Finding and Serving the Savings Frontier
As COVID-19 restrictions make it more difficult for informal savers to access financial services, using data to locate and serve potential customers in hard-to-reach areas is now more important than ever before.
Work by our Savings at the Frontier (SatF) team prior to the pandemic shows how the growing availability of satellite imagery, population and infrastructure data and free geo-location tools can help financial service providers (FSPs) close the proximity gap between clients and agents – without having to invest in substantial skills development or new software. FSPs have a real chance of greatly improving uptake and usage of their services by using satellite maps to better identify underserved areas and locate potential customers, as well as determine the most viable catchment zones for branches and agents within these areas.
Since 2015, SatF has been working with nine FSPs in Ghana, Tanzania and Zambia to improve services to savings collectors and savings group users through additional touchpoints (such as branches and agents) at the geographical frontier - areas that are currently financially underserved. This work has taken us on a winding path through many lessons learned but no obvious practical way of taking the underserved on board. All too often solutions to address the proximity gap that are put into the public domain need specialist software and skillsets to deploy them. SatF focused on developing practical applications of the learning and tools for expanding outreach and increasing usage that work with ordinary office software and widely available web apps.
Defining and locating the frontier
Initially, SatF and our partner FSPs struggled to gather sufficient data on outreach into frontier areas. FSPs argued that the frontier did not equate to the designated border between areas officially classified as rural and urban. For example, people living in rural market centers may not be at the frontier because they have many touchpoints nearby, whereas urban people in self-organized or peri-urban areas may well be at the frontier because they do not dare to enter nearby banking halls.
To overcome this obstacle to data compilation, we worked with partner FSPs to agree on a definition of “the geographical frontier,” based on a combination of settlement size, local population density, organization and infrastructure. But the challenge remained: FSPs did not know how and where to start locating the newly-defined frontier areas and felt it would be expensive and complicated to do so. There was also no organizational or regulatory incentive to invest in identifying the frontier.
So we worked with selected partner FSPs to design and test low-cost and practical methods for locating potentially underserved areas. These methods were based on standard Microsoft Office software and freely available census, satellite and crowd-sourced data, so FSP managers were able to apply them in-house.
Learning how to gather geo-coordinates for clients and touchpoints
As we looked more closely at why FSPs were reluctant to invest in identifying the frontier, we discovered that they lacked location data for existing touchpoints and clients. National and local surveys to collect geo-coordinates for touchpoints and clients have been fragmented and sporadic. Only one SatF partner FSP had information on the location of some of its own touchpoints, and no FSP had coordinates for its clients. The other partner FSPs believed that collecting geo-coordinates would be complicated and labor-intensive, requiring a dedicated budget and staff.
To dispel this belief, we introduced our partner FSPs to various free smartphone apps for collecting geo-coordinates. Thanks to apps like What3Words and Pluscode Finder, geo-location can now be established at no cost, by anyone, in an instant. It was an eye opener!
Identifying viable catchment zones for branches and agents
One further obstacle stood in the way of applying the method: FSPs associated the frontier with low usage and dormant accounts. This was largely blamed on the “proximity gap,” i.e. the distance between clients and touchpoints (such as FSP branches and agents). Rather than simply identifying frontier areas, FSPs also needed a method to identify spots in underserved areas that had sufficient clients in the vicinity to make financial service provision worthwhile.
To respond to this need, we showed partner FSPs how they could use publicly available mapping data and free-to-use web-apps to identify viable catchment zones for branches and agents within these areas. Not only do these simple approaches enable FSPs to measure outreach into frontier areas, they also enable them to increase their outreach!
In a recent FinDev Gateway webinar, Marvin Chibuye of Vision Fund Zambia demonstrated how his organization has used client mapping to identify locations where partner mobile network operators could place agents within greater proximity of its savings groups.
Next step: an automated tool to identify viable touchpoint locations
Even though these methods do not require new software or sophisticated analytical skills, the level of manual effort involved in using them can be cumbersome. We therefore engaged NIRAS, a Danish consultancy firm, to use machine learning to generate tailor-made maps from a range of satellite images and publicly available data, as well as to produce an Excel-based tool that FSPs can use to extract useful information from these maps.
To provide proof of concept, the tool has been applied to the Greater Accra region in Ghana. During the webinar mentioned above, Abednego Darko of DSS, a digital application system for susu collection in Ghana, showed how plotting the location of susu collector clients on the multi-layered map generated by NIRAS has helped his organization identify overlaps in collector routes, as well as opportunities for expansion.
Spreading the word
At a peer learning event with partner FSPs in September 2019, we presented the proof of concept for these methods and tools. Participants were able to try out the methods and tools themselves through practical exercises to locate frontier areas, collect coordinates and identify viable catchment zones.
Partner FSPs have found the presentations and exercises so compelling that they have all started collecting the geo-coordinates of their ISM clients. But not only that: they have also started taking over from SatF in pushing the use of these methods and tools. SatF is now looking for potential partners interested in co-funding the mapping of other regions and countries, as well as in developing an Excel plug-in that uses satellite mapping and machine learning to grade locations and overlay useful population and infrastructure data. The goal is for this functionality to become available to all practitioners in the office software they use daily.