Latest entries

Conversational Interfaces: Revenue Models

It is often easier and less costly for CI providers to white label their conversational interface to a revenue partner rather than go direct to customer.

CIs can be broken into two business structures:

Business to customer [b2c]:

These CIs (i.e., PIA, extend access to basic financial services directly to customers without the support and backing of a more recognized brand. Because of that, CIs have to build their credibility directly with the customer through either word of mouth or through advertising ( radio, TV, or Facebook ads). Under this business model, the CI would generate revenue through either transaction or subscription fees: the customer would pay to send money or to access learning modules.

We suspect that operating a CI through this business structure makes it easier to get the conversational interface online and ready to use but considerably more difficult to gain traction among customers. Though it is relatively simple to construct a CI, it is quite to difficult to convince a user that the conversational interface can solve one of their pain points and is worth trying, let alone using routinely.

Business to business [b2b]:

Under this business structure, conversational interfaces (i.e., Juntos, Arifu) license their product directly to a partner, who then redesigns the interface under their brand and specifications. The partner markets the conversational interface under their brand to their customers through different analog [e.g., brick and mortar advertisements, billboards, radio, and television] and digital [e.g., Facebook page, company’s website] means. And as part of the licensing agreement, the CI provider will likely need to meet different key performance indicators—from usage to the number of new subscriptions purchased to capacity building—r to receive payment.

Under this business structure, the CI provider leverages the brand recognition of the third-party partner to reach customers. This is significantly easier and more efficient than trying to reach customers directly but requires not only educating a potential partner about conversational interfaces but convincing them of their value proposition and return on investment.

Sign up for notifications of future blogs in this series

MoMo: The Orange Money Digital Assistant

This is post number 4 of the Conversational Interfaces Blog Series and has been prepared by Teller Technologies, Inc., with whom some of our research was conducted, based on their pilot with Orange Money Madagascar.

Two of every three adults worldwide are financially illiterate, meaning they lack the knowledge and skills to use their financial resources effectively for lifetime financial security. This staggering share of people is disproportionately concentrated in developing countries, where formal financial services and financial assistance programs are often lacking. Among the financially illiterate, women, the poor, and lower educated respondents are more likely to suffer from gaps in financial knowledge [1].

In Africa, there are very few resources available to access and learn about financial services. The resources that do exist, such as getting help from banks, present their own problems. Talking to a bank teller can be extremely intimidating for a number of reasons: the bank may be far from home, there is a lot of confusing paperwork and financial jargon, and you may feel judged for having a small balance. In addition to the many studies that support this, we have also seen this first hand while interviewing low-income people across Africa [2].

Our company, Teller, aims to tackle financial literacy by bringing financial services into the hands of people in Africa, through integration with their favorite messaging apps. We partner with financial institutions to launch an automated ‘chatbot’ that can simplify account opening, provide tips for first-time customers, and make getting help easier — all via messaging apps like SMS. N

Our case study covers the details of our pilot with Orange Money in Madagascar, one of the leading mobile network operators and mobile-money providers. Orange Money currently provides feature-phone enabled financial services to over 40 million customers, mostly in francophone Africa. Teller helped Orange Money launch “MoMo”, an SMS-based financial assistant that helps customers learn how to open an account, look up fees, check their balance, and access customer service.

The full case study is available here

A webinar featuring the Teller’s experienced in building bots for the emerging markets, an overview of their work with Orange Money Madagacar and a short WhatsApp based demo!

Sign up for notifications of future blogs in this series

Partnership Considerations for Conversational Interfaces

Conversational interfaces (CIs) are currently being deployed to help extend access to financial education and services across emerging markets with strong results. In one pilot in Tanzania, CGAP found that farmers who used Arifu’s interactive learning platform saved at rates five times that of farmers who did not use the platform. Although these findings are preliminary, they suggest these tools have a significant potential to change behavior.

While many partner organizations are excited by such findings, they often struggle to deploy a conversational interface because of their inability to answer two questions:

  1. Where to find evidence behind the value of using conversational interfaces?
  2. How to decide whether to build a conversational interface in-house or outsource to a third party?

Below are our thoughts on these two questions:

Where to find evidence behind the value of using conversational interfaces?

Multiple organizations told us that despite their curiosity about what conversational interfaces could do for their businesses, they were unsure of how CIs could be integrated and with what results. Many did not understand the inner mechanics of how a conversational interface worked, the design considerations needed when building for different demographics, and, more broadly, the value proposition they offered in practice. The lack of publicly available primers, case studies, and evidence exacerbated their confusion and misunderstanding, which could be addressed if more research reports were made available.

