Key Lessons from Trialling Our #TabunginAja (#JustSaveIt) Behavioural Intervention Campaign for Improving Financial Inclusion


A comic strip shared via WhatsApp during the #TabunginAja campaign

Pulse Lab Jakarta and the Behavioural Insights Team (BIT), in partnership with the Secretariat of the Indonesian National Council for Financial Inclusion (S-DNKI) and a private Indonesian bank undertook a year-long research, focused on fusing behavioural science and human-centered design (HCD) to advance financial inclusion in Indonesia. The research resulted in the trialling of a behavioural intervention in the form of an 8-week WhatsApp campaign, aptly called #TabunginAja (translated as #JustSaveIt). The call to action urged banking agents (who are typically trusted community members running a shop or food stall) to encourage customers to save their change after making purchases. The research report is now published, but unpacking everything from it would need another blog or two. We thought we’d therefore share our key lessons and the full report itself — happy reading!

The policy challenge: encouraging usage of bank accounts

Financial inclusion, the state in which individuals have access to and actively use a range of quality, affordable, and convenient financial services, can help economically vulnerable communities avert the impact of financial shocks, as well as assist them in becoming more resilient following environmental hazards. Agent-based banking in particular has been useful in lowering both the perceived and real barriers to opening and using a conventional bank account, such as what some persons might deem as the intimidating procedures and requirements of formal banks. Agent-based accounts rely on agents, which as we’ve described above tend to be trusted community members who often run a shop or a food stall as their primary source of income. The roles of these agents include rendering assistance to neighbours in signing up for accounts, performing bank transactions such as deposits and withdrawals, and providing general financial services advisory and support.

Whilst banking institutions sometimes succeed in attracting the unbanked segment of the population — individuals with low income who usually work in the informal sector — for true financial inclusion to occur, the financial services they offer such as bank accounts need to be useful and accessible to individuals who own them. Despite 56 per cent of Indonesian adults owning a bank account in 2018, 30 per cent of them had not used it in the previous year. Our research project therefore aimed to encourage bank account usage, more specifically regular deposits made into agent-based bank accounts. Various studies on the economics and mechanics of the agent-based model of the banking system Indonesia have been conducted, yet due attention has not been given to the potential of harnessing behavioural insights and human centred design to further improve financial inclusion.

Behavioural insights, which draws from disciplines such as psychology and economics, examines how people behave in practice as opposed to how they should behave and designs and tests interventions to encourage target behaviours. Human-centred design, a problem-solving approach that focuses on the experiences, emotions, and rapid prototyping with service users, on the other hand can help to quickly develop and test an intervention with end-users. We’ve previously shared our thoughts on the opportunities of combining both approaches in a blog post reflecting on a workshop we ran at the invitation of S-DNKI in 2019. Our Banking on Fintech research revealed that agents who encourage smaller, more frequent deposits are more successful in getting users to be more active in using their bank accounts. This thus formed the basis of our research question: since agents are mostly neighbourhood shop owners, would sending a message urging them to encourage their customers to save the change from purchases at their shop help in forming a new habit of bank account usage?

The intervention: #TabunginAja (#JustSaveIt)

To test our ideas in the real world, we partnered with a large Indonesian bank that offers agent-based banking. Together with S-DNKI (which provided policy advice), and the bank partner (that assessed the feasibility of our ideas based on their experiential knowledge), BIT and PLJ developed an 8-week WhatsApp campaign aimed at disseminating a set of key messages to bank agents. The messages, consisting of graphics with short messages, comics, and written text, were sent to bank agents, urging them to encourage their customers to save the change from purchases made at their shops or street food stalls.

The messages shared as part of the #TabunginAja (“#JustSaveIt”) campaign were based on a number of insights from behavioural sciences, for instance related to setting specific goals and applying rules of thumb to prompt the target behaviour. To come up with the messages, we sketched five formats, tested them with agents in two locations in Indonesia, and refined the messages based on feedback received. Upon surveying the various communication channels used by the bank, we thought WhatsApp might be the most effective medium to deliver the messages given its success in reaching the majority of agents. Following six weeks of prototyping and testing, we began trialling the intervention in March 2019. This was done as a cluster-randomised controlled trial which is described further in the report. To complement the quantitative analysis, we also conducted an in-depth qualitative evaluation to assess the implementation and perceived impact of the #TabunginAja campaign.

Conducting user testing with agents to refine messaging for the campaign

The qualitative evaluation indicated that the implementation of the campaign was labour intensive with insufficient resources needed to reach all the agents. Among the agents who received the messages, the reception was positive — indeed, some agents were already actively encouraging customers to save their change or had previously done so. However among the agents interviewed, for those of whom the idea was new, none of them actually followed through with encouraging customers to save their change as a result of the campaign. This was in part due to structural barriers, such as a lack of liquidity among agents. The results of the quantitative evaluation also showed that customers in the intervention areas were less likely to make a deposit in May 2019 than those in control areas, with a 10 per cent difference. We believe, however, that this is due to randomisation failure as deposit rates had already begun to diverge between the Treatment and Control areas before the start of the trial, and it seems highly unlikely that a relatively light-touch intervention would have such a large impact on the outcome of interest.

