average time-to-fill?", the answer will be months longer than reality because it mixes in evergreen roles.
  • Even a handful of multi-year reqs can double your reported average.
  • Solution: tag evergreen programmatically (GH custom field or a naming rule), exclude from standard dashboards, report separately.

Remote metadata layer Conflicting signals

Remote work is encoded simultaneously in location strings, offices and a Remote? custom field. They disagree more often than you'd like.

Remote vs on-site mix (Remote? field)
Chart E
The intent is clear: a third of the catalog is remote-friendly. The question is whether downstream reports can trust this flag.
  • Roughly one-third of roles are remote-friendly at the metadata level.
  • This mix is strategically important (talent pools, salary bands, recruiter coverage).
  • We'd make Remote? the single source of truth and derive everything else from it.
Remote metadata conflicts
Chart F
When three fields all try to describe "remote vs on-site", they inevitably drift apart. That is exactly what we see in the snapshot.
  • ~24% of roles have Remote? values that conflict with the location string.
  • ~16% conflict with office settings.
  • We'd clean the current catalog, then enforce rules so one field drives all remote reporting.
Location string quality
Chart G
The candidate-facing locations are good; the analytics-facing structure behind them needs one more layer.
  • 40+ distinct location strings, including multi-country blobs.
  • We'd treat these as display labels, and derive reporting from structured Country / City / Region fields.

Taxonomy & next steps From "interesting" to "operational"

With only the public job feed, we can already see enough to propose a concrete, low-friction plan: clean the schema, then wire it into recruiter performance and data-quality monitoring.

Department alignment (GH vs External-Facing)
Taxonomy
Two parallel taxonomies describe the same work: Greenhouse Department and External-Facing Department. They agree most of the time – and disagree just enough to matter.
  • Examples: Sales Enablement / Enterprise Data & Analytics roles exposed as Strategy & Operations externally.
  • At least 1 in 7 roles changes bucket depending on which field a dashboard uses.
  • We'd define a single, governed mapping and enforce it via templates and validation.
What WezOps would do
Proposal
A practical, two-phase plan: fix the schema, then turn it into a monitoring and performance layer Phil's team can actually use.
  • Phase 1 – Job & metadata cleanup
    • Redesign & lock the department / sub-department / location / remote model.
    • Normalize current open roles (remote conflicts, multi-location strings, evergreen flags).
    • Introduce a small, opinionated set of data-quality KPIs for the job catalog.
  • Phase 2 – Recruiter performance & data-quality dashboards
    • Once connected to internal Greenhouse data (stages, offers, owners), layer on time-to-fill, funnel conversion, and workload metrics.
    • All indexed by cleaned metadata – so "Sales, EMEA, remote" means the same thing everywhere.
  • Outcome: a TA system where dashboards reflect how the team actually works – and where data quality is monitored like a product.