The measurement problem
Energy has no shortage of data. Every module, battery and inverter ships with a datasheet. Every market publishes a price, every grid a dispatch record, every industry roadmap a projection. What the sector lacks is a record: a durable, auditable account of what was claimed, what was measured, and what actually happened.
Without that record, the failures compound. Manufacturer marketing overstates performance, because nothing checks it later. Roadmaps overpromise a technology mix that quietly revises itself inside a year. Market forecasts are published with a confidence no one returns to mark. Households then commit five-figure sums, and institutions commit rather more, against numbers nobody has gone back and settled against reality.
Solar Analytica is the correction. We aggregate the evidence in one place and keep it honest: every figure named to its source, nothing modelled where it can be measured, and the record of what each projection said before the data arrived to settle it. Attribution is the floor, not the standard. Wherever the source is public, the number is reproducible: run the same records through the published method and you arrive at the same figure. That is the difference between a citation, which asks to be trusted, and a record, which asks to be checked.
Held that way, the record does two things to forecasting. It feeds the forward view better, because a projection built on reconciled primary data starts from what actually happened rather than from someone's summary of it. And it holds the forward view to account, because a projection published over a durable record can be marked when the data arrives. We do not claim to see the future more clearly. We make claims about the future accountable.
Australia is the forward edge
The most useful place to build an energy dataset is the market that reaches the future first. On present evidence, that market is Australia.
Rooftop solar supplied 12.8% of Australia's electricity generation in the first half of 2025, from 26.8 GW installed across 4.2 million homes and small businesses (Clean Energy Council, Rooftop Solar and Storage Report, September 2025). The Australian PV Institute counted more than 4.29 million PV installations nationally at 31 December 2025.
Storage is following the same curve, faster. SunWiz recorded 221,000 residential battery systems installed during 2025, roughly a threefold increase on 2024, taking battery ownership to 4.6% of Australian homes. By May 2026 the federal Cheaper Home Batteries Program had driven more than 400,000 installations and 11.2 GWh of cumulative residential capacity (SunWiz, Battery Market Report Australia 2026; Department of Climate Change, Energy, the Environment and Water).
Penetration at that level produces problems before other grids meet them: collapsing midday minimum demand, voltage rise at the feeder end, export curtailment, and the open question of what several million household batteries do when they are dispatched together. Australia is not a small market that happens to like solar. It is the working prototype of a high-penetration grid, and it is running now.
That is why a dataset built here travels. What is observed in Australia this year, most markets will read as a forecast for theirs.
Four layers, one method
The work is ordered deliberately. Nothing is projected that has not first been recorded, nothing is measured that has not first been sourced, and nothing reaches the field that the record cannot back.
| Layer | What it does | Where it lives |
|---|---|---|
| The record | Observed data from primary sources: NASA and NOAA climate, market and grid dispatch data, retail tariffs, and manufacturer datasheets. No projection, no interpretation. | Solar Analytica |
| The instrument | Deterministic measurement of the record: market indices and climate readings on the data side, product scores on the review side. Reproducible by anyone holding the same sources. | Solar Analytica · review.solar |
| The projection | Scenario models: low, base and high, aggregated from published roadmaps rather than invented. Where the roadmaps disagree, the disagreement is preserved as the spread. | Solar Analytica |
| The bench | The applied layer: fault codes decoded, technology explained, and calculators that show their working, for the people who install and keep the hardware running. | wattbench |
The third layer is the one most often faked. A single confident number for 2030 is easy to produce and impossible to check. A range, with each endpoint traced to the source that argues for it, is harder to write and considerably more useful to plan against. The fourth layer returns the evidence to the roof: the same record, written for the person holding the multimeter. The three properties are one instrument read at three altitudes.
We mark our own homework
A forecast nobody returns to is an opinion with a date on it.
Our 2030 monocrystalline-silicon scenario model names six roadmap projections that observed data has already overtaken, including figures we would otherwise have relied on. Among them: n-type wafer share reached roughly 82% in 2025, against 2024-vintage roadmaps expecting a slower climb, and CPIA's own 2030 heterojunction forecast fell from 20% to 7.5% inside about a year (ITRPV 17th edition, June 2026; TaiyangNews comparative analysis of CPIA, Exawatt and ITRPV).
These markings now live on a register of their own: the Forecast Ledger holds the industry's dated projections and our own scenario endpoints side by side, each marked in public as the data settles it.
The same discipline runs through the market work. The TBx Capture Index measures what every battery in Australia's National Electricity Market actually earned against a perfect-foresight benchmark, so a contract structure can be priced against the record rather than a vendor's model. We publish the corrections because they are the interesting part. Where a methodology changes, it receives a version number, a changelog entry, and a re-run of every figure it touched. A number moves when the data moves or when the method moves, and we say which one it was.
Ownership and independence
Solar Analytica is operated by Solar Analytica Pty Ltd (ACN 653 738 115), a privately held Australian company. The following commitments are permanently in effect:
- We accept no payment, sponsorship, or consideration of any kind in exchange for coverage, placement, ratings, or conclusions.
- We hold no position in any asset, market, product, or company we write about. Where a report names a real asset or counterparty, it states facts from the public record and draws no conclusion about any party's private position.
- Every figure is named to its source and, wherever the source is public, reproducible from it. A reader with the same records can arrive at the same number.
- On the product side, the same standard runs the review.solar rubric: every product with a public datasheet is eligible, inclusion cannot be bought or blocked, and scores are derived mechanically from published sources.
- When a methodology changes, we publish a new version, document what changed, and re-run what it affects. Movements are attributable to a data change or a method change, never to an undisclosed adjustment.
We describe ourselves as evidence-first. We do not claim to be without commercial interest: we license data, as set out below. Reproducibility, rather than a promise of purity, is what actually protects the reader. Check the number.
Why the name is set in lower case
The wordmark is solar_analytica: monospace, lower case, the two words joined by an underscore. It is written the way you would write a variable, because that is what it is. The value it holds keeps changing. The method that assigns it does not.
Beside it sits the index mark: eight readings on a baseline, seven of them grey, one in Signal Blue. Most data is grey. The work is finding the single reading worth acting on, and being honest that the other seven were not it.
Energy-data brands tend to reach for one vocabulary: shields, locks, navy blue, the word "trusted" set in a heavy sans serif. That vocabulary asks to be believed. We would rather be checked. So the numbers are set in a monospace face, the argument in a serif, and everything else recedes into grey.
Publications
Reports is Solar Analytica's own output, written for procurement, investment and policy rather than for a single household purchase. The library now runs to three lines: market-data reports covering behind-the-meter storage, the lithium-ion battery decade, monocrystalline silicon to 2030, and residential electricity prices across 92 cities; The Production Climate, a ten-part series reading the environmental variables between the sun and the socket against twenty years of NASA and NOAA record; and Market Structures, opening with the TBx Capture Index. All free, no login.
review.solar is the product altitude of the network: manufacturer-neutral scorecards for panels, batteries and inverters, derived mechanically from published datasheets under a published rubric, maintained on its own domain. Every scorecard links back to the rubric that produced it and the datasheet it was derived from.
wattbench is the applied bench: fault-code libraries, plain-English explainers and browser calculators for the people who install and keep solar running. The library was built here and is moving to its own property; it stays free, with no login and no lead capture. The network, mapped shows how the three fit together.
All of it is readable by machines as well as by people. The site publishes an llms.txt map for AI assistants, full-text endpoints, and the fault-code library as open JSON and CSV. Data that cannot be checked is not evidence, and data that cannot be fetched will not be checked.
Licensing and data access
Solar Analytica data is licensed to platforms, publishers and tools that want neutral, source-anchored numbers inside their own product. Licences are structured to preserve the integrity of the data: a licensee cannot filter, suppress, reorder, or modify a figure. Where a number is wrong, the correction runs through the methodology, in public, for everyone at once.
For licensing, white-label integration, bulk data access, media requests, or methodology questions, see Contact.