COVIDSafe: App worked less than 25% of the time on locked iPhones at launch

We already knew the COVIDSafe app Aussies were urged to download had some problems, but now we know exactly how it performed.

The COVIDSafe app that Australians were urged to download was launched even though in some cases it worked about a quarter of the time on the phones used by nearly half the country, new documents show.

The documents tabled in parliament in response to questioning from South Australian Senator Rex Patrick, show that the COVIDSafe app administrator’s own testing found iPhones logged contact less than a quarter of the time when locked.

RELATED: App with ‘one job’ doesn’t work

The COVIDSafe app has been improved since first launching. Picture: Saeed Khan / AFP

The Digital Transformation Agency, the COVIDSafe administrator, is the only agency outside of state health authorities legally permitted to access the app’s data, which has to be kept on protected servers in Australia and never leave the country.

On the day of the app’s launch, Federal Government Services Minister Stuart Robert told Australians to keep the app running in the background, but that “your phone does not need to be unlocked for the app to work”.

But the Digital Transformation Agency’s own testing of the app prior to release shows it clearly performed better on some phones than others, and that having your iPhone locked rendered the app essentially useless.

A test returned an “excellent” result if it logged between 80 and 100 per cent of all encounters with other devices.

A 50 to 80 per cent success rate was regarded as “good”, while “moderate” success was a log rate of 25 to 50 per cent of all encounters.

A test returned a “poor” result when the app couldn’t log any more than 25 per cent of encounters.

RELATED: Reasons we’re not downloading app

The app got better over time.

On April 26, the day the app went live, it was only able to consistently detect at least 50 per cent of contacts on Android phones communicating with other Android phones.

iPhones were only able to achieve “excellent” status if both phones had the app open and active.

If both phones were locked then the tests returned “poor” results.

A few weeks later on May 14, Android phones performed “excellent” across the board, while iPhones had been improved to at least “moderate” success.

A month after the app went live, on May 26, locked iPhones were achieving “moderate” success communicating with other locked iPhones and Android phones, but had raised iPhone to iPhone success rates to “good” when only one of the phones was locked.

The DTA has said previously it has taken an “iterative approach” to the app’s rollout, to make it available as quickly as possible.

It has since issued a number of updates to address some of the initial problems.

Low numbers of new cases means that the app still hasn’t actually been used to identify any contacts with confirmed cases that health authorities hadn’t already identified via manual contact tracing.