Many sub-Saharan African cities have limited access to data about people’s movements and modes of transport (mobility data) because they lack adequate funding and capacity to conduct large-scale household travel surveys.
Without this data, it is difficult to plan, prioritise and optimise transport improvement projects and to justify their funding requirements. The High Volume Transport Applied Research (HVT) Africa Urban Mobility Observatory (AUMO) project set out to explore alternative data collection techniques, to give a better insight into mobility patterns in target cities.
The team wanted to test the use of existing technologies – particularly mobile phones and smartphones – to collect mobility data at scale, and thereby develop a more cost-effective wayto get this data than through conventional means such as household questionnaires. The target cities were Blantyre (Malawi), Gaborone (Botswana), Kinshasa (Democratic Republic of Congo), Lagos (Nigeria), and Maseru (Lesotho).
Technology-driven data collection techniques, such as those reliant on smartphones, are useful to researchers in part because they are agile and cost less than door-to-door surveys, but they also come with challenges in environments in which there is limited access to technology. It is therefore, important to identify demographic profiles that might be excluded as a result of relying on smartphones (like those in lower income brackets). The AUMO team assessed a range of technologies and their levels of usage among populations in the target cities to help capture data across a broad spectrum of users.
To take advantage of the power of smartphone technology, a tool called User Movement Analytics (UMA) was developed, which can collect smartphone data and detect origins, destinations and routes travelled, as well as travel mode and time. However, it was determined that to be more inclusive and provide more accurate results, tools were needed that would be compatible with older models of low-end mobile phones still in use in the target cities (including 2G GSM devices developed in the 1990s).
USSD (Unstructured Supplementary Service Data) was identified as a mobile phone network communication protocol which could help overcome the lack of internet connectivity on low-end mobile devices, while still supporting interactive multiple-choice surveys. Although there was little precedent for USSD-based surveys, it was understood that they would be technically possible.
A key challenge with USSD is that subscribers must initiate the survey session. To address this, a marketing campaign was developed using text messages sent to subscribers on a location basis, to inform potential participants about the survey and provide the necessary instructions to take it.
Other limitations associated with USSD sessions are that the survey must be completed within three minutes, with a limit of 160 characters per screen. As a result, the survey needed to be abridged, so some of the questions , such as those relating to passenger comfort and experiences of criminal activity while travelilng, were excluded.
During the fourth quarter of 2021, a pilot was held in Blantyre to test the process and technology. Respondents were recruited with the support of a local mobile network operator, who sent out targeted marketing messages to subscribers based on their location. As an incentive, a small airtime voucher was offered in exchange for completing the survey.
On the first day of the campaign, a total of 3,480 promotional texts were sent to subscribers in four of the city’s 23 cellular network defined zones. Unexpected technical issues resulted in the majority of the recipients being unable to connect to the USSD server, so the campaign was halted. 147 individuals were able to connect successfully, 37 of whom completed the entire survey.
Through the pilot, it was established that USSD surveys can indeed be significantly more cost-effective than conventional field surveys, which require greater investment and resourcing manpower. A large proportion of the costs involved in USSD surveys are setup and initiation fees, and therefore cost per user reduces substantially as the scale of the survey increases.
In Blantyre, a 2,000 respondent USSD survey would cost approximately £3,200, while an equivalent conventional field survey would cost approximately £7,000. For 5,000 and 10,000 respondent USSD surveys in Blantyre, the cost would be approximately £4,100 and £5,700 respectively, while equivalent conventional field surveys would cost approximately £14,400 and £27,100 respectively.
Next steps
The technical challenges resulted in a substantial reduction in the number of subscribers able to connect to the USSD server (it is estimated that less than 10% were actually able to connect). Despite this, it is evident that USSD surveys have the potential to cost-effectively collect mobility data at scale. A solution to resolve the technical challenges has since been identified and a further campaign round is currently underway. The team is confident that USSD survey campaigns will yield deeper insights into a broader range of users’ travel habits, especially in cities where smartphone penetration is limited.
