When the replacement
is also superseded

What the register shows you, and what it does not

When a training product is superseded, training.gov.au records two things about it: the product that replaced it, and whether the replacement is equivalent or not equivalent. Both are published, both are free to read, and both are exactly one step wide.

What the register does not do is walk the line for you. If the replacement was itself superseded two years later, that fact lives on the replacement's own page, not on yours. Follow it far enough and you reach whatever the register lists today — but nothing tells you in advance how far that is, or what happened to the equivalence marking along the way.

So we measured it. Every superseded product in the catalogue, every hop, to the end of every line.

How often the replacement is itself superseded

The register holds 125,996 training products. 44,442 of them are superseded and carry supersession data; 42,693 of those declare a replacement. (The other 1,749 are superseded and point nowhere at all — the register marks them superseded and names no successor.)

Of the 42,693 that do declare a replacement:

Qualifications are the worst case, and they are also the case that matters most to a provider: a qualification is the thing learners are enrolled in. Two qualifications out of every three whose replacement you might be about to transition into are pointing at a product that has already moved on.

How long the chains get

Counting hops from the superseded product to the point where the register stops forwarding you:

Just under half of all supersessions are a single step. The other 52.3% are not. For qualifications on their own, 1,706 of 4,590 — 37.2% — need three hops or more.

Here is one of the two seven-hop lines, in full:

TLI31910 → TLI31911 → TLI31913 → TLI31915 → TLI31918 → TLI30220 → TLI31920 → TLI31921

Certificate III in Mechanical Rail Signalling — every hop marked equivalent, and TLI31921 is current

That one has a happy ending: seven supersessions, every one of them marked equivalent, and the product at the end is live today. A provider holding TLI31910 on scope is looking at a replacement six generations out of date, but the destination is real and the line is clean.

The equivalence marking is about one step, not the journey

This is the finding with teeth. The equivalence flag is what tells a provider how much work a supersession is: equivalent means the replacement was judged to be the same product in substance, not equivalent means it was not. On a scope report you see the marking for the first hop — from your product to its immediate replacement.

Of the 22,337 chains that take two hops or more:

• 11,945 (53.5%) are marked equivalent at every hop.
• 10,392 (46.5%) contain at least one hop the register marks not equivalent.
• And of those 10,392, in 5,570 the first hop — the one on your scope report — is marked equivalent.

So for 5,570 training products — 24.9% of all multi-hop chains — the register shows you a reassuring marking on the step in front of you, and a step further along the same line is marked not equivalent. For qualifications alone that is 1,066 of 3,026 multi-hop chains, or 35.2%.

Verified example, hop by hop:

PUA30100 —E→ PUA30110 —E→ PUA30112 —N→ POL31115 —E→ POL30119 —E→ POL30122 —E→ POL30125

Certificate III in Public Safety (Aboriginal or Torres Strait Islander Community Policing) → … → Certificate III in Aboriginal and/or Torres Strait Islander Community Policing. E = equivalent, N = not equivalent

A provider with PUA30100 on scope sees one word: equivalent. It is correct — PUA30110 really was judged equivalent to PUA30100. But the current product at the end of that line, POL30125, sits on the far side of a hop that the register marks not equivalent, and it does not even carry the same name.

Where the chains end

Following each line to the point where the register stops forwarding:

537 qualifications lead, hop by hop, to a product the register lists as deleted. In 93 of them every hop along the way is marked equivalent — the line looks clean the whole distance, and the destination is gone. For skill sets the deleted ending is more common still: 227 of 1,255, or 18.1%.

AUM20100 → AUM20108 → AUM20112 → AUM20113 → AUM20118

Certificate II in Automotive Manufacturing — four hops, every one marked equivalent, and the register lists AUM20118 as deleted

We are not going to tell you what “deleted” obliges anyone to do; that is not ours to say. What we can say is what it means practically: following the register's own trail does not always land you on something you can deliver, and the trail gives no warning that it is about to run out.

