Methodology

Axis	What it measures	Source	Secondary weight	Primary weight
Academic	Mean of latest NCEA L1 + L2 + UE achievement %, weighted 0.4 · 0.4 · 0.2	Education Counts	35%	—
Outcomes	% of leavers to University + Te Pūkenga + Industry Training (counts normalised to %)	Education Counts	25%	—
Retention	% of Y9 cohort still enrolled at Y13	Education Counts	15%	—
Wellbeing	100 − (standdowns/1k + 2× suspensions/1k + 5× exclusions/1k), floored at 0. Small primaries with no rows = "no incidents reported" (100)	Education Counts	12%	35%
Trajectory	5-year change in total roll, scaled −40% → 0, 0% → 50, +40% → 100	Education Counts roll history	8%	25%
ERO review	Composite score (0–100) from the most recent Education Review Office report, AI-extracted from the narrative	ERO via Wayback archive	5%	40%

The composite is the weighted average of the axes that have data; weights renormalise when one or more axes are missing. Schools with roll under 30 don't get a rollup score (per-axis data still shows on the profile) because the rollup would be dominated by noise.

Synthetic ERO placeholder

About 40% of NZ schools (mostly small primaries) don't yet have an ERO composite in our dataset. Without one, the composite would collapse to wellbeing + trajectory only — both of which saturate at 0 or 100 for small schools and polarise the result.

To keep every school comparable, we substitute the national-average ERO score (~76/100) as a placeholder for any school missing an ERO review. This is flagged on the school's profile with an amber banner ("ERO review not in our dataset") AND on the ERO axis itself with a small est badge. When a real ERO review is ingested, the placeholder is replaced and the score may move materially.

Absolute scoring + per-axis peer context

The headline composite is on an absolute national scale — Auckland Grammar scores 80/100 because its underlying numbers are objectively strong, not because it's mid-pack among elite peers. Alongside each axis we show a (p47) style peer-percentile suffix indicating where the school sits within its EQI band (12 socio-economic bands across NZ). The peer note is contextual — the headline number is not adjusted by it.

What the score is NOT

• Not an official rating. ERO and the Ministry of Education don't publish school rankings. This composite is our derivation from open data.
• Not directly comparable across school types. Primary schools are scored on 3 axes (wellbeing + trajectory + ERO); secondaries on 6 (adding NCEA-derived academic + outcomes + retention). A primary at 90 and a secondary at 90 aren't measuring the same thing.
• Not a prediction. It describes the school as at the most recent data refresh. Education Counts data is annual; ERO reviews are quarterly.

Trajectory

When a school has more than one ERO review on file, we compute the year-over-year delta:

• score delta ≥ +8 → improving
• score delta ≤ −8 → declining
• otherwise → stable

Source of every score

Every input on the composite has a source URL on the school's profile. If you think a number doesn't reflect reality, tell us — we'll re-read and update.

Different rankings for different questions

The composite is one number, but it's not the only useful ranking. The /rankings page has quick-view preset pills for the common parent intents:

• Best overall — the composite default, secondary schools tab
• 🎓 Academic rigour — sorts by University Entrance achievement, secondary tab. Surfaces the schools traditionally associated with strong academic outcomes
• NCEA Level 2 pass — sorts by NCEA L2 achievement
• 🛡 Best behaviour — sorts by stand-down rate (low first), across all school types
• 📈 Highest stay-on rate — sorts by Y9→Y13 retention, secondary tab

One click sets the bucket + sort. Use the preset that matches your question instead of trying to read everything off the composite alone.

Known limits of the current methodology

We're up front about the gaps because parents and journalists deserve to know:

• Synthetic ERO is a placeholder, not a guess. It pulls schools toward the national median when we don't yet have their real review. Coverage is improving every refresh — the goal is to retire the placeholder.
• Cross-type rankings dominate small primaries. Small primaries with low/no incidents and growing rolls can hit 90+ on the composite even when their underlying data is thin. Use the /best-schools/[region]/[type]/ pages for type-segmented rankings that compare like-with-like.
• Peer-percentile context only when the band has ≥10 comparable schools. Schools with unusual EQI levels (especially extreme high or low) may have sparse peer comparison.
• Specialist + small schools at the extremes. The composite isn't designed for Specialist Schools, Teen Parent Units, Activity Centres or Correspondence — their educational mission is different from mainstream schools. Take their scores with extra caution.

We re-evaluate the methodology after every major data refresh. Substantive changes are noted on the contact page along with the date they took effect.

4. Address → school matching

We link every school to its suburb so we can show suburb safety and demographics in context. Matching uses (in order of preference):

1. The school's exact street address, where we have it

2. Geo-nearest suburb to the school's latitude/longitude

3. Name-similarity with the suburb of record

Where we can't make a confident match, we don't display the suburb context.

The MoE publishes enrolment-zone polygons separately. Until we have those zones licensed for every school, our "find by address" feature returns the schools nearest to your address by distance, not strictly by zone. That's clearly labelled on the Find by address page.

5. What we publish about people, and what we don't

About schools (institutions): we publish everything that's in a primary source — ERO findings, audited financials, statutory-manager appointments, roll, NCEA, attendance. These are facts about the institution.

About named individuals: we publish only factual, public-role information from primary sources:

• The principal's name and appointment date, from the school's own website
• ERO directors who signed reviews (a verifiable factual count)
• Board members named in audited annual reports

We never publish:

• Evaluative judgments about named individuals ("strict reviewer", "underperforming principal", etc.) even if statistically supported
• Anything that combines a named person with a negative outcome
• Reviews of named individuals — only of the school

Verbatim quotes from ERO reports are checked character-by-character against the source document before publication. If a quote doesn't exactly match the source, we don't show it.

If you're named on this site and believe a claim about you is wrong or defamatory, contact us. We respond within five working days under the NZ Privacy Act 2020.

6. How fresh the data is

Every fact has an "as at" date stamped on it. Refresh cadence varies by source: