Evidence health · Living benchmark

94% of active benchmark criteria have a direct source link.

Catalogue growth is easy to count. Evidence depth is the more important number. Across 229 active products, 4,818 of 5,137 scored criterion rows currently resolve to at least one direct source URL. This report rebuilds from the published dataset and makes the remaining work visible.

review.solar

Direct criterion-level source linkage

Evidence depth differs materially by product category

Panels92%

80 products

Batteries98%

72 products

Inverters96%

69 products

Plug-in ESS45%

8 products

Chart data

Series	Value	Note
Panels	92%	80 products
Batteries	98%	72 products
Inverters	96%	69 products
Plug-in ESS	45%	8 products

How to read the percentage

Source linkage is an audit measure, not a quality score

1. 01
   Directly linked

   4,818 criteria publish at least one criterion-level URL in the scorecard audit trail.

2. 02
   Reviewed absence

   509 criteria link to the exact material reviewed, but the manufacturer or test programme does not publish the qualifying detail.

3. 03
   Still unlinked

   319 criteria may be estimated, derived, sourced only in prose, or among the 1 rows where no direct source has been located.

4. 04
   Product score

   Unaffected. Adding a URL does not alter the criterion value or score unless the underlying evidence changes the fact.

5. 05
   Research order

   Risk-led. Australian buyer priorities, weak categories, safety issues and high-impact warranty claims receive attention first.

Category health

What this number can and cannot say

A direct link makes a claim auditable. It does not automatically make the claim current, exact-model, primary or correctly interpreted. Those questions remain visible in each audit row's evidence state, regional overlay and source note. Conversely, a low linkage percentage does not mean the product is poor; it means the publication has more verification work to do before presenting the record with high confidence.

A reviewed absence is not an unresolved research request. It means the linked exact-product material was checked and did not publish the qualifying detail. It records non-disclosure without pretending that absence from one reviewed source proves the feature can never exist.

The next research queue

Work is prioritised where evidence gaps can change a purchasing decision: current Australian products, safety notices, warranty scope, model identity, grid compatibility and criteria with large scoring weight. This prevents raw catalogue expansion from outrunning the publication's ability to defend what it publishes.

Sources and methodliving dataset

• The active panel, battery, inverter and plug-in ESS scorecards.
• Published methodology, including evidence-status and missing-data rules.
• Publication governance and corrections policy.

Coverage values are generated at build time from public audit rows. They are descriptive publication-health measures and never feed a product score.