ArticlePDF Available

Informal businesses and digital technology in Sub-Saharan Africa Uses and Value - A newsletter about research in economic and social sciences - Orange Labs


Abstract and Figures

Orange Labs
Editorial Kevin Mellet
The vast majority of businesses in Sub-Saharan Africa belong to the “informal” sector. The dening
features of these diverse enterprises – which tend to be small and undercapitalised, with no standard
accounting system, and escape tax to some extent – have been studied and are now well-known. But
we still know little about how they have used the new technologies which, in less than 10 years, have
become a huge part of their daily lives. How is mobile technology reshaping the activities and interactions
of African businesses? Looking to the future, how could new usage practices and applications support
these rms’ development? First of all, they could prove a crucial factor in Africa’s socioeconomic
development, modernising and improving the performance of informal businesses. They could also be
key for a carrier like Orange, which has a substantial presence in Africa and is investing not only in
building network infrastructure, but also in the development of service oers and innovations tailored to
the informal sector.
The six articles in this issue shed light on these matters. The rst two present the results of a major
quantitative survey on mobile use in the informal sector in Dakar, revealing considerable contrasts in
mobile adoption patterns. The third article examines the reality of the informal sector through a case study
of Ivory Coast’s scrapyards that exchange spare car parts. The fourth article reports on the e-commerce
application "Mon Business avec Orange", tailored to sole traders in the informal sector and trialled in
Ivory Coast in 2017. The fth, meanwhile, reports on a datamining initiative designed to identify informal
businesses within Orange’s consumer customer database in several African countries. The
sixth and nal article looks at infrastructure challenges, and specically the opportunities and
possible involvement of telecom carriers in the electrication of Sub-Saharan Africa.
Informal businesses and digital
technology in Sub-Saharan Africa
Mobile phones and the informal sector in Dakar:
contrasting professional practices
> Jean-Philippe Berrou,Thomas Eekhout
What kinds of digital usage practices do we find
in supplier relations in the informal sector?
> Blaise Boton
Use of digital technologies in informal business
activities in Ivory Coast
> Philippe N’Guessan N’da K., Jean-Marc Josset, Alain Rallet
"Mon business avec Orange" and relations between
formal and informal businesses in Ivory Coast
> Servane Crave, Bruno Conquet
Identifying small informal businesses with machine
learning: a challenge for Orange MEA
> Romain Trinquart, Ismaïl Rebaï
Access to electricity: challenges and opportunities
for telecom carriers
> Erwan Le Quentrec, Georges Vivien Houngbonon
uses and value
A newsletter about research in economic and social sciences
© Anton Ivanov / Shutterstock
January 2019uses and value 60
Since the early 2000s, Africa has been
engaged in an unprecedented digital
transformation which has profoundly
changed the business environment and
the daily lives of local communities.
The rapid spread of mobile phones and
internet use has been a real technological
leap forward for the continent. These
technological shifts are taking place
in a unique economic context in which
the informal sector dominates the
productive structures and the labour
market. Yet the use of digital technologies
within that context has not yet been
sufficiently studied. Questions remain
unanswered about the specific uses of
mobile technology by entrepreneurs in
the informal sector: how do they use
mobile phones in their business? What
are the impacts on the productivity and
performance of their production units?
Answering these questions is the focus of
an ambitious research project launched
in 2016 by the research laboratory “Les
Afriques dans le Monde” (LAM - CNRS -
Sciences Po Bordeaux), with the support
of Orange Labs. The first results of this
research are reported here1.
Jean-Philippe Berrou
(LAM CNRS, Sciences Po Bordeaux)
Thomas Eekhout
(GREThA CNRS, Université de Bordeaux)
With the support of Kevin Mellet
(Orange Labs).