Reports such as this, as well as FIBR’s Artificial Intelligence: Practical Superpowers; GSMA’s Messaging as a Platform; Accion Venture Lab’s Can Chatbots Promote Financial Inclusion; and CGAP’s Interactive SMS Drives Digital Savings and Borrowing in Tanzania are important first steps in addressing the learning curve. The next step is to supplement these reports with in-depth case studies that capture the challenges to be addressed, the value proposition offered by a CI provider, and results observed thus far. This deep dive would not only familiarize partners with the different companies working in this space but also better orient the companies around how best to build, design, and implement CI for different customer demographics.

How to decide whether to build a conversational interface in-house or outsource to a third party?

Below are a few of the considerations organizations should factor into their choice:

  • Security considerations: As described in the previous blog posts, conversational interfaces can provide customers access to information, transaction services, and tailored advice. For the first category—information—the conversational interface can likely provide answers to targeted questions without needing access to personal information, such as transaction records or bank account information. But  the CI will probably need access to the user’s personal information to offer transaction services or tailored advice. This would require the user to submit the personal information over the messaging application, a potential security concern when managed by an outsourced entity or conducted over a third-party messaging application like Messenger or Telegram. To work around this, some use cases will support using a one time password for authentication—similar to PIA when using the interface to access M-PESA. But for use cases that require more in-depth information about the user, organization will most likely need to develop their  CI application in-house and in adherence with the company’s security protocols.
Figure 1: CI PIA requiring an OTP to transact over its application
  • Experience of the technology team: Through the many different open source tools (natural language processing engine; cloud storage; analytic software), non-technical staff can effectively build CIs. However, in some cases, the open source tools either do not work or do not exist for emerging markets. For instance, one company we spoke with built a customized two-way SMS gateway for one of their emerging market clients because the service did not exist in that country. In situations like this—particularly when the technology has not been used before—it may be appropriate to outsource the work to an organization with experience building and working with conversational interfaces rather than trying to build in-house.

As the field of conversational interfaces evolves and their value proposition becomes more clear, organizations will probably have less of a learning curve than those pioneering the technology have now. But to address and minimize the learning curve, reports such as this, along with access to grant support and technical mentorship, can help minimize potential risks while building capacity for organizations looking to trial conversational interfaces.

Sign up for notifications of future blogs in this series

The role of language in the construction of a Conversational Interface

As mentioned in the primer, using CIs to interact with customers through freeform African languages can be challenging given the limited amount of localized digital content to source.

Of the conversational interfaces we tested, it was consistently challenging to have a human-like conversation with the interface. Although the CI generally understood basic commands that follow standardized spelling and grammar conventions, it often could not understand or follow the cadence of a human conversation. Training a CI to understand syntax, sentiment, and jargon in languages that are well represented online requires sourcing enough content to understand different scenarios, which is both time-consuming and challenging. Given how new these technologies are in emerging markets, it is possible that the technical teams were not familiar with or did not fully understand how to build complicated natural language processing engines.

This becomes considerably more complex in emerging markets where either the user mixes words from their local language into their English-based conversation, or chats with the CI entirely in their local language. Moreover, the limited amount of content in local languages online makes sourcing the requisite content for the CI difficult, if not impossible, and renders the local language interaction ineffective. This means that, until there is a stronger presence of underrepresented languages online, it will likely be impossible to build in native languages in sub-Saharan Africa.

Example 1: Using buttons and dropdown menus to interact with Simply’s conversational interface

To account for these challenges, we have seen conversational interfaces structure the interaction around a combination of natural language, dropdown menus and image based buttons. As seen in the screenshot from Simply’s CI, the user can interact with the interface by using buttons or selecting options on the dropdown menu

In these cases, the natural language processing engine is used for very basic commands that are well understood and require an almost binary response, while the buttons and dropdowns are used for more complicated responses. Using these two features limits how much nuance the user can exercise in responding to a CI and ensures that the CI is able to provide an immediate and accurate response to queries that could otherwise get lost in trying to understand syntax.

Sign up for notifications of future blogs in this series

Criteria to consider when choosing between SMS and messaging applications when building a conversational interface

Though functionally similar, it’s important to understand how SMS and messaging applications differ prior to engagement.

When deciding whether to design a CI on SMS or on a messaging application, it is important to consider the following:

Ease of deployment:

SMS: SMS CIs must be able to integrate their product with the mobile network operator’s system. To do so, CI providers can partner either directly with the MNO or with a third party aggregator, like Africa’s Talking, that works on behalf of companies to push their service onto the MNO’s system. In either case, the CI provider depends on another organization to push their service to customers. Accordingly, the CI provider cannot control the delivery. Among the organizations we spoke with, many struggled with the lack of transparency around knowing whether their message was ever sent to the user and, if not, what the bottleneck was.

Messaging applications: Currently, messaging applications companies have different rules and parameters that CIs need to follow in order to build on their service. On one end of the spectrum are companies like WhatsApp that does not allow external, open access to the public API; on the other end are companies like Telegram that offer open and public access to their API. In the middle are companies like Messenger that review each CI built on their application and can deactivate the CI if it breaks any of their rules.

Target audience reach

SMS: SMS offers incredible reach to a large audience of potential customers— regardless of the type of phone used—in one geography that can be easily accessed by users through a shortcode and therefore does not require a data bundle to access. However, SMS cannot easily be shared across different geographies, limiting the number of potential global users who can access the CI.

Messaging applications: Messaging applications are relatively easy and simple to share among other users [regardless of network] through a link over social media applications, have a rapidly growing user base that tends to be digitally savvy and literate, and can easily reach across multiple geographies. However, despite the rapidly growing user base, the vast majority still do not own the feature or smartphone necessary to access a conversational interface. Moreover, WhatsApp, arguably the most popular messaging application in sub-Saharan Africa,does not have an open API, making it incredibly challenging to use the application as a distribution channel.

Service offerings:

SMS: By design, SMS CIs can offer only a limited number of features and services to the user. Each SMS is capped at 160 characters that cannot link to information outside the message and is limited to interacting with the user purely through  text ( not visual or audio). These restrictions make it difficult to offer a robust and dynamic experience for the user. However, its simplicity paves the way for a more instructional experience, particularly around building user competency in different topic areas, like financial literacy.

Sign up for notifications of future blogs in this series

Introduction to the Conversational Interfaces Blog Series

As the cost of accessing the latest in machine learning and artificial intelligence drops, more and more organizations are depending on technology to reduce the cost and improve the quality of core business functions. By leveraging this technology in interactive conversations via conversational interfaces (CIs), customers across different demographics can now receive current, guided assistance, whether they want to know more about the latest agricultural practices or new financial services.

With the advent of these technologies, financial service providers (FSPs) are  digitizing more of the support and services that extend basic financial services to new and future customers, especially to last-mile customers. These technologies—data dissemination platforms, data collection and management tools, alternative credit scoring platforms, digital payment platforms, and e-learning platforms —have the potential to transform FSPs in the following ways:

  • Increase the number of potential customers that can access their services by building solutions that leverage the reach of mobile networks;
  • Decrease the cost of extending services to new and current customers by using low-cost digital tools and services;
  • And improve the accuracy of customer service through automation by using chatbots and conversational interfaces to handle customer inquiries.

Over the past few months, the Mastercard Partnership for Finance in a Digital Africa (FiDA Partnership), delivered by Caribou Digital, has been investigating the role that one of these technologies—conversational interfaces—are playing to contribute to financial inclusion across Africa in three ways:

Over the next few weeks, FiDA will release the following blog posts highlighting our findings across different business considerations on the topic:

  • Key considerations when thinking about using a conversational interface: These three sub-posts, spread across a week, will highlight different design and revenue considerations for those building CIs.
  • Partnership considerations: This post will discuss the different knowledge gaps large organizations tend to have when considering using a conversational interface.
  • Future considerations: With different legislation and technologies emerging, this post looks at the role these might play in the future of conversational interfaces in emerging markets.

Sign up for notifications of future blogs in CI Blog Series

What can Nigeria learn from China’s digital economy? Insights on big data, mobile payments, and what China’s model means for financial inclusion in Nigeria

This blog is authored by Ashley Lewis, Investment Officer, West Africa and South Africa, Accion Venture Lab with support from Tunde Kehinde, co-founder of Accion Venture Lab portfolio company Lidya.

As I prepared for my first visit to China, I was curious about what I would discover and hoping that what I learned would help me in my work to build more inclusive financial systems in Africa. I would be joining a delegation of digital finance professionals in Africa, including Tunde Kehinde, co-founder of Accion Venture Lab portfolio company Lidya. The trip was organized by the Mastercard Foundation Partnership for Finance in a Digital Africa (FiDA)  to learn about the ecosystem in China and bring insights and inspiration to our workplaces.

