Where this platform departs from the Ocean Accounts framework, and why

Accounts are compiled by a country for its own territory. This follows from SEEA being a satellite of the System of National Accounts: to speak to economic and financial data, it inherits the economic territory and the residency principle. Pilots are national or sub-national — a bay, a province, an EEZ.

A restoration funder's portfolio does not respect borders. The question “of everything we funded this year, what moved most, and where is nobody working” requires one comparison across many jurisdictions. No country compiles an account for another country's waters, and no one compiles for areas beyond national jurisdiction — roughly two thirds of the ocean. Assembling forty national accounts would not produce the answer either, because they are compiled to different vintages, at different resolutions, and with different reference conditions.

Allocation in this sector is measurably poor. Across some 14,600 conservation projects over 25 years, 6% of threatened species received funding while 29% of funding went to species of least concern; 87% of reptile conservation funding went to seven marine turtle species. Marine protection covered less than 2% of ecoregions with high abatable threat — about 59% less protection in high-threat areas than random placement would have produced.

One global variable set, compiled once, applied to every project regardless of jurisdiction.

Our figures cannot be entered into a national account, and they are not official statistics. They are a management view, not a statistical product.

Summary indicators are published individually, each with its context and its data quality caveats. There is no weighting scheme and no total. This is deliberate: aggregating across incommensurable variables hides more than it shows.

A funder comparing forty grants across different variables needs commensurability. “This grant moved coral cover by 3 points, that one moved mangrove extent by 200 hectares” is not a comparison. Without a common scale there is no portfolio view, and the portfolio view is the product. This is not a question we invented: GOAP's own account of why ocean accounts matter includes activating private finance for restoration by giving investors comparable, regularly updated project data.

We compute a weighted index — with three constraints. The index is never displayed without its coverage. We do not aggregate across framework components: ecosystem condition, pressures from economic activity and designated use are computed and shown separately, because the framework is right that they are not commensurable. Every variable carries the source and the reference basis of its score, reachable in one click from the number.

A single number invites being quoted alone. We mitigate this in the interface; we cannot prevent it.

Five components: pressures, assets (extent and condition), ecosystem services, ocean economy, governance.

We measure movement in ecosystem condition attributable to restoration effort. Services and economy answer a different question — the value of what the ocean provides to people — for a different user, and they require valuation methods that are contested and slow to apply. One of those methods is closed to us in any case: restoration cost is explicitly rejected as a valuation approach, on the grounds that a restored ecosystem is unlikely to provide the same services as the one it replaced. Since restoration expenditure is exactly what our users hold, monetising contribution through it is not available.

We hold assets and pressures. Services, economy and governance are absent.

We cannot answer “what is this worth”, only “what moved”. For a funder deciding between grants, the second question is the operative one; for a treasury deciding between sectors, it is not.

The IUCN Global Ecosystem Typology is recommended as the reference classification for ecosystems.

Our variable set has to do two things a scientific typology does not: be navigable by a project lead entering data who is not an ecologist, and carry weights. GET is built to classify ecosystems correctly, not to be a weighting structure or a data entry interface. A classification a user cannot map their project onto blocks entry, and no entry means no data.

Six categories with our own hierarchy, weights normalised per row, and a crosswalk to GET rather than a replacement of it.

Comparability with ecological literature organised by GET, until the crosswalk is complete.

Data quality caveats accompany published indicators; vintage of data is among them. But an account records what is compiled, not what could not be.

In an account, an empty cell says nothing about why it is empty. For a funder reading a map of effort, this is not a cosmetic problem: an unmeasured region and an unworked region look identical, and the decisions that follow are opposite.

Every one of the 170 variables carries either a published global value or a recorded reason why none exists, classified by mechanism. Of the 170: 84 have a global value, 63 are meaningful only locally, and 23 are measured nowhere in the world.

Nothing, but it costs compilation effort that an account is not obliged to spend.

The distinction is not academic. Data that exist but are withheld close through requests and regulation; an absent observation network closes through project measurement; a technical barrier does not close at all and needs a proxy. Treating these as one category — “no data” — produces the wrong response to each.