Mobile phones
and the informal
sector in Dakar:
The African digital transformation raises hopes as it
could have signicant benets on local economies in
terms of growth, innovation and structural changes. The
number of active SIM cards in Africa rose from 12.4%
of the population in 2000 to 74.6% in 2016, while the
number of touchscreen-phone users increased from
1.8% in 2010 to 22.9% in 20162. This technological
transformation is also, to a larger extent, taking place
on the fringes of the “informal sector” in which operates
a considerable proportion of the population. According
to the International Labour Organisation (ILO), 70 to
80% of non-agricultural jobs are in the informal sector,
with a particular prevalence of self-employed workers
and micro-enterprises. An extremely diverse range
of informal businesses operating, in part, outside the
scope of public regulations (crafts, repairs, catering,
small shops, etc.) has sprung up in the recent decades,
particularly in urban areas. Today, according to the
International Monetary Fund (IMF), such businesses
contribute from 25 to 65% of Sub-Saharan economies’
GDP (Figure 1).
Senegal provides an eloquent illustration of how
intertwined the informal economy and digital technology
are: the number of active mobile phone accounts in
the country is equivalent to its entire population, while
the informal sector accounts for 97% of all production
units (businesses)3. For this reason that we chose to
carry out in 2017 an unprecedented quantitative survey
on a representative sample of 500 businesses from
the informal sector in the Dakar region4. Through this
survey, we were able to put together a detailed picture
of the revenue, workforce and capital of businesses
that do not engage in formal accounting. Respondents
were also surveyed at length about their telephone and
internet devices and how they use them. In addition to
the questionnaire administered to all 500 businesses,
we held around fty qualitative individual and group
interviews (focus groups) with informal entrepreneurs.
Our initial analysis identied the dierent types of
businesses in Dakar’s informal sector and the dierent
mobile practices of the sector’s entrepreneurs.
Focusing on the characteristics of entrepreneurs and
their business establishments, our rst statistical
analysis showed the highly varied nature of the
informal sector in Dakar. More specically, we identify
four types of informal production units.
The rst distinction we can draw is between “small-
scale, subsistence-level informal businesses” (29%
of the sample) and the “top performers” of the
Orange Labs
Mobile phones and the informal sector in Dakar: contrasting professional practices
informal sector (22%), two segments with starkly
dierent characteristics that have already clearly been
identied in the literature of this topic. The rst group
mainly comprises self-employed workers and very
small, highly insecure, low-performance businesses,
operating primarily in the sales and food processing
sectors. The second group is dominated by relatively
large businesses (80% of them have more than ve
employees) with a high level of capital. They are well
established and often involved in production activities,
with relatively elaborated accounting systems and
signicantly higher economic performance levels than
the other segments. The heads of these businesses
are not necessarily highly educated, but did attain the
highest scores on the cognitive and entrepreneurial
behaviour tests conducted during the survey.
Between these two extremes, our statistical analysis
reveals an intermediate segment of gazelles, with
comparable capital levels to the top performers,
but much lower performance levels whether we
measure them in terms of revenue or protability. In
this intermediate segment, we distinguish two sub-
groups. The rst, the “inexperienced gazelles” (21%),
is made up of entrepreneurs who are younger and
less experienced, but highly educated. Their business
activities are more recent and often in the trade sector.
The second sub-group, the “mature gazelles” (28%),
is composed of older entrepreneurs with more stable
and more long-standing business activities. Though
less educated, they perform well on the cognitive tests
and demonstrate markedly entrepreneurial behaviour.
They tend to be engaged in service activities.
We performed a separate statistical analysis on
professional use of mobile phones among our sample
of 500 businesses in Dakar’s informal sector.
Several aspects of mobile phone usage were
the devices used (simple GSM phone or
touchscreen phone/smartphone);
the main purposes for which mobile phones are
used in business5: i) bilateral coordination (one-
to-one with suppliers, customers and partners) or
multilateral coordination (one-to many on the internet,
searching for and sharing information, promotion and
sales); ii) mobile money services (payment – sending
and receiving money – and savings); and iii) internal
management (accounting, managing accounts and
stock, personnel management);
diversity of use, i.e. the number of applications
and interfaces an entrepreneur uses (voice, instant
messaging, social media, video calls, etc.);
intensity of use, i.e. how frequently the identied
functions are used.
Figure 1: estimates of the informal economy (% of GDP) in Sub-Saharan Africa (Average gures 2010–14)
Between 20 and 30%
Between 30 and 40%
Between 40 and 50%
Over 50%
Figure 2: proles of mobile users in Dakar’s informal sector
Multilateral coordination practices
(one-to-many via the Internet)
Bilateral coordination practices (one-to-one)
Intensity and
diversity of use
Through the statistical analysis of the variables
associated with each of these dimensions, we identied
four distinct user proles, as shown in Figure 2 below.
