Neo-demographics: Opening Developing World Markets by Using Personal Data and Collaboration

The NEO-DEM project will use non-standard data and novel methods to impact business efficiency, encourage community collaboration and provide scholarly insights into consumer behaviour. UK businesses can struggle in the developing world, despite excellent track records at home. The following reasons explain a good deal of this failure in retail, service and consumer oriented sectors:

* It is not possible to directly transfer domestic business models into emerging economies due to cultural, infrastructural and behavioural differences. Companies need to generate new analytical and strategic models that identify the differing needs of customers based on an understanding of the novel behavioural and consumption patterns exhibited.

* In the developed world consumer oriented businesses are increasingly data-driven. They rely on cross-referenced geo-demographic, socio-graphic, and psychographic data as well as transactional data (e.g. Tesco & Boots in the UK); their use is enmeshed within company strategy. In many countries this kind of data are incomplete or non-existent, their absence inhibits growth and means that targeting and resource use is sub-optimal. Replicating the kind of data that is readily available in the UK will often be impossible or expensive and impractical. Even when transactional data is forthcoming (e.g. Tesco Clubcard Malaysia) there is limited scope to cross-reference them with reliable geo-demographic data-sets and models that are taken for granted in the UK (e.g. Experian's Mosaic).

Despite lagging behind in infrastructural developments, developing countries have experienced digital revolutions; providing a largely untapped opportunity to generate business intelligence. In 2010 of the 5 billion mobile phones in the world 80% were in developing countries and this proportion is continues to grow. African countries have embraced new financial technologies such as mobile payment: over 17m Kenyans use mobile money; around 25% of the country's GNP flows in this way. Crowd sourcing systems such as Ushahidi lead the way in the aggregation of social factors. The project will create a decision support and market segmentation platform generated via personal data, collaborative aggregation and crowd-sourced feedback, that will allow the generation new models of consumer behaviour to support innovation.

Our work will hinge on three case studies in exemplar developing economies (Tanzania, Malaysia and China) where we will develop example behavioural segmentations via novel computational and clustering methods and in partnership with a range of data providers and internationally significant companies including: Alliance Boots, Dairy Farm International, Bakhresa Group, Boots, E-fulusi, Tesco, Marks & Spencer and Experian.

Academic research into consumer behaviour patterns will be significantly advanced by the techniques developed, their application in this field is novel. There is scope to exploit advanced forms of computation and clustering that more readily account for market complexities. There is a very high chance that the project will provide insights into consumer behaviour that have hitherto remained obscure. So the contribution to research in this area could be both methodological and empirical and contextual (robust insights into developing world consumers are more rare). This expeditionary collaboration is likely to open the door to and on-going conversation between the fields of business/consumer analytics and computational analysis.

[Industry Impact] The NEO-DEM project brings together academic researchers, UK industry and data providers in emerging economies to produce a platform that generates behavioural segmentations and archetypes, to help UK firms to identify unmet needs in developing nations. Because behavioural segmentations do not currently exist in much of the developing world (Experian, for example, are unable to offer geodemographic "mosaics" for many emerging economies because the infrastructure isn't sufficient there to use traditional methods) we will help to fill a significant gap in knowledge, supporting both academic research and industry business modelling (especially in the retail sector).To emphasise this, in 2012 the UK retail sector was valued at £300 billion, equivalent to 20% of the national GDP: 1-in-12 companies are in the retail sector, and they employ 1-in-9 of the UK workforce (source: ESRC). Given that recent reports indicate that western multinationals expect 70% of their growth to occur in emerging markets, and these markets continue to grow in size (with populations greater than their developed counterparts), the scope for impact is extremely good.

[Academic Impact] This research provides an unparalleled opportunity for social science, business and data science/machine learning researchers to work on huge datasets in parallel to create a joined-up picture of behaviour in the developing world. Getting good data is hard in academic research - it is normally held protectively in silos by those that generate it. To be able to correlate and mash-up a plethora of huge data streams in tandem is an immense opportunity for researchers. For information science the scope to combine datasets (such as pharmacy data with mobile phone use; or nutritional value of purchases with health metrics) to generate new insights is particularly exciting. From a computer science perspective we aim to extend current state-of-the art in clustering and time series analysis; and with respect to consumer business research the ability to use objective (quantitative) data to provide robust insights or establish 'rules' of behaviour is significant. Impact will be realised through publications, capacity building workshops and seminars.

[Knowledge Transfer] The timely nature of this work, and the potential to understand the rapid social changes occurring in emerging markets, has led to a substantial number of highly significant project partners joining us, sharing both invaluable datasets as well as expertise. We will work closely with our partners' "customer insight" teams, creating much opportunity for ongoing knowledge transfer through the project. Importantly, while project partners will benefit from the models and toolkit derived in the most direct way possible, they will still be in the vanguard - we will disseminate all of our findings (methods and behavioural groups alike) in an open and transparent fashion. This will be through practitioner conferences and forums (e.g. NEMODE and The Demographic Users Group), workshops, as well as through organizations such as the OKFN and UN Global Pulse. All source code produced will be released on open source licences and all research though open publication models.

[Policy Impact] Our societal/policy impacts lie in a greater understanding of what can be elicited from behavioural data streams, contributing to current policies on data privacy as well as helping to understand the potential that behavioural segmentation has for (social) policy making. The project aims to impact on future digital economy research, informing initiatives that integrate business and social science disciplines viacomputational approaches, and to continue its impact beyond the end of the project (with the market insights being dynamic, based on the ever-changing data held by our partner organizations rather than requiring continued input from the research team).