The UK government’s strategy to contain the spread of the SARS-Cov-2 virus has, until now, relied on extreme social distancing, via stay-at-home orders (aka lockdown). Such measures aim to slow down the rate of infection (i.e., flatten the curve) to avoid overwhelming the health services, and to allow for treatments and a vaccine to become available.
Given the very high infection rate of this virus, on the one hand, and the lack of an effective treatment or vaccine, on the other, these lockdown measures need to run for a long period of time. However, extended lockdowns have significant negative impact on the economy. If people don’t go out, they buy less products such as clothing or fuel, and they spend less on services such as hospital, transport, beauty or leisure. In turn, this leads to business bankruptcy and unemployment. Extended lockdowns have negative public health effects, too, in terms of mental health, strokes, heart attacks or cancer symptoms, among others.
For all of these reasons, extended lockdowns are unsustainable in the long-term. That is why the UK government is now looking at deploying a contact tracing app, to support the monitoring and management of the COVID-19 disease. The UK is not alone. This type of apps has been trialled in South Korea, Singapore and Australia, while the EU countries have plans to launch one, too.
The growing interest in contact tracing apps has been accompanied by discussions about technical aspects of these apps. For instance, whether to use GPS or Bluetooth technology; or whether the apps would produce a worrying number of false positives because proximity is measured in crude terms, and doesn’t take account of factors such as the presence of walls between app users. There is also a growing number of articles discussing various governance problems associated with this type of app. For instance, the privacy risks of centralised vs decentralised data storage; or whether contact tracing could be used as a veil for generalised population surveillance initiatives.
Marketers’ voices are largely absent from this discussion, however. And, in my view, this is a regretful omission. In this post, I discuss why marketing scholars and practitioners need to engage with this discussion. I start by review the purpose of contact tracing, and the rationale for a COVID-19 digital contact tracing app. If you are familiar with these topics, you may wish to go directly to the section “Why the marketing perspective is important”.
Before we begin, a note on terminology. In this post, I am using the World Health Organisation’s notation: “SARS-CoV-2” refers to the novel corona virus; while “COVID-19” refers to the disease caused by the virus.
What is contact tracing, and how does it work?
Contact tracing refers to identifying people that may have been in contact with someone who has been diagnosed with a communicable disease, in order to alert them for the fact that they may have the disease, so that they can avoid passing on the disease to others and, if relevant, to seek treatment. Contact tracing also helps understand the spread of a disease, and can guide decision making, such as informing public health campaigns, or directing resources to the areas most affected by the disease.
Contact tracing is routinely used to control the spread of many diseases, such as sexually transmitted diseases, tuberculosis, measles, chicken pox, HIV and Ebola, among many others. It is usually done manually. A trained health professional interviews the infected person, and prompts them to remember where they were in the days or weeks before the diagnoses, and who they interacted with. The goal is to create a list of at-risk individuals, who are subsequently contacted in order to repeat the process of identifying places they have been to and people they interacted with.
This process works well when there is a small number of infected people, when there is a small number of potential contacts, and when there is a short time between being infected and being diagnosed. However, it has several weaknesses, such as :
- It relies on the infected person’s memory accuracy, and honesty. This is particularly problematic for diseases that are highly infectious, have a long incubation period, or suffer from social stigma.
- It requires quick and accurate diagnoses. In systems that only test individuals that developed symptoms, where there is a long incubation period, and where tests have low reliability, by the time contact tracing starts, several chains of infection might have already occurred.
- It is very time consuming. This is challenging for diseases that are highly infectious, and have spread to highly populated areas. It is also difficult when health professionals are over-stretched.
- It requires trained personnel for the interviews and the follow-up. These take time to recruit and train, which is a problem for cash-strapped health systems, and for highly infectious diseases.
- It is not anonymous and is done without the consent of the person being identified as a contact of the infected person.
