Three in Five: What the Battery Inspection Rate Counts

What the rating counts

Batteries became eligible under the Small-scale Renewable Energy Scheme on 1 July 2025, and the CER's battery inspection program started a fortnight later, on 15 July. To 30 June 2026 it had assessed 3,425 systems: 1,266 adequate, 2,133 substandard and 26 unsafe. That is the 62.28 per cent now in circulation.

The regulator defines the middle rating precisely. Substandard systems "do not meet one or more clauses in Australian Standards or industry guidelines but are safe to remain in operation". One clause, anywhere on the checklist, rates the whole system. That is a reasonable way to run an inspection program and a poor way to read one, because it means the system rate and the defect rate are different measurements of different things.

The item-level figures show how far apart. Across 285,889 assessed checklist items, 267,282 complied, 14,938 drew a recommendation for improvement, 3,639 required rectification and 30 were unsafe. So 93.49 per cent of everything inspectors looked at was compliant, and 1.28 per cent needed fixing. Those 3,639 rectifications are spread across 2,133 substandard systems, which is fewer than two per system.

The rate that did not move

If installation quality had genuinely collapsed in a new product category, the unsafe rating should carry some of it. It does not. Batteries were rated unsafe in 26 of 3,425 systems, 0.76 per cent, and at item level in 30 of 285,889, which is around one item in nine and a half thousand.

The rooftop solar program is the closest available control, run by the same regulator against the same three ratings. In its 2024-25 program the CER inspected 2,586 solar systems, of which 2,532 were a random sample. Of those, 223 systems installed in 2024 were rated substandard (18.6 per cent) and 5 were rated unsafe (0.4 per cent); for systems installed in 2023 the figures were 262 substandard (19.6 per cent) and 12 unsafe (0.8 per cent). The battery unsafe rate sits inside that band. Over the life of the solar program to 30 June 2025, the unsafe share is higher still: 300 of 11,281 inspections in New South Wales and 332 of 12,129 in Queensland, both near 2.7 per cent.

So the two categories differ by a factor of more than three on substandard and by nothing meaningful on unsafe. Whatever is driving the battery figure is being caught by the rating that explicitly means "safe to remain in operation".

Where the flagged items sit

The category breakdown says what it is. Labelling is compliant on 76.28 per cent of assessed items, which makes it the weakest category by a wide margin, but most of that gap is recommendations for improvement, 20.73 per cent of assessed items; rectifications account for 2.98 per cent. Mechanical protection carries the highest rectification rate at 4.02 per cent, with overcurrent protection next at 1.71.

One caution on reading that ranking as a ranking of causes: these are rates within each category, and the CER does not publish how many items each category contributes. A category can carry the highest rate and still account for few rectifications overall. What the table does support is where the unsafe items are, and they are narrow: wiring (0.03 per cent of its items), installation requirement (0.02) and isolation (0.01). No other category recorded an unsafe item at all.

Clean Energy Regulator, battery inspection results report, checklist criteria results by category, covering 1 July 2025 to 30 June 2026. Percentages are of assessed items within each category, not of all items. The CER does not publish the item count per category, so these rates cannot be converted into shares of total rectifications.

The spread the national figure hides

The per-state counts are the part of this dataset I would look at first, and they are not what a single national rate suggests. Queensland rated 554 of 718 inspected systems substandard, 77.2 per cent. The ACT rated 64 of 154, or 41.6 per cent. That is a spread of thirty-five points inside one national scheme with one checklist. The Northern Territory rates higher than Queensland, 27 of 34, but on a denominator too small to carry the comparison.

Counts from the CER's published per-state CSV; the substandard rate is calculated from those counts and is not published by the regulator. The Northern Territory (34 inspections) and Tasmania (44) still carry the smallest denominators by a wide margin, the next smallest being the ACT at 154, and are left out of the chart above for that reason.

What I am not willing to conclude

Four limits, and the first is the one that matters most. The CER states that it selects battery installations for inspection randomly from the small-scale technology certificate applications it receives, that it also inspects installers already found to be non-compliant, and that it will inspect "a statistically significant sample" of battery systems claiming certificates. What it does not publish for batteries is the split between those two streams. For solar it does: 2,532 of 2,586 inspections in 2024-25 were the random sample. Without the equivalent count for batteries, the share of the rate carried by installers already flagged as non-compliant is unknown, so the battery figure cannot be treated as an estimate of the installed population, and extrapolations that scale it to national totals do not follow from what is published.

Second, the solar comparison has a definitional break in it, and it falls where the solar series improves. The CER states that it refined its checklist so that installations "previously rated as substandard based on broad general clauses or at an individual inspector's discretion can now be assessed more precisely". Part of solar's improvement is a change in how the rating is applied. The update also carries two substandard series that do not agree: a summary series of 24.6 per cent for 2022, 22.6 for 2023 and 18.7 for 2024, and the random-sample series by install year quoted above. The report does not reconcile them.

Third, the battery figures are cumulative across the program's first year, so no period rate exists yet and no trend can be read. Four readings are now on the record, each containing the last.

Successive states of the same CER page, the January reading recovered from the Internet Archive. Subtracting one reading from another does not give a period rate, because the counts move as earlier inspections are finalised rather than only as new ones are carried out: between the last two readings Tasmania went from 1 inspection to 44 and New South Wales added 305 while Victoria added 27, over twenty-two days. That is backlog clearing, not a national flow. What the series does support is that the headline rate is stable, with substandard between 62.0 and 62.45 per cent across a sample that grew almost fivefold while unsafe drifted down from 0.9 to 0.76.

Fourth, this is a new checklist applied to a new product class by inspectors and installers both working to it for the first time. Solar's own record is the useful precedent here rather than a rebuke: on the CER's summary series it sat at 24.6 per cent substandard in 2022, and its life-of-program unsafe share in the two largest states is near 2.7 per cent, well above where batteries are now.

Why this is a fleet property

This belongs in a dataset because installation quality is a property of the installed fleet, and we have been treating it as uniform. A rate that runs from 41.6 per cent to 77.2 per cent across states is not uniform, and it is now measured. Any assumption about how a battery fleet performs or degrades that is applied evenly across jurisdictions is resting on something the regulator has started to publish and we have not been reading.

Three things I will look for in the next release. Whether a period rate becomes readable once the finalisation backlog stops moving the earlier counts. Whether the CER publishes the battery program's split between random and follow-up inspections, as it does for solar. And whether the state spread keeps narrowing, because it has already come in from nearly forty points to thirty-five, and if it keeps closing across a larger sample it is telling us something about jurisdictions rather than about batteries.