Tunde and I both live and work in Lagos, Nigeria, home to over 20 million people — one of the world’s megacities. We were fascinated to see how China has galvanized their government, businesses, and citizens to adopt financial products and services and how they’re creating a more seamless financial ecosystem for all. During the trip, we took note of everything we discovered and we’ve distilled our findings into lessons for inclusive finance in Nigeria.

From big data to bigger data

Tunde and I were both impressed by the tremendous amount of valuable data sources companies could tap into for lending decisions. As we met with business leaders from companies such as JD Finance, Ant Financial, and WeBank (the first Internet-only bank in China) , we were surprised to learn that they leverage from 30,000  to 100,000 data points on consumers and merchants in their underwriting process. As a comparison, Lidya uses 100 data points, and that’s a pretty robust data set in a region where high-quality and consistent data can be hard to obtain. Despite their advances, one Ant Financial Executive we met claimed that they are “still in the small data age of big data”, and that in 50 years we will look back and laugh at the datasets we have at our disposal today.  

Payments as the underbelly of a rich dataset

During the trip, we saw how a strong payment infrastructure is crucial for collecting data to build successful customer profiles to lend to the underserved. China’s impressive data collection starts when customers make a payment. The QR code system has been widely adopted for transactions in China, making it extremely simple for consumers to scan and pay. The low cost of implementing QR code payment infrastructure at the merchant level has led to high adoption rates, rich data sets, and minimal fraud. In comparison, Nigeria  has two primary approaches to payment facilitation: point-of-sale (POS) devices and USSD payments. Unfortunately, high costs and and inconsistent internet connectivity, meant that POS devices are hard to come by or unreliable.

QR Codes for locker storage and JD Headquarters

Collaboration for advancement

One thing that Tunde and I both found surprising was the high level of collaboration in the Chinese fintech sector. We expected companies to be highly competitive, secretive, and unwilling to work together in their race to the top. Surprisingly, that couldn’t be further from the truth. Instead what we learned is that many share potential customer pipelines. These companies have started to specialize in finance offerings for clients at various income levels and businesses of different sizes, and they found that it was more productive to share a pipeline so each customer segment gets the most valuable experience from the optimal provider for them.

Tunde and I have both traveled a significant amount, but we both walked away transformed by our time in China. It was a thought-provoking trip — one that allowed us to see the tangible progress of a market that has made significant strides in fintech and financial inclusion. At one point I started to panic, wondering if any African nation, let alone the entire continent, could ever catch up. However, what I came to accept was that our collective goal is not necessarily to imitate China’s model but to forge our own path, centered on the needs of African consumers and small businesses. That path may look entirely different from China’s. There is no one solution for building a financially inclusive future, and each region and country will have their own challenges and successes along the way. While our visit to China inspired us, Tunde and I returned home most excited for what’s to come here in Nigeria.

Wrapping up the Evidence Gap Map impact insights series—The value of community and conversation

What we shared

The FiDA partnership launched version 2.0 of the Digital Finance Evidence Gap Map (EGM) in October 2018. With 55 studies examining 60 products, there are many insights to navigate. To show the types of analysis the EGM makes possible, we published a number of impact insights on a range of topics. These are the headlines of our insights series:

  • Insights from digital savings studies show promising results from coupling savings products with two-way SMS, mobile learning platforms, and client training. Various types of savings accounts—accounts with default contributions, locked accounts, and commitment accounts—were found to improve the savings behaviors of users. Additionally, one study forefronted the need for careful planning when transitioning from an analogue to a digital service, and another raised the issue of differential impacts of digital interventions between women and men.
  • The evidence on digital credit was limited compared to evidence on traditional microcredit. Yet insights for promoting ‘healthy borrowing’ surfaced. These included optimizing ways to frame and time SMS repayment reminders, making borrowing T&Cs more salient and accessible, and providing interactive educational content on financial literacy. However, beyond borrowing behavior, there is limited evidence for the longer-term impacts of credit.
  • With 18 studies, Person-to-Person (P2P) payments and transfers have the most fully formed, evidenced based pathway to client impact. The insights on this topic are numerous and often consolidating, allowing us to have more confident conversations on the impact of P2P. Numerous tests highlighted how users share risk and smooth consumption through quick access to remittances, while others qualified the ease of storing value and making and receiving transactions in privacy. However, money comes from somewhere and some studies also raised concerns regarding the pressure on those who do the sending.
  • Most recently, we reflected on sources of variance when testing for  impact, that is, each digital finance study tests the the impact of a product, in a country, for a population, via a channel. On the product side we noted that digital credit and insurance products, are under-evaluated for client impact. In terms of geography, learning has been concentrated in East Africa; while this knowledge may be transferable, there are far fewer studies from other regions. Lastly, we shared that studies that disaggregated impact on more excluded groups (women, lower income, less education) showed that different populations can experience different outcomes. This insight challenges an assumption that a given product will have the same effects across the board.

