Fewer NSW public offices can produce a complete view of their highest risk records than could two years ago. Overall record-keeping compliance across the state still edged upwards.
The finding comes from State of recordkeeping in NSW 2026, published in July by State Records NSW. It reports the fourth Recordkeeping Monitoring Exercise. That exercise is a self-assessment against 19 questions drawn from the State Records Act 1998 and the standards made under it.
Of the 379 public offices asked to take part, 321 lodged submissions between 2 and 27 March 2026. That is a response rate of 85 per cent, up from 77 per cent in 2024.
Responses came from 156 agencies, authorities or departments, 125 local councils and 15 local health districts. Ten universities, seven state owned corporations and eight courts or tribunals also took part. Courts and tribunals participated for the first time.
Question 1 asks whether an organisation has formally identified its high risk and high value information assets. Compliance with it fell from 73.4 per cent in 2024 to 68.5 per cent in 2026. Question 2, covering information risk, fell from 71.3 per cent to 66.4 per cent.
The report calls the Question 1 result concerning. High risk and high value business areas "should be the priority for records and information management", it states. Compliance requires a complete, comprehensive, single view of those assets. Ten other questions in the assessment tool depend on that view existing.
Public offices said the work is now bound up with the "crown jewels" identification required for cyber security attestation. Information risk is increasingly handled inside cyber security or ICT risk initiatives. Information risk registers are being folded into the main enterprise risk register.
The overall records management maturity score for the jurisdiction is 2.94 out of 5. That is up from 2.93 in 2024 and 2.79 in 2023. A score of 3 represents baseline compliance with State Records NSW requirements.
The overall compliance rate is measured as the share of all responses scoring 3 or above. It sits at 69.85 per cent, against 68.57 per cent in 2024 and 62.31 per cent in 2023.
Of the 321 participating offices, 12 scored at the initial level, 149 at developing, 147 at defined and 13 at managed. None reached the optimising level. The number sitting at managed has not moved since 2024.
By sector, courts or tribunals recorded the highest average at 3.08. Local health districts followed at 2.99 and agencies, authorities and departments at 2.97. Local councils and state owned corporations sat at 2.90, universities at 2.89.
Self-assessment now gets checked
State Records NSW cautions that the improving trend may reflect the new assurance measures introduced this year rather than changes in practice.
Three assurance measures were added for 2026. A new mandatory question asked whether the submission was accurate and had been approved or endorsed by the chief executive. In response, 318 offices, or 99 per cent, answered yes.
That question also generated a jump in requests for deadline extensions. Offices cited delays in obtaining organisational approval. State Records NSW says this suggests some had not sought that approval for previous exercises.
Every submission was then checked to confirm supporting evidence had been named. The check found 25 public offices had identified no evidence at all, or none across a major part of their response. Nineteen supplied further information when contacted.
State Records NSW also ran desktop verification on a sample of 10 offices, about 3 per cent of participants. The sample covered the questions on records and information strategy, quality of records, and retention and disposal. Maturity scores were not changed as a result.
The verification found some offices had no strategy or roadmap for their records management program. Those offices relied instead on a strategic direction stated inside a policy document.
Others lacked documented disposal procedures, including how disposal actions are recorded and how destruction is made secure and irreversible. Some supplied evidence that was overdue for review, or that referred to superseded retention and disposal authorities.
The exercise also found offices selecting a maturity level without meeting all its criteria. Others did not recognise that they must also meet the criteria of the level below. Levels in the assessment tool are nested and build on each other.
Transfer stays the weakest area
Question 18 covers transfer of State archives to Museums of History NSW. It remains the lowest scoring question at 2.60 out of 5. Compliance improved from 46.7 per cent to 53 per cent. State Records NSW credits engagement work by Museums of History NSW on transfer planning requirements introduced in 2024 and 2025.
Comments from public offices show why progress is slow. Many have lodged transfer plans, but transfers have been limited, ad hoc or have not started. Paper accounts for the majority of transfers, with very few digital transfers completed.
Reasons given include limited resourcing, competing record-keeping priorities and no developed process for transfer. Some offices hold records identified as State archives that are still in regular use. Some said they have no intention of transferring records.
Performance monitoring, Question 9, did not move at all and held at 2.63. Offices reported heavy reliance on automated reporting from their EDRMS and on tools such as Power BI. Others said non-compliance is usually picked up by chance during other activities.
Five questions have stalled at compliance rates between 63 and 57 per cent. Ranked from best to worst, they are quality records and information, search and discovery, records and information strategy, retention and disposal, and performance monitoring.
State Records NSW attributes the pattern to inconsistent approaches across different areas of an organisation, and to a lack of resourcing. Those questions will be the focus of further guidance in 2026-27.
The next Recordkeeping Monitoring Exercise runs in March 2028, following the move from an annual to a biennial cycle.