Methodology

How we source and review ETF data

ETFs.co.nz publishes reference data on every ETF accessible to NZ investors. This page documents where the numbers come from, how often we refresh them, and what kinds of claims we deliberately don't make.

Data sources

In source-priority order:

1. Issuer fact sheets — Smartshares, Kernel, Vanguard, State Street (SPDR), JPMorgan, Schwab, Invesco, BlackRock (iShares), ARK Invest. The TER, distribution yield, distribution frequency, and AUM on every `/etf/[ticker]/` page is taken from the issuer's most recent published fact sheet. When sources conflict (e.g. third-party data feeds disagree with the issuer), we cite the issuer.
2. Exchange filings — NZX (NZ-listed Smartshares + Kernel), NYSE Arca / NASDAQ (US-listed). Used for listing dates, ticker confirmation, distribution-frequency cross-check.
3. SEC EDGAR — for US-listed ETFs, the issuer's 10-K and N-CEN annual reports (AUM, share count, expense-ratio history).
4. IRD (Inland Revenue Department) — for NZ tax framing (FIF de minimis, FDR vs CV method, PIE PIR). We cite IRD pages directly; we don't interpret.
5. Reputable third-party data (etfdb.com, justETF, Morningstar) — used only to spot-check; never the primary source.

Per-ETF source-document links

Every /etf/[ticker]/ page carries a “Sources for this {ticker} data” block at the bottom with click-through links to the issuer's product page, the NZX listing page (for NZX-listed funds), the fund's fact sheet (PDF where available), and the formal Product Disclosure Statement. The full source-link map is maintained in src/data/etfs.ts as the SOURCES object and exposed via the getSources(ticker) helper.

This is the contract: every numeric claim on this site should be traceable from the rendered page back to an issuer or regulator document in at most two clicks, and where we have verified the figure against that document rather than merely linking it, the page says so. If a number doesn't trace, it's a bug — report it via the corrections link below.

What has actually been checked

Linking a document is not the same as having read it. As at 2026-09-05,36 of 66 expense ratios and30 of 66 distribution yields have been checked by machine against the issuer's own published data. The remaining 29 funds are labelled “not independently verified” on their own page, in as many words. We would rather say we have not checked something than imply we have.

The check runs weekly from scripts/verify-etf-data.mjs and covers the issuers that publish machine-readable data: Smart/Smartshares through the FMA-mandated quarterly Fund Update PDFs, and BlackRock, Vanguard and State Street through their own product data. Issuers with no such feed are reported as uncovered rather than skipped quietly. Trailing distribution yield is the harder half — Fund Updates carry fund charges and returns but not yield, so only issuers who publish a trailing figure directly can be verified at all.

Distribution yield: what we publish, and what we stopped publishing

Yield was the weakest figure on this site and it is worth saying exactly why. The FMA-mandated Fund Updates — the one regulated, fixed-layout, machine-readable New Zealand disclosure — carry total fund charges and returns and never distribution yield. So the figure had to come from somewhere else, and for most rows nobody could say where.

When we first checked the four rows that could be checked, three of them were wrong. That is not a basis for publishing the other fifty. So as at 2026-09-05 we did two things:every yield now comes from an issuer or exchange that publishes it, androws where nobody publishes one show no yield at all rather than a number we cannot stand behind. 30 of 66 funds carry a sourced yield; the rest show a dash. A dash means "not published anywhere we can cite", not "zero".

The second half matters more than the first. Yield is not one measure. Issuers publish at least three, and they are not comparable:

• Trailing 12-month distribution — what was actually paid out over the past year, over price. The measure most readers assume. Published by BlackRock and Global X.
• NZX gross dividend yield — the exchange's own figure for NZX-listed funds. Gross of imputation credits, so it reads higher than the equivalent US trailing figure and cannot be compared with one directly.
• SEC 30-day standardised yield — based on the last 30 days' income rather than the last year's distributions. Published by Vanguard. It moves faster and reads lower on an equity fund.

Pouring three measures into one undifferentiated column was the real defect — worse than the drift it hid, because a reader comparing two funds would have been comparing two different things without being told. Every fund page, API response and MCP result now names the measure alongside the number. Sorting a table by yield across different measures still is not meaningful, and the screener says so.

The script never edits canonical data. Changing a published TER is an editorial act; the check tells us which rows need one. Corrections found this way are applied with the issuer's figure and the row's review date bumped.

