Making contact: the inside story of how the government’s Covid app was developed

More than 20 million UK citizens have downloaded the NHS Covid-19 app. One of the leaders of the development process discusses its impact and future plans with Sam Trendall
Danny Lawson/PA Wire/PA Images

By Sam Trendall

29 Mar 2021

For the 20 million citizens that have downloaded the NHS Covid-19 app, a digital service has surely never reassured and terrified them so much at the same time.

Launched across England in September, the app is updated weekly. Its most significant revamp took place in October and aimed to reduce instances of so-called ghost notifications – contact alerts which disappeared when users tried to open them.

Further code updates have been implemented more recently to integrate with and support the introduction of policies such as the tiering system – which is split by local authority areas, and not postcodes, as the app previously was – and the launch of isolation payments of up to £500.

The app has also been amended to reflect the more widespread usage of lateral flow tests, which are quicker but somewhat less reliable than PCR tests. Users who enter a positive lateral flow test result are now encouraged to take a further PCR test to confirm the result.

But this reporter missed these new features because, as I tell Wolfgang Emmerich, UK chief executive of Zühlke UK, the main commercial partner in building the technology, of the dread of opening the app, which I hadn't done in weeks before we spoke.

He laughs in recognition, and says my experience of app-ordered confinement – which lasted for nine full days, concluding at midnight on Christmas Eve – came during a particularly busy period for notifications. (He is yet to receive one, he reveals – although his daughter has.)

Further improvements to the app may cover the provision of information on whether users have been vaccinated or not, Emmerich says, although no definitive plans for this have yet been made.

In the nearer term, a focus will be aligning technological operations more closely with the manual contact-tracing programme.

“The role the app plays is that we can do contact tracing extremely cheaply – orders of magnitude smaller than the cost of manual contact tracing,” he says. “But there is a case for manual contact tracing and the app to work together, and we have built some of those mechanisms. For example, when you scan your QR code [to enter a venue]…. the app can detect outbreaks, without contact tracers having to go into the back end.”

Emmerich adds: “But there is also a need for manual contact tracing because not everybody will be able to run the app or have the technology to do so.”

Zühlke’s involvement in the programme began a year ago; between March and May 2020 it was awarded three contracts worth a cumulative total of almost £3m to work on the development of the original app, which was ultimately ditched. Unlike many deals entered into by government organisations over the last year, these contracts were awarded following a full competitive process in which the Switzerland-based firm beat a number of other bidders, Emmerich says.

NHSX led the process to develop the original app, which the government hoped would allow data gathered to be pooled in a centralised repository for study by public health authorities. This technology was ultimately scrapped in June after it was discovered to work only patchily on Android devices, and barely at all on iPhones.

The Exposure Notification system jointly developed by Apple and Google was then used as the basis for the development of a second app, which runs on a decentralised model, where data is only stored on individual users’ phones.

Emmerich says that the Apple-Google technology was not available until April – by which time work was already well underway on the government-led app.

“We provided independent field testing on the first app and also helped them assess the outcome of the first trial, on the Isle of Wight,” he says. “As much as we would have liked that to succeed, we came to the conclusion that it was in the best interests of the country to stop. This was no fault of the development team… it was not possible to fight [Apple’s] policy.

“The first app was stopped on our recommendation and… [we then] rolled out a national feasibility study and we developed the second app using the contract that was set aside for the first app.”

Working together

After the initial app was ditched, the programme to develop a second was moved from NHSX to then-recently launched Test and Trace programme.

Development work was conducted by a large team that included 70 staff form Zühlke, alongside civil servants, representatives of other suppliers such as Accenture, and various external parties that contributed along the way.

This included the incorporation of a QR code-generating feature developed by Rush Digital and the New Zealand Ministry of Health for use in the NZ Covid Tracer app.

“We also worked with organisations in Europe who use the Google and Apple API – such as NearForm, which built the Irish, Northern Irish and Scottish app,” Emmerich says. “And we reused code from Germany, whose app was built by SAP.”

He adds: “We brought policy into the development to ensure we worked together very closely with the policy team and user researchers. What is unique here is that we had to bring together a team within a day or two and build the first release of the app within six weeks… which is very unusual in a government development.”

When its initial contract with the Department of Health and Social Care ran out in September, the company was awarded a further £9m deal which is due to conclude at the end of this month.

Emmerich says that his firm’s engagement with the government is likely to continue for a while yet, as focus increases on driving further uptake of the technology.

Capacity is not a problem, he adds, with the Amazon Web Services environment in which the app is hosted allowing it to grow “more or less infinitely”.

The importance of ensuring even wider adoption is demonstrated by recent research from The Alan Turing Institute that found the app’s average of 4.4 close-contact notifications per positive case could have prevented as many as 900,000 extra cases this winter when compared with purely manual tracing. For every percentage point increase in the proportion of app users, the amount of infections is liable to be reduced by 2.3%, the research claims.

Emmerich says: “We are trying to encourage people to use the app because of the superlinear impact that has on cases… 56% of the population has it – but that still leaves 44%.”

This article first appeared in the March edition of Civil Service World

Read the most recent articles written by Sam Trendall - Home Office warns settled EU citizens of benefit cut-off

Share this page
Read next