- It could be an invasion of the infected person’s privacy. This is particularly the case when contact tracers have permission to check devices (e.g., mobile phones, public transport tickets) or services (e.g., ride-sharing apps, credit card payment records) used by the infected person, to obtain a more complete picture of the person’s behaviour.
SARS-CoV-2 is highly infectious, air-born and has a large incubation period. Moreover, it is a novel virus, which produces a variety of symptoms. In fact, some of those infected are asymptomatic, meaning that they may not be aware that they have the disease and may be infecting others.
Researchers at the University of Oxford modelled the proportion of cases who would need to be isolated, and the proportion of their contacts who would need to be quarantined, in order to bring the COVID-19 epidemic under control in the UK (i.e., to reduce R0 to less than 1), and found that “for a 3-day delay in notification assumed for manual contact tracing, no parameter combination leads to epidemic control”. In other words, manual contact tracing is insufficient to map and contain the spread of the COVID-19 disease.
The case for digital contact tracing apps in managing COVID-19
Given the limitations of manual contact tracing, and the fact that most of us use digital technology in our daily lives, various countries are experimenting with digital contact tracing approaches, such as examining data from electronic cards, location-based apps (e.g., ride-sharing), or developing specific apps (e.g., Australia’s COVIDSafe). Digital contact tracing apps can run on dedicated devices (e.g., GPS-enabled wearables), or on smartphones. Given the ubiquity of the latter, many developed countries have either launched, or are developing, COVID-19 digital contact tracing apps to use on smartphones.
Digital contact tracing apps purposefully developed for COVID-19 promise to be faster, cheaper and more reliable than manual contact tracing. They could “enable a week’s worth of manual detective work to be done in an instant” (Source: BBC). They are touted as one of the reasons why countries such as Taiwan or South Korea have managed to contain the epidemic.
However, it needs to be emphasised that this is a new, yet unproven, technology:
“At present there is very little evidence that contact tracing apps will be a success. This is because using Bluetooth for contact tracing has not been done before and the first apps to be rolled out are in their very early stages.” (Source: WIRED Magazine)
Despite the lack of evidence regarding their effectiveness, and given that the vaccine is still a long time away, these apps are increasingly seen as a key tool in the management of the epidemic, as part of a package of other measures.
How COVID-19 digital contact tracing apps work
How digital contact tracing apps work, precisely, depends on the type of app. Their design is still evolving, with some features being dropped and others being added. For instance, the app originally proposed by the University of Oxford for the UK included the use of QR codes to record locations, but this feature has now been dropped. The current version records proximity via Bluetooth Low Energy (BLE) technology and sends alerts. It is possible that future versions of the UK app may “include options for people to provide more information to the NHS, such as their exact locations, collected through GPS technology, of where their contacts with others took place.” (Source: WIRED Magazine).
Regardless of the specific features, the general principles of COVID-19 digital tracing apps are the following:
When a phone running the contact tracing app is within a pre-defined distance from another phone running the same app, for a set period of time, the two phones exchange tokens with each other via Bluetooth technology. When they do that, they are deemed to have been ‘in contact’ with each other. Later, if the user of one of the phones is diagnosed with COVID-19, and that information is entered on the app, an alert is sent to the users of the other phones which are deemed to have been in contact with that one.
The University of Oxford researchers estimate that, at least, 60% of the population of a country will need to download and use the app correctly, for it to be affective.
A simple but very effective description of how these apps work is available here (long article – Scroll down to “What this scheme will do (more or less, at the time of writing)”). For a full description of the app being considered by the UK government – and how it works – see this paper.
The apps currently being deployed or developed vary in two main ways: 1) what they track; and 2) how they store data.
In terms of 1) what they track, apps that use “absolute tracking” record the exact GPS coordinates of the phone at different points in time. So, whoever has access to the data can say where the app user was on day A, between times X and Y. In contrast, apps that use “proximal tracking” record which phones were in proximity of each other. So, whoever has access to the data can say which phones were in proximity of this one, on day A, between times X and Y, but not where this took place.