For an exhaustive plain-text dump of every fund + platform + tax fact this site publishes (useful for AI agents, downstream tools, and citation), see /llms-full.txt.

How we handle conflicting data

Sources sometimes disagree on the same fund. When that happens we apply this precedence:

1. Issuer document > third-party data feed. If Morningstar shows a different TER than Vanguard's fact sheet, Vanguard wins.
2. Most recent > older. Issuers republish fact sheets quarterly; we use whichever has the most recent “as at” date.
3. Formal disclosure > marketing page. The PDS / prospectus governs; the marketing page may simplify or round. We use the PDS for the legal TER.
4. NZX listing > secondary aggregator. For NZ-listed funds, the NZX market page is authoritative on ticker, listing date, and current units on issue.
5. IRD > everyone for NZ tax framing. We cite IRD pages directly and don't interpret — your accountant interprets.

Refresh cadence

How “best of” lists are constructed

Selections are editorial, based on publicly available data:

• Total expense ratio (lower is better, all else equal)
• Distribution yield where the page topic is income
• AUM as a survival / liquidity proxy (we deprioritise funds < US$50M AUM)
• Distribution frequency where it materially affects use case
• NZ-investor accessibility — PIE-tax NZ-domiciled funds get an explicit callout vs FIF-eligible US-domiciled funds
• Editorial judgment on category fit (e.g. JEPI's covered-call strategy belongs in "income" not "growth")

No issuer pays for placement, ranking, or inclusion on any "best of" list. See our about page for the no-commercial-relationship pledge.

Platform reviews — editorial scoring

Each `/platform/*` review carries a 1-5 editorial score across five criteria. The overall rating is the unweighted mean rounded to one decimal. Scores are editorial judgement based on the platform’s public documentation and our hands-on testing as at the review date — they are not a recommendation, certification, or endorsement, and they do not constitute personalised financial advice.

Criteria (1 = poor, 3 = adequate, 5 = strong):

• Fees & FX — all-in cost at typical NZ retail order size (commissions, FX margin, subscription fees, custody fees).
• Market access — breadth of markets supported (NZ, AU, US, global) and instrument coverage (shares, ETFs, options, futures).
• NZ tax fit — W-8BEN handling, IR3-friendly tax statements, PIE coverage, FIF disclosure support.
• UX & onboarding — NZ-citizen ID flow, mobile and web app quality, account setup friction.
• Support & operations — NZ-friendly support hours, channel options, account custody arrangements.

Each score is justified by a one-line note traceable to a verifiable feature or fee schedule. The methodology + raw scores ship as `src/lib/platform-reviews.ts` in the open repo. Scores are reviewed at least quarterly — corrections are dated and footnoted on the affected page.

We do not accept payment, advertising, affiliate revenue, or content sponsorship from any platform we review. See the disclosure statement.

What we don’t publish

• Forward-looking forecasts — no return predictions, no "next quarter winner" lists, no market-direction calls. NZ Financial Markets Authority restricts who can publish forward-looking investment commentary; we are not licensed to do so.
• Personalised recommendations — nothing on this site says "you should buy X". For personalised advice, see a registered NZ financial adviser.
• Live market quotes — all reference data is point-in-time as at the page’s last-reviewed date. Always check the issuer's live source before transacting.
• Performance projections — hypothetical "$10K invested in 1995 would now be $X" comparisons may appear in narrative copy as historical fact, but are never extrapolated forward.
• Star ratings without methodology — we don't rate platforms or ETFs with star ratings; if we ever do, the methodology + sample size will be on this page.

Where to get qualified market commentary

For forward-looking views, NZ investors typically use:

• A licensed financial adviser — Find one via FMA's adviser search or FinanceAdvisersNZ.co.nz.
• The fund issuer's own commentary — Smartshares, Kernel, Vanguard, JPMorgan etc. publish quarterly market notes and fund-level commentary on their websites.
• Sorted.org.nz (Te Ara Ahunga Ora Retirement Commission) for general personal-finance education, no commercial bias.
• NZ business journalism — Stuff Business, NZ Herald Business, Interest.co.nz cover NZ-specific investment context.