Again, for further discussion on our intervention, evaluation, and recommendations, here is where you can read the policy brief and full report.

The lessons: our reflections

As we were not able to determine the overall impact of the campaign and the fact that its implementation required an adequate amount of budget and human resources to send out the messages, we’d reserve recommendations on its roll-out across the agent-based banking system in Indonesia. In fact, BIT reminds us that less than half of all behavioural interventions are successful in generating the desired behaviour change — and that this is probably the case for a lot of public policy interventions that are never rigorously evaluated. Therefore even if the results were not quite what we had hoped for, the #TabunginAja campaign provided several useful lessons that might be adopted for advancing financial inclusion in Indonesia.

Here are some joint reflections from PLJ, BIT, S-DNKI, and our bank partner.

Enable testing before scaling

In the initial stages of the project, we aimed to narrow down our ideas to an intervention that could encourage customers to use their bank accounts more actively via agents. We created a long list of ideas; chose the most promising ones based on a set of criteria (that included high potential impact, low cost, use of existing systems, and implementability and to scalability if successful); developed and tested prototypes with some agents; and finalised the messages based on feedback received. This process helped us to come up with an intervention that has high probability of success, and the criteria we established made it easier for us to come up with an intervention that would be highly likely to be adopted by policy makers should trialling be successful.

#TabunginAja utilised WhatsApp as a channel for communicating with agents, who would encourage customers to make more frequent deposits and save their change when making purchases. Though WhatsApp was deemed the most suitable communication channel for reaching the target groups, the roll-out also demonstrated that the manual process of sending out WhatsApp messages was labour-intensive and sometimes inconsistent. This further highlights that even if a system has been used for a while, it is worth examining whether it is functioning as intended and what can be done to improve its effectiveness and efficiency.

On the other hand, we received positive feedback on the content of the messages, such as its success in helping agents to explore new ways of encouraging their customers to save, and reinforcing existing agents’ beliefs and practices about sharing information about savings. We learnt that agents paid more attention to messages that were short and visually-appealing, as opposed to 6-panel chat boxes that were perceived as too long and videos that required sufficient internet data to be downloaded. The latter ideas were both abandoned.

The lessons from the trial roll-out pre-empt the potentially costly mistakes, such as spending considerable amount of time creating unnecessary video content or overly labour-intensive WhatsApp messaging, that could be made should this intervention be scaled up widely. Yet in another sense, the lessons allow for an iterative approach and suggestions to improve existing communication channels to find the most effective strategies to encourage the desired behaviour.

Uncover hidden enablers and barriers

Behavioural insights interventions can be powerful, but are likely to be effective only if appropriate systems are in place. Using a behavioural insights and human-centered design approach to identify and address such barriers has the potential to lead to a more effective and fit-for-purpose agent-based banking system.

In the case of #TabunginAja, the process of rolling out the intervention illuminated several enablers and barriers for agents to pass along the message, and customers to save their change. For example, agents are already familiar with the practice of assisting their customers with savings transactions and draw upon that experience, but the qualitative evaluation of the intervention also revealed that agents are often inhibited by structural barriers such as issues with liquidity. This finding is in line with other studies on agent banking in Indonesia, which offer recommendations to the government to allow third-party agent network managers to help banking institutions better manage their agents.

Leverage existing channels and reduce friction

Introducing new ideas in the hopes of changing behaviours often creates some friction for the intended beneficiary or the people who have to deliver it. Applying behavioural insights is inherently about removing or reducing those frictions. One way of doing that is by using existing channels (such as methods of communicating, existing interactions, etc.) and introducing the behavioural intervention via those avenues. This ensures that the interventions we design are cost-effective and feasible for organisations to implement.

WhatsApp was chosen as the communication channel for the intervention because it is already regularly used by agents to communicate with bank personnel. WhatsApp messages are cheaper and easier to send out than physical posters (which often remain unused in an agent’s shop) and are more likely to be seen than going through the mobile banking app (which has low engagement from agents). Another way of reducing friction is through understanding when a behaviour change is most likely to take place and using those breakpoints to build a habit for the desired behavioural change. For #TabunginAja, that meant ideally targeting new customers who were more likely to form a savings habit, and because agents would be more willing to try something new with new customers.

What’s next?

Through this collaborative work, we learn the value of designing interventions that can close the gap between responsiveness and action — and test whether they successfully do this through actual behavioural data. We also see that the combination of HCD and behavioural insights has the strength of breaking down large scale challenges and testing solutions that have the human experience as the focal point. Even behavioural interventions that do not result in the desired outcome may still provide important lessons on the importance of conducting testing before scaling, uncovering hidden enablers and barriers, and leveraging existing channels to reduce friction in introducing new ideas.

In 2020, PLJ and BIT are planning to further research on the combination of HCD and behavioural insights to improve financial inclusion and address other development goals. Stay tuned for more news, or ping us if you’d like to collaborate!

Pulse Lab Jakarta is grateful for the generous support from the Government of Australia



UN Global Pulse Asia Pacific
United Nations Global Pulse Asia Pacific

UN Global Pulse Asia Pacific is a regional hub that aims to drive data innovation and sustainable development to ensure that no one is left behind.