On the horizontal axis of the graph, a clear distinction
can be seen between “entrepreneurs with simple
mobile practices” and the “digital entrepreneurs of the
informal sector”. The entrepreneurs in the rst group
make very little use of their mobiles for professional
purposes, and very few of them (less than a quarter)
have a touchscreen phone. Conversely, the “digital
January 2019uses and value 60
Mobile phones and the informal sector in Dakar: contrasting professional practices
entrepreneurs” are extremely advanced and intensive
users of mobile phones in their business activities.
They use mobiles for nearly all possible purposes, and
draw on all the interfaces available. To this end, they all
have a touchscreen phone or smartphone. Nearly two-
thirds go as far as promoting their business and selling
their goods and services online.
Between these two groups in terms of intensity and
diversity of use are the “networking entrepreneurs” and
the “connected entrepreneurs”, who stand opposed
on the graph’s vertical axis. The former are particularly
dynamic when it comes to bilateral coordination
with their customers, suppliers and partners. In
other words, they make daily use of their phone (a
touchscreen model, in 60% of cases) to communicate
one-to-one, orally or writing, with their business
contacts. Conversely, they use the Internet very little in
their business activities. The connected entrepreneurs
are much more widely equipped with touchscreen
phones (90% have one). They stand out primarily for
their internet use, with 90% of them using the web to
nd information, and 40% to sell their products and
services and promote their business. Facebook and
WhatsApp are some of the tools they use most.
These rst results provide some answers concerning
the relationship between mobile use and the
performance of small and micro informal businesses in
an African context. If we cross-reference user proles
with the dierent segments of the informal economy
(Table 15), we can see a stark polarisation between
entrepreneurs with simple practices who are strongly
represented in the subsistence-level activities, and
digital entrepreneurs, most of whom are top performers.
This conrms that not everyone is beneting from the
potential of mobile phones in the same way. There are
barriers that prevent some from using the dierent
functions mobile technology oers, be they nancial
(not being able to aord a smartphone) or cognitive
(not having the knowledge and skills required to take
full advantage of mobile functions).
1- For further details about the results presented (particularly in terms of
data collection and statistical analysis methodologies), see the report here:
2- Source: ITU (2017), Measuring the information society. Volume 2, ICT
country proles. Two notes of caution to bear in mind when interpreting
these data: rst, the number of touchscreen phones is measured by the
number of mobile broadband subscriptions, but it is common for Africans
to have active SIM cards with competing operators. We estimate that if we
would to take into account the number of mobile carriers in each country
to offset the effects of multi-SIM users, the average mobile penetration rate
would be less than half as high. Secondly, it is common for a single mobile
to be shared by several people, which could mean actual mobile use has
been underestimated. The inadequate means we have of measuring these
practices make it difcult to come up with accurate estimates of mobile
penetration in the region.
3- ANSD, Senegalese National Agency of Statistics and Demography),
Summary of results of the General Census of Businesses. ANSD, Dakar,
Senegal. 2017
4- Informal production units are considered to be those that are not
registered and/or which do not keep formal written accounting records
in accordance with the standards of the West African accounting system
5- For a detailed presentation of the subject,
see our analysis of the current situation
6- Each cell of this table presents the results of
a Khi2 test in a simplied and schematic form.
A + sign means a cell is signicantly statistically
overrepresented (threshold of 5%).
Mobile use among gazelles is more varied.
Inexperienced gazelles are as likely to be connected
entrepreneurs as digital entrepreneurs. This can be
explained, in particular, by the characteristics of
these gazelles, who tend to be younger than other
entrepreneurs, and quite highly educated. It is not
surprising, either, that networking entrepreneurs are
prominent among experienced gazelles, given their
better integration in the market. Nevertheless, these
gazelles often display economic performance levels
well below those of the top performers. Proof, if any
were needed, that mobile technology alone does not
provide miracle solutions to improve the economic
performance of informal businesses.
