Data collection methods along the Last Mile often face significant barriers in gathering reliable, transparent and accurate data. Focus groups, surveys, questionnaires, interviews and other baseline studies form a vital part in any company’s entry into a new market or geographic area. Yet along the last mile targeting bottom-of-pyramid customers is often where companies struggle most. This is further exacerbated by limited or no publicly available secondary data, such as reliable census, parish records and electoral statistics. These challenges can vary significantly including logistical, financial and safety constraints depending on the climate, landscape and even season.
1. Logistical issues
A major barrier to conducting in-person data collection, particularly in sub-Saharan Africa (SSA), is the issue of population density. Africa remains one of the continents with the lowest population density (at approximately 87 people per square mile) and thus collecting data efficiently and methodologically proves difficult. Despite the increased use of sales agents, conducting interviews or focus groups can be time consuming (typically lasting 1–3 hours) if considering multiple communities dispersed over a significant area. Additionally, if conducting data collection on a tablet or smartphone, data connectivity issues may impede the download of surveys and the upload of responses. Respondents may also be uncomfortable with answering sensitive questions, such as those related to income and profession without a prior relationship with the surveyor.
2. Data quality
Collecting data in foreign countries may encounter unexpected hurdles. For example, a translation service may be required to translate surveys into a local language; this is complicated further by the sheer variety of languages spoken by different ethnic groups. In Africa, for example, it is estimated that there are over 2,000 languages spoken by different communities and ethnic groups (Nigeria alone has over 250). Similarly, inferring and interpreting responses must also be done in the local language, meaning that often a local translator may be needed. Lastly, if conducting interviews, it is difficult to write detailed answers at the same time; thus recording hardware and software is required to record the interview, which then must be transcribed later.
Secondly, the skillset requirements for conducting data research are significant. The interviewer must understand the business model of the company in order to ask the correct questions of the end-user. They require good language skills, in order to understand nuanced and detailed answers, and must know when and how to probe the participant further (for example, if answers are inconsistent, short or irrelevant). In addition to this, they also require significant IT skills as well as being able to negotiate the interview plan. Thirdly, quality-checking survey data is an important process, usually reserved for a regional or survey manager. This helps address any irregularities in interviewee responses and can lead to a reformulation of interview questions if necessary. This will require employing someone with managerial experience that understands how data must be collected while overseeing call centre staff or agents in the field. Lastly, a great emphasis must be placed on appropriate survey design. Not only does this mean asking the correct questions (for example, structured or open-ended) but they must create tangible data that can be compared across regions or countries.
3. Safety issues
The deployment of agents to conduct surveys and other research methods raises a number of safety concerns. Firstly, agents may be at risk of theft or robbery if they are known or seen to be carrying tablets and smartphones; this is of particular relevance of agents visiting areas far from their headquarters in unknown territories. Agents must be aware of existing ethnic or community tensions and security concerns that exist in an area before visiting. Often those who require off grid products the most reside in conflict-prone areas (where reliable data is most difficult to obtain).
A second safety concern is the condition of infrastructure particularly during wet seasons. Roads may be muddy which may impede accessing a research site by tuk-tuk or motorcycle. Sporadic weather events, such as flooding or landslides, have a higher chance of occurring during the wet season and thus precautions must be taken to ensure safe access to and from the research area. A key cost of data collection is travel, with many public transport options lacking suitability or reliability forcing data collection teams into expensive private vehicle hire.
4. Cost issues
As alluded to in the previous paragraph, deploying sales agents is costly for businesses. This may relate to fuel and other transport costs, as well as the cost of equipping each sales agent with a tablet or smartphone. Deploying sales agents over even a small area can be an expensive process, particularly if follow-up interviews or surveys must be administered, or if the intended participant is occupied or unwilling to participate.
5. Agent Training
When talking about digitalisation, its necessity is often obvious for the distributor. To scale, it will need to transform and improve its procedures, it will need to build capacity. Onboarding office staff suffering from slow and tedious processes is straightforward; the same cannot be said about the field agents who can easily feel overlooked in the process. Assuming they will automatically get on board is a common oversight – especially as they are likely to be the most impacted by this transformation. We spoke about this challenge and how to make sure that agents are not left behind in the previous article.
Upya Collect
With Upya Collect, our offline solution for data collection, you will improve project KPIs and tackle the challenges by implementing the best practices and trends of data collection using all-in-one place digital tool.
With Upya you can address the full cycle: from discovery and onboarding surveys to aftersales and audit, always having respondent profiles updated with the latest data and events. Also, eliminate the risk of double counting, ensure data accuracy, and facilitate seamless management of multiple implementation partners across diverse project locations.
