Methodology

The renewal score (0–100)

Three additive components, capped at 100:

• Proximity (max 35) — how soon the contract ends. ≤90 days scores 35; ≤180 scores 27; ≤270 scores 18; ≤365 scores 9; beyond that, 4.
• Value (max 35) — scaled on a log₁₀ curve between $1k and $10M, so $100k ≈ 17 and $1M ≈ 26. Larger contracts are worth more effort.
• Incumbent entrenchment (max 30) — how many contracts this agency has awarded this supplier in this category, in tiers: 1 award scores 5; 2 scores 12; 3–4 scores 18; 5–9 scores 22; 10–24 scores 25; 25–99 scores 28; 100+ scores 30.

Bands: High ≥70, Medium ≥45, otherwise Watch.

Measuring entrenchment

Entrenchment is counted from a historical award index: every Commonwealth contract notice published over a rolling 24-month window (2024-08-01 to 2026-07-31), grouped by buyer and supplier and matched on ABN so that name variants ("X Pty Ltd", "X T/As Y") count as one company.

Crucially it is category-specific. A supplier deeply embedded in an agency's IT spend is not an incumbent on that agency's catering contract, so awards are counted within the contract's own UNSPSC Segment, not across every notice the pair has ever exchanged. Counting all categories together would overstate how hard the incumbent is to displace.

9,260 contracts were scored on their category's award count. The remaining 4,921 fall back to repeat contracts inside the expiring pool: either their awards predate the index window, or the supplier has fewer than two awards from this buyer in this category — which is precisely the "not an incumbent here" case, and is meant to score low.

The counts are never added together: they measure the same quantity over different windows, so the larger is used. The pool fallback is a floor, not a ceiling — it can only understate an incumbent, never overstate one.

A supplier's award count is not a prediction that they will win again. It measures how established the relationship is in this category, and therefore how much displacement effort a challenger should expect. It is tiered rather than flat — a supplier with 400 awards in a category outranks one with 4 — but it is still capped at 30, so anything past 100 awards is treated as equally entrenched.

Clustering: notices vs opportunities

Agencies frequently publish one contract notice per line item. Six separate notices for "Vehicle Spare Parts" to the same supplier, in the same category, all ending on the same day, are one re-procurement — not six.

The radar therefore groups notices sharing buyer + supplier + category + end date into a single opportunity, sums their disclosed value, and recomputes the score from that combined value using the formula above. The raw notice count is shown on the row, and the underlying records remain individually linked. Total value is unaffected by grouping.

~ATM (approach-to-market estimate)

Set at 120 days before the contract end date, reflecting the typical Commonwealth lead time for re-procurement. This is a rule of thumb, not a published date, and it will be wrong for panels, extensions and options. Treat it as "start paying attention", not as a deadline.

Entity resolution

Suppliers are merged on ABN, which is authoritative. Where a record carries no ABN, its normalised name (case-folded, legal suffixes and "trading as" clauses stripped) is matched against known ABNs; it is merged only when that name maps unambiguously to exactly one ABN. Otherwise it stays separate.

Currently 778 resolved suppliers, with zero ABNs split across multiple display names.

Categories

Commonwealth records carry a UNSPSC code, rolled up to its official 2-digit Segment label so that free-text descriptions aggregate into real categories. Records without a UNSPSC code fall back to the source's own category field, or to the raw description.

What the radar shows, and what it leaves out

The AusTender contractEnd feed returns tens of thousands of releases over a 12-month horizon and is not ordered by end date — a partial pull is a broad sample across the whole window, not the soonest-expiring contracts. Ranking such a subset would quietly bury the largest and most imminent opportunities.

So every contract in the pull is scored before anything is discarded. Aggregate figures (contract count, total value up for re-procurement) are computed over that full pool. Only the top-scoring slice is then published as radar rows.

The API uses cursor pagination and exposes no total-count endpoint, so an exact population figure cannot be obtained without walking every page. Totals are therefore published as a lower bound (≥) whenever the feed still had pages remaining — never as a complete count. The practical consequence: the radar is the highest-scoring slice of a large sample, and a contract's absence from it is not evidence that it does not exist.

Limitations — read these

• Renewal scoring needs a published end date. Coverage varies sharply by source (see the table on the home page). Where end dates are sparse, absence of a contract from the radar means nothing.
• Truncation is disclosed, not hidden. If a data pull hits its record cap or its page cap, the generated report says so and marks its totals as a lower bound.
• Coverage is Commonwealth-first. Several state portals prohibit commercial reuse or automated access and are excluded by design — so this is not, and does not claim to be, national.
• "Location" is the supplier's registered state in Commonwealth records, not where the work is performed.
• Disclosed value is the contract value at publication. Variations, extensions and exercised options may not be reflected.
• Entrenchment is measured within the expiring pool, not across the supplier's full history, so it understates long-standing incumbents whose other contracts end outside the window.

These numbers are a research and business-development aid built from public records. They are indicative. Before you commit bid resources, open the linked source record.