There are still many more insights to uncover

The EGM insight topics of FiDA’s mini-series represent only a small sample of what can be gleaned from the EGM. For instance, we did not develop product-level insights for digital products like Government-to-Person transfers (G2P), Business-to-Person transfers (B2P), or digital insurance There are reasons for this.

The main reason was the low volume of studies on these topics. Additionally, in the case of G2P and B2P, we observed that, of the few studies available, each tested different outcomes, in different ways, in different markets. The studies have not consolidated enough to derive actionable insights. The digital finance community is in a nascent stage of testing and learning about the effects of these products. And although these specific digital payments are a niche line of inquiry, we will continue to track new studies for those who work in this space.

Beyond a product-level analysis, the EGM comprises other learning, such as the types of product design and delivery mechanisms gaining traction, segmenting findings by population types and market level, or even methodological approaches to measuring digital finance impact. The EGM can be used as an entry point for all of these inquiries.

What would it take to conclusively prove that digital finance has development impact?

In October of this year, I spoke at the Mastercard Foundation’s Partnership for Financial Inclusion Learning event, convened by IFC. A question posed to the panel was “What would it take to conclusively prove that digital finance has development impact?

While I raised a few points on the need for funding mechanisms, quality partnerships between researchers and digital finance services providers, and the value of sharing null and negative results, my standout insight came from EGM analyses: No single study can prove a digital finance impact pathway. This is why synthesis documents are so incredibly valuable in advancing learning.

In the discussion I offered the following scenarios to illustrate the importance of continued conversations on the impact of digital finance:

Imagine the resources for 50 impact studies a year were available.

The digital finance community can only extract a certain amount of value from the 50 studies if the 50 studies are unaware of each other. This unawareness might mean that each study has different conceptualizations of digital finance, the effects it may have on certain clients, and how to measure outcomes; they might even focus on different assumed knowledge gaps rather than actual knowledge gaps. They would not be learning from each other.

Alternatively, imagine that, for the same 50 studies, there is an incentive to cross check and coordinate to ensure that each is in dialogue with the others, testing the real gaps, finding things that the other studies did not, consolidating previous learnings, and creating a mini-community that talks about what is working and what is not. Through open communication, the digital finance community can get far more than 50 studies’ worth of information from the same 50 studies. That is how the digital finance community will come closer to an understanding of the impact of digital finance.

The greatest impact we can have through resources like the EGM, is in using the knowledge of the digital finance community to make better choices on both testing impact and designing better products.

FiDA is publishing a series on insights derived from an analysis of the latest Digital Finance Evidence Gap Map (EGM) update. This is the sixth blog.

Previous blogs:

The studies in the EGM, represent our best knowledge of digital finance impact insights. New studies are ever emerging and thus the EGM will continue to evolve. If you have questions on the EGM, are interested in discussing research priorities, or know of relevant digital finance impact studies that meet the inclusion criteria, please contact

Four things to be aware of in the search for digital finance impact

In previous insight pieces on savings, credit, and, Person-to-Person (P2P) transfers, we synthesized what we learned from studies in the Digital Finance Evidence Gap Map (EGM) using a product lens. However, the product lens is just one perspective, and the digital finance impact landscape is more varied and layered than this. Here we share four factors that digital finance researchers should consider when testing the impact of a digital finance product.

Figure 1: Source of variance on impact

1. Digital finance is not one thing

Digital finance includes dozens of diverse products, and researchers need to disentangle each digital finance product from the broader category in order to deepen the digital finance community’s understanding of what changes each of these products catalyze in the lives of low-income users. When the 60 product studies within the Digital Finance Evidence Gap Map (EGM) were disaggregated, we saw vast differences among the products that have been tested to date. P2P payments and transfers account for 30% (18) of the impact studies, savings for 18% (12), and credit for 12% (7).

Figure 2: Number of digital finance products represented in the Digital Finance Evidence Gap Map

Results from studies on P2P transfers cannot contribute to our understanding of the impact of digital credit. Indeed, every product comes with its own theory of what changes as a result of interacting with it. To advance knowledge in the digital finance community, we need to identify the gaps in product impact knowledge and allocate resources to begin correcting it.

