Using Behavioural Insights to Improve Service Delivery

ETHOS Issue 17 June 2017

What is Public Service Delivery?

The Public Service does not choose its customers. It serves all citizens and businesses, continually carrying the responsibility of meeting diverse needs and creating a better life for Singaporeans. In a similar way, citizens and businesses sometimes depend exclusively on public services for their needs. This unique relationship sets public service delivery apart from any other service provided by other organisations.

Often, “service delivery” means providing a service efficiently, according to what citizens need at that moment — the quicker, the better. However, public service delivery has to go beyond quick fixes to satisfy immediate needs. Good “service” involves improving connections, building trust and enhancing experiences. For citizens, service experiences matter most at key milestones in their lives, such as marriage, childbirth, and retirement. In Singapore’s context, an engaged couple might be looking to purchase public housing, or families with children could be looking to the government for support towards living and education expenses. A positive and effective experience of public services during these important life-stages can help build trust in the Public Service and strengthen the government’s relationship with citizens.

“Delivery” describes the actual transaction process, which should be as convenient and easy as possible for the public to understand and use. Many service innovations have incorporated digital channels, including mobile technology, to increase the accessibility and convenience of light-touch services. Recent advances in artificial intelligence and automation could potentially eliminate even the need for many transactions, and render public services even more seamless and fuss-free.

A common challenge faced by the Public Service is striking a balance between delivering easily accessible services that meet our citizens’ needs, and improving manpower and fiscal efficiency to make the most of limited public resources. The draw of Behavioural Insights (BI) is that they have been able to help governments around the world do both — achieve substantial improvements in service outcomes, at relatively low cost. BI approaches have come to be increasingly important tools in designing public services. Governments around the world have successfully “nudged” citizens to contribute to desired service outcomes, such as paying their outstanding taxes or signing up for organ donation programmes.

Insights from these efforts challenge us to rethink how we can design and deliver better services from two broad angles: how to make services more salient, easy and convenient, and how to improve experiences and connections so that public services continue to be aligned with what really matters to our citizens.

Making Services Salient, Easy and Convenient

Making services salient and easy to access is key, because even the best-designed services cannot achieve their outcomes if they are underutilised. Most people are already preoccupied with things that demand their attention – family, work, and leisure time. How do we make sure that public services stand out from the daily “noise” and are easy to use, so that citizens can and will make the most of public programmes and services meant for their benefit?

In the healthcare sector, hospitals face the problem of patients who persistently miss their outpatient appointments, but who do not cancel or rearrange the appointments in advance. This creates wastage: time and resources that could have been used to serve other patients in need. The UK Behavioural Insights Team worked with the Department of Health and Imperial College London to increase the saliency and ease of arranging for appointments. They included a specific phone number in text messages for patients to use for cancellation and rearrangements. On top of that, the cost imposed on the system was highlighted in the message by stating, “Not attending costs NHS 160 pounds approx”. This redesigned message reduced the incidence of missed appointments from 11.1% to 8.4%.

In Singapore, which has one of the highest mobile phone penetration rates in the world, our ability to reach a large majority of our citizens through text messages means that we have the potential to scale up such low-cost interventions for many services. In fact, Khoo Teck Puat Hospital has replicated the UK’s hospital appointments trial. Their SMS message, stating “Missed appointments keep others waiting”, reduced the incidence of missed appointments by 6.9%.

Another important way of improving the ease of services is through digital channels. Traditional service channels, such as over-the-counter interactions, are labour intensive and require citizens to make the time and effort to visit physical service locations. Merely offering a digital alternative, however, is usually not enough. There needs to be a shift in people’s behaviour, which often “sticks” to the status quo. Experience shows that BI methods can effectively shift more users online, as seen in the Singapore Housing Development Board’s efforts to prompt online payment of upgrading fees.

