The Numbers Behind Canberra's Duplicate Image Problem: What the Data Actually Shows

Canberra's public sector has a clutter problem — and it lives on hard drives. Across ACT government directorates, Australian Public Service agencies clustered along Constitution Avenue, and research institutions including the Australian National University, digital asset libraries have ballooned over the past decade to the point where duplicate images now account for a significant share of total storage overhead. The practical and financial consequences of that redundancy are only now being tallied with any rigour.

The timing matters. The ACT government is midway through a digital asset management review, and the federal government's own data-governance push — driven partly by the 2023 Data Availability and Transparency Act — has pushed agencies to audit what they actually hold before they can responsibly share it. Duplicate imagery is proving to be one of the most stubborn line items in those audits.

What the Numbers Look Like

Industry benchmarks cited in a 2024 Gartner report on enterprise content management put duplicate or near-duplicate image files at between 20 and 35 per cent of total unstructured data in large public-sector organisations. Apply that range to a mid-sized ACT directorate running, say, 50 terabytes of unstructured storage — a conservative figure for departments handling planning imagery, infrastructure photography and communications assets — and you are talking about 10 to 17 terabytes of essentially redundant data that still attracts licensing, storage and backup costs.

At current hyperscale cloud storage rates sitting around AUD $0.025 per gigabyte per month on platforms used by several ACT government entities, 15 terabytes of duplicate imagery translates to roughly $375 a month per directorate — or close to $4,500 annually — for files that add no operational value. Multiply that across a dozen directorates in Civic and Barton and the number starts to attract attention from chief financial officers.

The ANU Library's digital collections team acknowledged the issue publicly in its 2025 annual report, noting a multi-year digitisation program across the Chifley and Menzies buildings had produced significant duplication in photographic archives. The university did not publish a specific dollar figure for remediation, but the report flagged that a deduplication project was scheduled to begin in the second half of 2025. Progress on that project, as of early July 2026, has not been publicly confirmed.

Local Agencies Starting to Act

The ACT Government's Digital Strategy, updated in late 2024, includes a specific workstream on image and media asset rationalisation under the broader Digital Records Framework. The Majura Technology Park precinct, which houses several data centre facilities servicing ACT government workloads, has seen increased activity around storage tiering and automated deduplication tools since that framework went live.

Automated deduplication software — tools that scan libraries for perceptually identical or near-identical images using hash-based or machine-learning comparison — now costs between $8,000 and $40,000 annually for an enterprise licence, depending on the volume of assets processed. At the lower end of that range, a single directorate could theoretically recover its investment within two years purely on storage savings, without accounting for staff time saved during image retrieval.

The University of Canberra's library and information science program has incorporated digital asset auditing into its postgraduate curriculum, partly because the local job market — dominated by public service roles in Civic, Barton and the inner-north — increasingly demands those skills. Graduates have been placed in records management roles across the ACT public service in each of the past three years, according to UC's 2025 graduate outcomes data.

For ACT government agencies and APS departments beginning a formal audit, the practical starting point is establishing a baseline: how many image files exist, in how many repositories, and what proportion share a perceptual hash with at least one other file in the library. Without that baseline, no remediation project has a defensible scope or budget. The good news is that the tools to run that count are cheap; the harder part is the organisational will to act on what they find. The data, at least, is no longer a mystery.