2. Product design and delivery varies

Digital finance products can be and are designed and delivered in many different ways. The difference with digital is the opportunity to enhance services, and hopefully outcomes, by integrating design and delivery elements. An analysis of the EGM confirms that product designers use a multitude of innovative digital, and non-digital, design and delivery mechanisms. The EGM shows that across the 60 products studied, 28 different design and delivery features were observed. On average there were two design and delivery mechanisms described per product in the EGM.

Figure 3: Number of design and delivery mechanisms used across 60 products in the Digital Finance Evidence Gap Map

This means we might expect different client outcomes from digital credit products with a mobile learning component versus digital credit products without. Even within the same category of digital finance products, how they are designed and delivered may differ, and these differences result in variance in user outcomes.

However, most studies are not engineered to show whether the driver of change was (1) the way they recruited clients, (2) the training content, (3) the price of the SMS, or (4) some combination of these elements. These design and delivery choices typically become a single data point in a study so that we learn about the sum of the whole, rather than the sum of the parts. This is fine for deployments but a challenge for meta-analysis and impact assessment unless these variations and distinctions are made explicit.

3. Markets vary

A product owner could test a similar product in two different countries and find that client outcomes diverge. Regulation, infrastructure, capital, and, of course, the client’s ability and need to engage with the product vary across markets. The EGM represents 24 countries, but Kenya accounts for over a quarter of these impact studies (27%) and East Africa as a whole for over half (52%).

Figure 4: Number of countries represented in the Digital Finance Evidence Gap Map

Unsurprisingly, Kenya, the site of much innovation in digital finance, is the most studied country. But we must acknowledge that we simply know less about other regions and countries and that much of our learning comes from a particular region. So, before we transfer learnings from Kenya or Tanzania to other contexts, we must carefully consider the various social, economic, and cultural differences involved.

4. Clients vary

We could also test the same product in the same country with different client segments and find different results. One question to ask is, with which population segments are we testing various digital finance products? From a review of the 60 products in the EGM, the user sample has been concentrated on a mix of urban and rural clients (30%; 18), followed by exclusively rural clients (22%; 13). Fewer studies, 12% (7), have focused on smallholder farmers and, 8% (5), on women.

Figure 5: Number of client categories tested

With whom we choose to test our products matters. Just as a specific market focus can limit our generalizability so can the client sample. This is not necessarily a bad thing: in the context of digital finance, a number of outcomes are unlikely to be experienced equally by all clients. Often, some people benefit more, others less. But we only know when we test for it. See FiDA’s Snapshot on How do advances in digital finance interact with dynamics of exclusion? for more details. A quarter of the studies in the EGM looked for, and highlighted, differential effects on some level within a market based on gender, location, income, or education as described below.

Cases in which the marginalized benefit more from a digital finance product

  1. In a study of M-Pesa in Kenya, results showed that M-Pesa users did not resort to reducing consumption as a coping mechanism when they faced a negative economic shock. Further, the effects were shown to be more evident for the bottom three quintiles of income distribution than for the top.
  2. A study on mobile money in Burkina Faso found that individuals living in rural areas were three times more likely to save with mobile money than those in urban areas. Additionally, women were six times more likely to save with mobile money than men, and less educated individuals were four times more likely to save with mobile money than higher educated individuals.
  3. A Kenyan panel study on M-Pesa reported that mobile money access reduced both extreme and general poverty and that the effects were more pronounced in women—women being affected more than twice as much as the average.
  4. A study in Tanzania examined the impact of P2P transfers in the event of a [weather related] shock. They found that the benefits of other people using mobile money in the community come largely from rural areas where, with 1/3 of the village using mobile money, household benefit by a 10% increase in consumption. In urban areas this effect is only 3% and not significant.

Cases in which the marginalized may benefit less from a digital finance product

There are also cases in which the benefits of digital finance accrue to high status, high skill individuals, rather than to marginalized populations.

  1. CGAP partnered with the Busara Center for Behavioral Economics and Jumo KopaCash—a mobile money marketplace that offers digital credit—to measure client responses to various repayment reminders. The messages’ results varied by demographic with positive effects on repayment for male borrowers and negative effects for female borrowers.
  2. In a panel survey looking at the savings behaviors of M-Pesa users and non-users in Kenya, 65% of M-Pesa users reported having some savings, compared to 31% of non-users. The results showed that those most likely to have savings are male, live in rural areas, and have higher levels of education and income. In terms of the value of savings, those who registered with M-Pesa save 12% more than those not registered. Furthermore, the survey showed that those who are male, urban, and have higher levels of education and income tend to have more savings.
  3. A study on the effects of ATM cards on bank account use in Kenya found that while ATM cards increased account use among male-owned and joint accounts, a negative impact was found on female-owned accounts.The hypothesis was that household pressures to share savings drove women to stop using their accounts when an ATM card reduced the costs of accessing their money.
  4. Testing the effects of a roving Point of Service (PoS) on savings account use in Malawi, researchers observed differential levels of impacts based on wealth levels. For households at the top of the wealth distribution, an increase in savings services was associated with less reliance on distressed asset depletion to cope with economic shocks. However, the effect was the opposite for those at the bottom of the wealth distribution.