In terms of 2) how they store data, data storage can be fully centralised, fully decentralised, or use a hybrid solution, as described in this table:
The relative merits of decentralised vs centralised apps include:
(E.g., Google-Apple app)
(E.g., UK, Japan, France apps)
|Difficult to customise app, and change features.||Flexibility in terms of app design, and features update.|
|Any user information would be available only to the user themselves. This makes it better in terms of privacy, especially if coupled with periodically changing the device-ID (as planned for the Google – Apple app).||Data is held centrally. This can provide additional insight to Public Health officials, but is bad for privacy.
|Alerts are sent by the user’s phone. So, they can’t be easily adapted.||Alerts can be tailored by the app developed – e.g., in light of evolving scientific understanding of the disease, or based on additional information about the circumstances of the contact (e.g., presence of a wall).|
|Not vulnerable to central hacking (though individual phones may be hacked).||Likely to be a target for hacking.|
Why the marketing perspective is important
As this (not so) brief description of how contact tracing works, and what a COVID-19 app might look like, show, for such an app to work in practice it relies on a number of not trivial factors, and is likely to impact both those users and non-users of the app.
Marketers should join the discussion drawing on our work on adoption of mobile apps, and particularly health and fitness apps. Key issues include:
- Potential users: Bluetooth technology does not work on all smartphones, let alone all phones. What is the adoption rate of ‘the right type’ of smartphone in the country? Not just the overall rate, but the geographical distribution and the socio-demographic penetration rate.
- Potential adoption: Of those that have the right phone, who is likely to download and install the app? Again, what is the distribution of this rate across the country and socio-demographic groups?
- App use: Of those that download the app, how many (and where and who) will use it in conditions that meet the requirements for effective contact tracing? For instance:
- Will they use it all the time so that their contacts can be continuously traced
- Will they always have network coverage (this is likely in urban areas, but more challenging in rural ones) or access to broadband (less likely in socially deprived households)?
- Will they develop symptoms (many don’t)? Will they recognise the symptoms (new information continues to emerge about different symptoms associated with the disease – e.g., Covid toes)?
- Will they get access to testing (at the moment, in the UK, testing is only done if someone is admitted to hospital, or is a key worker)?
- And will users update the app with their diagnoses so that others can be alerted (they might not be well enough to do so)?
- Will other users still be using their apps so that they can receive an alert?
- Will they take the alert seriously, and self-isolate (unlikely, if there is a high number of false positives)?
- Will those users, themselves, have symptoms, have access to tests and update the app?
- The app’s system: The app doesn’t work in isolation. It only works if there is access to quick testing, and people are able and willing to self-isolate.
Another field where marketers can add value is in terms of effective communication. Clear messages are extremely important in encouraging public health behaviours. For instance, think of the effectiveness of the message to wash your hands for 20 seconds (or, even better, to do so while singing “happy birthday” twice). In contrast, think of the confusion generated by the government’s latest slogan:
In addition, we need to look at important questions around consumer vulnerability, for instance:
- Which citizens are likely to be excluded from using the app because of factors beyond their control, such as their social status, their level of comfort with technology, their economic condition or their location?
- Which citizens will be unable to use the app properly because they do not have access to testing?
- Which citizens will be unable to self-isolate due to their employment status or their living conditions?
While adoption and use of the app is, ostensibly voluntary, some service providers may limit access to their premises to only those that can demonstrate that they are not in a chain of infection. For instance, in China, “users were allowed to go into public spaces or on public transport only if they had installed” the app. Restaurants, offices, etc… might demand the same, in the interest of public health. It is, therefore, critical to consider which consumers / citizens are more likely to be at risk of being unable to access key services, or continue to work because they can’t use the app. Or which users are more likely to be classified “at risk” (e.g., because they can’t work from home, and they need to use public transport), and, therefore, find themselves regularly excluded.
These are some of the ways in which we – marketing scholars and practitioners – can contribute to the discussion around COVID-19 digital contact tracing apps. But I am sure that there are many others. What have I missed?