Skip to content
d daycarenearme.co.nz

Methodology

Daycare Near Me is an independent directory of New Zealand's 4,375 licensed early childhood education services. We aggregate government-published data so families and operators can find, compare, and benchmark centres without scraping the source.

What we publish

Sources

  1. Ministry of Education ECE Directoryeducationcounts.govt.nz. Refreshed nightly upstream; we pull every change cycle. 49 fields per service. Roll data refreshed annually each July from the Annual ECE Census.
  2. Education Review Office (ERO)ero.govt.nz. We ingest every published ERO review (Education Review · Akarangi · Te Ara Huarau · Governing Body) for each licensed centre. Licensed CC-BY 4.0. Logos and emblems excluded from republication. See "How we grade ERO reports" below.
  3. Charities Services (Te Rātā Atawhai)charities.govt.nz. Open data API; we pull all 45,500+ registered charities and match by name/address to centre operators. Annual Returns provide revenue, expenses, employees, and govt vs non-govt income split.
  4. crimestats.co.nz — sibling Pangaea project; NZ Police-derived suburb safety scoring.
  5. IRD / MSD / parents.education.govt.nz — for the FamilyBoost + Childcare Subsidy + 20 Hours ECE calculator policy parameters.

How we match charities to centres

Two-pass approach in scripts/ingest/match-charities-to-centres.mjs + match-charities-by-address.mjs:

  1. Name fuzzy match using Jaccard token overlap (≥0.7 + suburb tie-break). Catches obvious cases like Geraldine Pre-School Charitable TrustGeraldine Preschool Incorporated.
  2. Address tie-break for unmatched ECE charities: normalise street + suburb, exact match. Catches cases where the charity's legal name doesn't share tokens with the centre's trading name.

We do not assert a match when confidence is low. Charity numbers, OrganisationIds, and source URLs are surfaced on every centre page so you can verify.

How we grade ERO reports

ERO publishes one categorical score per review — the overall placement on a 4-point scale: Not well placed, Requires further development, Well placed, Very well placed. Everything else in an ERO report is narrative prose.

On centre profiles we surface ERO's overall placement verbatim, as ERO's own judgement. We also derive a six-dimension letter-grade summary from the ERO prose — these dimension grades are our analysis, not ERO's scoring. Every dimension tile shows the verbatim quote it's based on with a link back to the canonical ERO review URL so you can verify.

The six dimensions

The letter scale

Four grades, mirroring ERO's overall 4-point scale so a Curriculum B reads consistently with a Well placed headline.

Guardrails

How we match centres to suburbs

Three-tier strategy (ports match-schools-to-suburbs.cjs from the SchoolsData Pangaea sibling): (A) exact suburb-name + region match, (B) name unique + geo nearest within 5km, (C) lat/lng nearest within 10km. 98% of centres resolved.

What we do NOT publish

Update cadence

API + LLM access

Every centre has a machine-readable companion at /api/centre/{slug}/facts.json and /markdown.md. The full URL index lives at /llms-full.txt. An MCP server at /.netlify/functions/mcp exposes lookup + search tools for AI agents — see /mcp-install.

Data sources & freshness (4 sources · last refresh 12 July 2026)

Service type, address, roll, ethnicity, Equity Index, 20 Hours, Takiwā, lat/lng

· Refresh: Nightly source · re-pulled on every deploy

Annual financial returns (community-based + umbrella orgs)

As of: 545 centres covered· Refresh: Annual filings · re-ingested with each refresh

Per-centre review reports + 6-dimension letter grades (curriculum / wellbeing / whānau / bicultural / health & safety / leadership)

As of: 10,663 reports / 4,194 centres · latest review 2026-05-18· Refresh: On review publication · CC-BY 4.0 attribution

Safety score, crime/1k, 2023 Census demographics

As of: 2023 Census· Refresh: Quarterly police data · 5-yearly census

Last build refresh: . See full methodology + data sources pages for licences, refresh cadences, and known gaps.