These observations offer the digital finance community a refined understanding of impact by determining the conditions under which impact is present, or stronger or weaker. They also underscore the need to examine how a digital finance product interacts with and affects various excluded groups.

The takeaways

  • Digital credit, particularly insurance products, are under-evaluated for client impact. While a growing community of practitioners are working to address gaps in credit knowledge (for example the Digital Credit Observatory, CGAP, FSD Africa), this is not yet the case for insurance. Digital insurance, a nascent product, has primarily, and understandably focused on business model testing. However, as a starting point, qualitative work should begin examining the insurance impact theory.
  • Much research has looked at what does and does not work in terms of product configurations and campaigns, but few impact studies are engineered to pinpoint whether the recruitment strategy, pricing, messaging, or something else that contributed to a given change.While thousands of choices go into product design and delivery, they typically become a single data point. Without understanding this nuance, it is difficult to discern exactly what we are learning.
  • The learning on digital finance impact is concentrated in East Africa. While this knowledge may be transferable, there is a notable absence of studies that focus on West Africa and Southern Africa.
  • Digital finance studies that have examined impact on more excluded groups show that impact is not homogenous. While a few studies specifically define a group of people, we lack research that disaggregates client segments. Acquiring this knowledge would advance our learning on product design and user needs.

Although these nuances may seem obvious to some and overwhelming for others, it is important to recall that while the impact landscape is complex, the studies themselves do not have to be. There is more than one way to gather impact insights, and if the digital finance community is going to be inclusive, we must ensure that we are expanding the conversation on impact to include, in an appropriate way, voices and signals beyond those that can be gathered by experimental methods (See FiDA’s Approaches to determining the impact of digital finance programs for more details).

To continue building the evidence base, the broader community needs to be more involved in contributing insights so that we can have active conversations about impact. Every study that targets a gap— in a product, a market, a design and delivery mechanism, or a client segment—brings the digital finance community closer to understanding the various impacts of different digital finance products and services. Developing an evidence base is like building a puzzle, no individual piece can reveal the picture, but, bit by bit, the picture will emerge as new pieces are added.

FiDA is publishing a series on insights derived from an analysis of the latest Digital Finance Evidence Gap Map (EGM) update. This is the fifth blog.

Previous blogs:

The studies in the EGM represent our best knowledge of digital finance impact insights. New studies are ever emerging and thus the EGM will continue to evolve. If you have questions about the EGM, are interested in discussing research priorities, or know of relevant digital finance impact studies that meet the inclusion criteria, please contact

Fiction or future for platformization? The implications we explored with banks, fintechs, and MNOs

Participants at our event exploring different platformization scenarios.

Platforms are a hot topic and for good reason. The digital “platformization” of markets is one of the defining forces of change in the shift to digital economies. Facebook and Google have leveraged network effects in social media and search to grow massive multi-sided markets in attention and advertising. Innovators such as Upwork, Uber and Kuhustle are revolutionizing the world of work. Amazon, Alibaba and Jumia are fundamentally changing how merchants sell to customers. These platforms are not only transforming online experiences but have real-world implications for users and the markets in which they work.

Platformization presents both new possibilities and new challenges for financial and economic inclusion in Africa and we had a chance to talk about this in new ways with a mix of dynamic startups and larger institutions active in the space. In early October, we organized an event alongside the IFC conference held in Dar es Salaam, Tanzania, where we invited leading representatives from MNOs, fintechs, and banks to explore the business strategy and financial inclusion implications of platformization. While some platforms bask in user bases in the millions and billions, astronomical transaction velocity, and stickiness so effective that people have to consciously tear themselves away from the screen, most African financial service providers struggle with low customer uptake, thin usage, and a dearth of digital data.

There is no doubt, then, that certain platforms can dramatically change the landscape for traditional financial institutions, presenting new opportunities for collaboration and competition. Likewise, some platforms will also likely need to make a shift towards thinking more about financial inclusion, because of their pivotal role in the digital economies of the global south and because their success relies on the ability of users to seamlessly transact.

