Tech Relief for the High Cost of Field Research

Client-centered research means, among other things, speaking to clients. This can be costly and time consuming. But when clients are in rural areas, it can also be impossible. Surveyors need to find locations that have no addresses, travel in various vehicles and on foot, all while trying to save money. Travel times and logistical routes can add to the complexity of the research. When we consider a randomized treatment, it can often exacerbate the logistical complexity as well as the costs.

The Microinsurance Centre is managing a mixed methods study, which includes a random assignment of two treatments to agricultural microfinance clients of Crezcamos in rural Colombia. One group is receiving an agricultural insurance product bundled with the loan (a traditional way of selling crop insurance), while the other is receiving the product at a later date, after the loan has been approved (or rejected). The purpose is to see whether the cognitive load, or amount of information that can be mentally processed, of a bundled vs. unbundled product may lead clients to take up or understand a product differently.

While it has been impossible to eliminate all of the complexity of working in rural areas, we have used technology to address some of these challenges as following:

Smart phones: We were lucky that Crezcamos issues company smartphones to all loan officers, who use this to GPS map their location when they sell agricultural insurance, call and text clients, and contact internal staff. This became the platform for our study, and the tools we would use to implement it.

The Randomizer: Loan officers often determine a route for marketing new loans based on where they have recently been implementing promotional activities, where referrals might have pointed them to, or where they are planning to carry out loan collections. These pre-determined routes are not conducive to our randomly selecting a treatment and control group from a list of clients in advance. At the same time, assigning a randomization such as selecting every other client, or client by first initial would be complex to track and verify. We used our “Randomizer App” (developed by EA Consultants and our IT partners) and loaded it onto loan officers’ phones. The application allows the loan officer to enter his/her name, the clients’ name and ID into the application, which provides a random assignment to each client at the time of the visit. It then uploads both the information and the random assignment in real time to a database in headquarters. This allows us to track the randomization and adherence on a daily basis.

The Video: One concern about our methodology was whether differences in the way loan officers marketed the insurance product would lead to differences in take up or understanding. It would be cost-prohibitive to audit a significant number of sales, given that a loan officer may make only 1-3 visits per day. To reduce the differences in sales techniques, instead, we designed a 3- minute video that explains the main characteristics of the product and offers a brief client testimonial. While the video was designed for the purposes of the study, Crezcamos plans to now adapt it to use for sales going forward. It saves time, standardizes sales, and offers clients some entertainment.

Phone vs. in-person surveys: Our study relies on in-person qualitative interviews, which entailed extensive travel to clients’ homes. However, these were done only for a selection of clients. To reach a larger number of our respondents, we used a local phone survey company that implemented brief surveys with short and mostly binary questions for end-clients. The company’s biggest challenge was locating clients by cell given the poor reception. However, calling numerous times was a much cheaper trade-off than traveling to their homes and finding they were not there, which happened a number of times during qualitative visits.

The intersection between research and business challenges in rural areas is significant as both require high touch visits that often involve longer travel time than the time of the visit itself. We borrowed from Crezcamos’ microfinance technology platform used for lending in rural areas, where a main server connects to loan officer smart phones, to layer on additional technology specific to our study. Here we were able to integrate video into a sales process, as well as introduce a “Randomizer” application to facilitate the study’s random assignment of treatments. It is well known that using technology in rural lending is critical to keeping costs down, and improving access to remote areas. We have found that this has become ever more critical in rural studies that can otherwise be extremely costly and challenging. We cannot apply technology without careful thought, however. Considering the trade-offs of technology is essential, as technology cannot eliminate a human touch. Using mixed methods in a study can help mitigate some of these trade-offs, by reducing the need to collect large samples of interview-sourced data. Phone surveys cannot replace a face-to-face interview, they must have a more narrow scope, which in the case if this study, we complemented with in depth interviews with a smaller number of clients. Without these interviews, it would be difficult to probe into the “why” or the reasons behind some of the behaviors that we observe, and qualitative open-ended interviews are a great way to get at these. Similarly, in offering rural financial services it is important to note that some “high-touch” is still needed. While the video informed clients about the agricultural insurance product, loan officers convinced clients of the specific value of the product to their crops, and offered pricing examples based on clients’ investments in insurable crops to close the deal.