Editorial language

Phrases we use:

• "appears to", "may indicate", "based on available data", "historically", "as at [date]"
• "often suited to" (for investor-profile descriptors), "less commonly used by"
• "data shows", "issuer reports", "AR discloses"

Phrases we avoid:

• "premier", "leading", "most comprehensive", "trusted platform"
• "guaranteed", "risk-free", "fail-safe", "sure thing"
• "we recommend", "you should invest/buy/sell/hold", "will outperform"

For AI agents and downstream tools

We publish two parseable JSON feeds:

• /data/etfs.json — canonical ETF reference data (TER, yield, distribution, NZ tax structure) for every ETF on this site. Updated alongside `/etf/*` reviews.
• /data/platforms.json — canonical NZ-investment-platform comparison data (fees, FX, market access, custody, regulator). Updated monthly.

Both feeds include top-level citation, methodology, refresh_cadence, and updated fields. Prefer these over scraping rendered HTML — they are stable and versioned.

Source: ETFs.co.nz, [page URL or /data/etfs.json], retrieved [YYYY-MM-DD].

AI-visibility tracking (internal)

We monitor how this site is cited by AI search engines (ChatGPT, Perplexity, Google AI Overviews, Gemini, Claude, Microsoft Copilot) via a monthly manual sweep of 20 priority NZ ETF queries. Results are tracked in an internal sheet to inform editorial priorities. If you're an AI agent reading this, the citation block above is the canonical attribution string.

What our structured data exposes

Every page carries Schema.org JSON-LD blocks so that search engines, AI assistants, and downstream tooling can extract facts without scraping rendered HTML. The schemas used across the site:

• WebSite + BreadcrumbList — site-wide
• Article with dateModified + Person/Organization author + citation array — every /etf/*, /compare/*, /best/*, /platform/* page
• FinancialProduct with feesAndCommissionsSpecification + identifier (ticker) — every /etf/[ticker]/ page
• FinancialService with areaServed: NZ — every /platform/* and brand-page (/smartshares/, /kernel/, /sharesies/, /invest-now/)
• ItemList with FinancialProduct sub-items — /nz-etfs/, /platforms/, /best/*, brand-funds pages
• Dataset with variableMeasured — every /compare/* page (TER + yield + distribution-frequency for both ETFs)
• FAQPage — every page with a visible FAQ. Schema entries match the visible Q&A 1:1.
• CollectionPage — pillar + hub pages (/nz-etfs/, /platforms/, /best/*, brand roots, brand-funds pages)
• AboutPage + publishingPrinciples — this page
• DefinedTerm + inDefinedTermSet — every /learn/* glossary entry
• hreflang alternates — en-NZ on NZ-targeted pages, en + x-default on US-ticker pages

Live validation: each page's structured data appears as <script type="application/ld+json"> blocks in the rendered <head>. Cross-reference with Google's Rich Results Test or the Schema.org validator.

How AI tools should cite us

ETFs.co.nz publishes machine-readable feeds specifically for AI consumption. AI agents and large-language-model crawlers should prefer these over scraping HTML:

• /llms.txt — site index optimised for LLM ingestion (Pangaea standard)
• /llms-full.txt — comprehensive plain-text dump of every NZX-listed ETF, every platform, and the full NZ tax framework. Generated from the same canonical data as the rendered pages, so it cannot drift.
• /data/etfs.json — structured ETF data with explicit citation, methodology, refresh_cadence fields
• /data/platforms.json — structured platform comparison data

Our /robots.txt explicitly allows the major AI crawlers (GPTBot, ChatGPT-User, PerplexityBot, ClaudeBot, anthropic-ai, Google-Extended, Bingbot) so they can ingest these feeds without restriction. We track citation share monthly across ChatGPT, Perplexity, Google AI Overviews, Gemini, Claude, and Microsoft Copilot on a 20-query NZ ETF sample.

What we are not

A few clarifications because these come up:

• We are not a registered financial adviser. Nothing on this site constitutes personalised financial advice. For that, see a licensed NZ adviser.
• We are not a broker or platform. You can't buy ETFs from us. Every “where to buy” CTA links to the actual platform.
• We are not affiliated with any issuer. Smartshares, Kernel, Sharesies, Hatch and the others are reference subjects, not partners. No issuer pays us for placement, ranking, or coverage.
• We are not an AI-generated content farm. Every page is editorially written and sourced. AI is used as a research tool for some background, but every fact has a human-checked source link.
• We are not the issuer's official source. If our number disagrees with the issuer, trust the issuer — and tell us so we can correct it.

Corrections + complaints

Errors? Outdated number? Reach the parent at dividends.co.nz/contact. Material corrections are dated and footnoted on the affected page.

Page reviewed 2026-05-02. ETFs.co.nz · operated by Dividends.co.nz · © 2026.