In truth, platforms and financial services providers in this digital community probably need each other and this became a theme of discussion among participants at our event. Representatives from JumiaPay, JD Finance, Mastercard Foundation, Uber, and the World Bank, among others, participated and helped shape the group’s understanding of platformization. The event was interactive at its core – an important element to stimulate the kinds of discussion needed to more wholly understand the opportunities and challenges platformization presents.

One lens, six scenarios, and endless implications

We first introduced the FiDA Partnership Lens for Platformization – a framework to move forward through the sometimes confusing landscape of platformization.

Jonathan Donner introducing the FiDA Partnership Lens for Platformization.

We then introduced the participants to six fictional scenarios that they, in groups, would have to navigate – wearing the hat of MNO, bank, or fintech. These scenarios – while fictional – we think are possible.

The fictional scenarios we dropped these MNOs, fintechs, and banks into were these:

1. Facebook’s WhatsApp launches Whatsapp zero-fee payments and transfers in sub-Saharan Africa, enabling customers to link bank accounts and mobile wallets to the service. Whatsapp begin to recruit merchants who will offer payments through a QR code system.

2. Indonesia’s motorcycle delivery service Go-Jek enters Kenya offering a wide range of services from transportation and courier services, to mobile payments and food delivery.

3. Walmart takes Flipkart (the Indian e-commerce platform in which it owns a 77% share) to Africa.  In a bid to ensure prices remain competitive, it introduces low-price private label products which put pricing pressure on small merchant businesses.

4. Three local labour platforms – similar to Dumaworks, Taskit, Kuhustle and Lynk – consolidate to overcome fragmentation in the labour market. The new, consolidated platform, called WorkWork, begins to develop other product offerings to enhance revenues, by partnering with several banks and fintechs.

5. WeChat pens a deal with OEM, Transsion (manufacturer of Itel, Tecno, and Infinix), enabling WeChat and WeChat Pay to be enabled on any Android device in five influential African countries.

6. Regulation requires online platforms and marketplaces to obtain banking licenses, or partner with banks / insurance providers, etc. in their quest to enter the lending, insurance and wealth management space. Jack Ma, co-founder and executive chairman of Alibaba Group visits Kenya, Nigeria and Rwanda in search of potential partnerships.

Each scenario above had a series of events that played out as participants thoughtfully moved through business strategy during the first session, and how they moved through financial inclusion implications in the second session, guided by a facilitator. These events impacted the way each representative responded to the others in the group and brought a level of real-worldness to the scenarios which was appreciated by participants.

As Peter Zetterli from CGAP put it:

“To have conversations that were quite concrete in nature – talking about specific scenarios was useful in teasing out some concrete questions and answers from what could otherwise be a fairly abstract and hypothetical conversation.”

One conclusion that came out of these conversations, noted by Sid Garg, founder of fintech startup Teller, was the realization that while platforms present an opportunity, it doesn’t mean that players can ignore local market context.

“The opportunity that platforms present – especially for small players – allow something like Teller, or another small startup, to scale really quickly but one of the things I noticed at the event was that you have to adapt to the regional and cultural characteristics of each market. You can’t just trust that a platform will give you the access and skill you need – you still need to do a lot of work to adapt your products.”

For many, this event highlighted that platformization is already underway with WeChat and WhatsApp, and could see these scenarios play out in the next six months. The prospect of WhatsApp turning on payments in several countries in Africa, or even all of them, isn’t far fetched and could have great impact on the business models of incumbents in the space, therefore potentially impacting financial inclusion. How companies adapt their strategies to platformization will likely be key to their ability to thrive in the shift to a digital economy.

“Before, as a platform founder, I felt my job was done when I connected both sides of the marketplace but I do feel strongly now that there’s a value you can add on both sides of the marketplace by adding financial services.”

Sam Gichuru, founder of Kuhustle

Over the next few months FiDA Partnership will conduct research to critically assess platforms’ prospects for delivering financial services and furthering the economic activity of underserved individuals, small businesses, and developing economies. We will explore how platformization is affecting partnerships with bearing for financial inclusion and impacting the livelihoods of micro and small enterprises.

We have shared our analytical framework and have started the conversation considering these themes with key players in the spirit of collaborative exploration. We’re thankful to those who participated in our event last month, and we thank you for being a part of this discussion as we move forward. Stay tuned to our blog for further updates on our platform research and please reach out with any comments or questions. This is a conversation that should involve all of us, and we look forward to hearing your input.