Chief Information Officer / Australia
Three restrictions and what each one cost
- Tenancies managed
- 8,640
- Owned and state-leased homes across Queensland, June 2026
- Annual revenue
- A$218m
- Year to 30 June 2026; not-for-profit community housing
- Technology colleagues
- 41
- Housing systems, security, data and service delivery
Decisions & consequences
01Cutting the contractor portal from forty-six tenant fields to eleven
In August 2021 a tenant in Logan told her housing officer that a plumber had mentioned her protection order while fixing a tap. He had read it in our contractor portal, which I had designed in 2016 to show 46 tenant fields, including free-text alerts, to anyone holding a login. A review found 1,940 active accounts across 212 firms, 38 per cent unused for ninety days, one of them held by a subcontractor's employee who had left seven months earlier.
The stakes were tenants' safety and our obligations under the Privacy Act, which require reasonable steps to protect personal information and limit its use to the purpose for which it was collected. Our general manager of assets disagreed with my proposal to reduce the portal to eleven fields. She argued that tradespeople would arrive unprepared, first-visit completions would fall and repeat visits would cost more than any privacy benefit. The contractor panel supported her.
The executive accepted my recommendation, with her cost concern recorded. We kept the address, a work-order contact number, preferred times, interpreter needs, pets, access instructions and three instruction flags: two workers, call the housing officer first and do not attend alone. Reasons stayed with tenancy staff. Each firm's principal now confirms its accounts quarterly, and unused logins expire after sixty days.
First-visit access fell from 91 per cent to 86 per cent over five months, and calls to our coordinators rose by about 1,100 a month. We hired three coordinators at A$260,000 a year, a cost I had underestimated. By the ninth month first-visit access had recovered to 90 per cent, and accounts had fallen to 1,180. No contractor-related privacy complaint has been upheld since.
What changed was the principle: a contractor now receives an instruction, not a history. I would do one thing differently. We wrote the flag wording ourselves and revised it twice after tenants found it confusing. Our tenant advisory panel and tenancy officers should have drafted it with us before launch.
Notifying 1,326 people after one mailbox compromise, not the 214 our lawyers listed
In March 2023 a housing officer in our Rockhampton office entered her password on a convincing copy of our sign-in page, and an attacker read her mailbox for nineteen hours before our monitoring ended the session. The mailbox held correspondence with 1,326 tenants and applicants. Attachments included income statements, identity documents and letters from support services.
Forensic investigators confirmed access but could not rule out downloading. Under the Notifiable Data Breaches scheme we had to assess whether the breach was likely to result in serious harm and, if so, notify the OAIC and the people affected. External counsel considered the threshold met for 214 people whose identity documents were exposed and suggested a general notice for everyone else. The chief financial officer supported that view, because the wider approach would cost more and draw attention to the incident.
I recommended notifying all 1,326. Much of the correspondence revealed income, disability or family circumstances, and our tenants are poorly placed to absorb identity misuse. I also proposed that housing officers telephone the 312 people with safety flags before anything was posted, because a letter opened by another member of the household could itself cause harm. The chair of the Audit, Risk and Technology Committee supported the recommendation, and the chief executive approved it.
We contacted everyone within nine days. Forty-seven people asked us to change contact details or replace documents, and we paid for identity-protection support for the 214. The response cost A$410,000, against about A$150,000 for the narrower option. The regulator made inquiries and closed the matter after reviewing our response.
Since then, attachments are moved into the housing record within five days and deleted from email, and conditional access blocks sign-ins from unmanaged devices. What I would do differently is preparation. Building the list of tenants who needed a telephone call took two days in the middle of the incident. We now maintain that list, checked each quarter, before anyone needs it.
Walking through every home before buying software to predict its condition
In early 2024 the board was asked to approve a ten-year maintenance forecast of A$186 million that rested on condition surveys carried out in 2016 and 2017 and aged by formula ever since. A software vendor offered a predictive platform for A$1.1 million over three years that would estimate component condition from our work-order history. The chief operating officer and the chief financial officer favoured it: it was cheaper, faster and avoided entering thousands of occupied homes.
I opposed it. The model would learn from repair requests, and many of our tenants do not report defects, whether from fear of being seen as difficult or because they have lived with worse. A forecast built on complaints would understate the homes of the quietest households. The state's minimum standards for rental housing also meant we needed evidence of condition, not an estimate.
I proposed a physical resurvey of all 8,640 homes over thirty months, using four in-house surveyors, two contracted firms and a mobile form built by my team. It cost A$3.4 million, and to fund it I deferred replacing our finance system by eighteen months, which the finance team resented with good reason. Surveys used the entry notice required by tenancy law, and the form allowed photographs of components only, never whole rooms.
By June 2026 we had surveyed 8,412 homes. Of those, 1,170 had at least one major component past its expected life, concentrated in older detached houses in central Queensland. The forecast rose to A$238 million. The board slowed a planned 120-home development programme to 80 homes over three years to fund the difference.
The board now sees the date of observation beside every condition figure. I would change one thing: survey a stratified sample of about 600 homes first. It would have shown the bias in the old data within four months and given the board an earlier, cheaper warning before it committed to the full programme.