What this changes in practice

• “We have identified the replacement” is not the end of the job.For 53.2% of superseded products — 66.0% of qualifications — the replacement you identified needs the same check run on it. Once.
• Read the equivalence marking as a property of a step, not of a destination.An equivalent first hop tells you the next product is substantially the same. It tells you nothing about the product two hops further on, and in 5,570 cases the line turns not equivalent later.
• Check that the end of the line is somewhere you can actually go. One in seven chains ends at a deleted product, and the register does not flag that when you start walking.
• The enrolment clock runs on the hop, not the line. The one-year limit for enrolling new students runs from the date the replacement went on the register — see the 12-month rule, which was repealed— and a chain of three hops has three such dates, not one.

Method, so you can redo it

Two public sources, no private data, no estimate anywhere in the numbers above.

1. The catalogue. All 125,996 training products in the register, downloaded from the training.gov.au search API in partitions of type × status — the API caps pagination at offset 100,000, so a single loop would silently drop about 26,000 products. Downloaded 13 August 2026. This gives the current status of every code.
2. The edges. The supersededBy record of every superseded product — 44,442 of them — read from the same API in 53 calls on 14 August 2026. Each edge carries the replacement code and its equivalence marking.
3. The walk. From each superseded product, follow the replacement while the product you land on is itself superseded, and stop when it is not. A hop is only counted if the destination exists in the catalogue.

What the walk found in the register's own data, reported rather than smoothed over:

• 0 cycles. No chain loops back on itself.
• 0 dangling references. Every declared replacement exists in the catalogue.
• 342 products declare more than one replacement. For those we follow the first one listed, which is a choice we are making rather than one the register makes for us. It affects 0.8% of the products measured.

What these numbers are not. This is a measurement of the catalogue, not of providers. It says nothing about how many RTOs have any of these products on their live scope — that is a different count, and it is the one in our study of the sector. And the register moves: re-run this next month and the figures will differ slightly, which is rather the point of the article.

Questions

How often is the replacement of a superseded training product itself superseded?

For 22,724 of the 42,693 superseded training products that declare a replacement on training.gov.au — 53.2% — the replacement shown is itself superseded. For qualifications specifically it is 3,029 of 4,590, or 66.0%. Measured across the whole national catalogue on 14 August 2026.

How long can a supersession chain get?

The longest chains in the register are seven hops. Certificate III in Mechanical Rail Signalling runs TLI31910 to TLI31911 to TLI31913 to TLI31915 to TLI31918 to TLI30220 to TLI31920 to TLI31921, which is the current product. Across all superseded products with a replacement, 47.7% resolve in one hop and 52.3% take two or more.

Does the equivalence flag apply to the whole chain?

No. The register marks each hop, not the line as a whole. Of the 22,337 chains that take two or more hops, 10,392 contain at least one hop marked not equivalent — and in 5,570 of those the first hop, the one shown on a scope report, is marked equivalent. The marking you see describes the step in front of you, not the distance to the current product.

Does every supersession chain end at a current training product?

No. 36,050 of 42,693 chains (84.4%) end at a product the register lists as current. 6,074 (14.2%) end at one the register lists as deleted, and 569 (1.3%) end at a product that is still superseded and declares no replacement at all. For qualifications the split is 88.2% current, 11.7% deleted.

Sources

• training.gov.au — Australia's National Register of VET. Every figure on this page is derived from its public search API: the catalogue of 125,996 products and the supersededBy record of each superseded one.
• The three example chains — TLI31910, PUA30100 and AUM20100 — were re-read product by product against the live API on 14 August 2026, twenty codes in all. You can check any of them yourself: each product has a public page at /q.

Measured on 14 August 2026 against training.gov.au: catalogue of 125,996 products read 13 August 2026, supersession edges for all 44,442 superseded products read 14 August 2026 in 53 API calls, and each chain walked to its end. The three example chains were re-verified code by code against the live API the same day. This describes the national catalogue, not any provider's scope, and not any provider's learners: how many RTOs hold these products is a separate count. Chains with more than one declared replacement (342, or 0.8%) follow the first replacement listed. We read the register; we do not assess compliance, and nothing here is a judgement about any provider's obligations.