Transforming Experiences, Making Better Connections

In the process of providing more channels and e-services to increase the ease and convenience of public services, the actual experience of the citizen can sometimes be compromised. What public agencies see as efficient and effective may not always be aligned with what citizens expect. To establish trust in the Government, the Public Service needs to be citizen-centric; BI can contribute to this effort by helping us to understand how people interact with our public agencies, make decisions and respond to different service modes.

For instance, providing job opportunities is not a simple matter of presenting information in a clear manner to the right target group. An unemployed jobseeker faces many challenges that need to be better understood before appropriate support can be provided. The Singapore Ministry of Manpower and the former Workforce Development Authority understood the need to connect with the jobseekers’ experience in order to design more effective employment services.

In another example, the Central Provident Fund Board (CPFB) found that engaging customers to understand their concerns helped in the redesign of communications for a retirement planning service, improving take-up rates.

Other kinds of pre-commitment and reframing devices have been used effectively by other public agencies to encourage citizens to sign up for programmes that are beneficial but not immediately salient to them. For example, the Ministry of Home Affairs worked with the Singapore Civil Defence Force, in collaboration with Harvard University’s Behavioural Insights Group, to encourage residents whose apartment block had just encountered a significant fire to participate in the Community Emergency Preparedness Programme (CEPP). Even though the CEPP was already well-subscribed to, those who attended were mainly from organised groups such as schools and companies, compared to members of the public who needed to pre-register to attend. Two BI-based modifications — message frames and follow-through prompts — helped nudge residents to participate in the CEPP. The use of a message frame, paired with pre-registration (allowing residents to sign up for a slot on the spot) was found to significantly narrow the intention-action gap and increase CEPP attendance among residents.

The Central Provident Fund Board (CPFB) in Singapore found that sending an invitation letter to all soon-to-be retirees to attend a newly introduced retirement planning service resulted in 15% of recipients making an appointment.

The Future of Behavioural Insights in Service Delivery

Technological change is rapidly transforming the business of the Public Service, creating new frontiers for service delivery. Citizens can now interact with organisations through digital channels such as chat bots, augmented reality, or virtual reality interfaces. The explosion of data generated and collected through service interactions allows organisations to customise and anticipate the needs of its customers through thoughtful use of analytics. This surge in capacity is matched by a rise in expectations of services that are faster, cheaper, and more intuitive. Citizens expect public services to perform as well as those of the private sector, be delivered in real-time and resolved quickly with minimal effort or disruption to their lives.

These changes present opportunities for the Public Service to be more sophisticated in how it applies BI to service delivery challenges. There is potential in combining a rich understanding of data with the implementation of BI. Existing descriptive data can be used to customise interventions based on customer profiles. We have already seen how we can encourage citizens to sign up for retirement planning services by providing personalised information. We could take this a step further by targeting and customising our invitations according to data about their age, employment status, income, and current level of retirement savings.

Looking forward, machine learning may yield algorithms that can predict specific outcomes based on a citizen’s characteristics. For example, we could one day predict the risk of re-admission of a patient upon discharge from the hospital, based on their demographic profile, symptoms, and treatment and discharge plan. However, the prediction of such ‘high-risk’ customers only points us to where the problem may be. It does not inform us on specific treatments that would reduce the likelihood of re-admission. How do we encourage better health behaviours amongst these high-risk patients? How do we provide information to healthcare professionals about these high-risk patients in a way that will support their medical decision-making? Answering these questions require a deeper understanding of user behaviour — including both patients and medical practitioners — to design effective interventions.

BI will not be able to resolve all issues with service delivery. The context of the problem will determine whether BI interventions are the most appropriate solution. Most importantly, we must constantly ask ourselves if we are solving the “right” problem. A BI-based approach allows us to achieve substantial improvement in service outcomes only when there are clearly defined policy outcomes. We should not use it as a blunt tool whenever we find gaps in service efficiency or effectiveness. We should always return to the policy intent, and question if it continues to be relevant. If our citizens are not engaging with the service, is it because the service is not effective at meeting their needs, or is it because the need has changed? It is only with such rigour and discipline, that we can design and deliver our services effectively, and continue to build trust in the Public Service.