It is still uncertain if the use of mobile technology
can help “gazelles” to overcome obstacles to their
development with regard to both external constraints
these businesses must face and the (un)suitability of
the mobile services oered in the specic context of
their business activities. Additional work currently in
progress is seeking to answer those questions. The
next wave of surveys will take place in Dakar in 2019.
Its aim will be to see how these gazelles have evolved
over the past two years, in terms of both economic
performance and digital practices. Particular attention will
also be paid to the relationship between professional and
social/family mobile use.
Simple users
level informal
Top performers
Table 1: users proles and segments of the informal economy
Orange Labs
One of the factors affecting companies’
performance is the nature of their
relations with suppliers. In the informal
sector, though we know the adoption of
digital technology has a positive effect on
businesses’ performance levels, we know
little about the relations these businesses
have with their suppliers or the way these
relations can shape their digital usage
practices. We examined these links
through case studies of businesses in
Dakar’s informal sector.
Blaise Boton
(Orange Labs)
What kinds
of digital usage
do we find
in supplier
relations in
the informal
According to the 15th International Conference
of Labour Statisticians (ICLS), the informal sector
is dened as “consisting of units engaged in the
production of goods or services with the primary
objective of generating employment and incomes to
the persons concerned. These units typically operate
at a low level of organisation, with little or no division
between labour and capital as factors of production
and on a small scale. Labour relations - where they
exist - are based mostly on casual employment,
kinship or personal and social relations rather than
contractual arrangements with formal guarantees.”
The Conference also excluded agricultural, animal-
rearing, forestry and shing activities from the informal
sector. The predominance of the informal sector in
Sub-Saharan Africa makes it a central component of
policies to ght poverty and the focus of much research.
The aim is often to better understand how informal
businesses work and what challenges they face, in
order to improve their performance and help them
formalise their activities. Despite the eorts already
made, there remains much to be done to understand
the ecosystem in which such businesses operate.
Today, this need for understanding is more relevant
than ever [1], due rstly to the changes brought about
in the daily lives of local people and the entrepreneurial
environment by the use of digital technologies [3] and
secondly to the fact that these usage practices bring
benets for development [2].
Our study was conducted with this in mind, focusing
on the links between digital usage practices and
supplier relations in the informal sector. It concentrated
on Senegal, where only 3.8% of jobs are formal [1].
The data used were taken from a quantitative survey
carried out in Dakar in the rst half of 2017 by Orange
Labs and the research laboratory “Les Afriques dans
le Monde”.
The survey concerned a representative sample of 500
informal businesses. They were identied using criteria
set by the Economic and Statistical Observatory for
Sub-Saharan Africa (AFRISTAT) and with reference to
the framework provided by the International Labour
Organisation (ILO) for measuring informal activity.
More precisely, two criteria were used:
not having a national identication number for
businesses and associations (NINEA);
not keeping formal accounts as per the standards
of the West African Accounting System (SYSCOA).
Apart from these criteria, and for practical reasons,
the informal businesses chosen all had xed business
premises and were listed by Senegal’s National Agency
for Statistics and Demography (ANSD). Of the 500
January 2019uses and value 60
What kinds of digital usage practices do we nd in supplier relations in the informal sector?
businesses in the surveyed sample, 40% operate in
the services sector, 33% in production and extraction
activities, and 27% in commerce. The owners of 27%
of these businesses are under 30, while 40% fall into
the 30-40 age bracket and 33% are over 40. 43% of
them are women.
Figure 1: geographical distribution of informal production units
in the Dakar region (Source: Berrou et al., 2017)
Based on these characteristics and the economic
performance of the businesses, this survey identies
four distinct segments within the informal sector:
subsistence-level informal businesses”, “inexperienced
gazelles”, “mature gazelles” and “top performers”. Of
these four segments, the subsistence-level informal
economy explains most of the disparities observed.
This segment comprises small businesses (75% of
them sole traders) that have barely any employees and
operate in highly precarious circumstances in terms
of both basic infrastructure (electricity, water, public
lighting) and the nature of their occupancy (tenants
or owners). Very few of these businesses have bank
accounts, and their economic performance falls short
of that of all the other segments. Subsistence-level
informal activities are particularly prevalent among
rural migrants.
Basing our approach on the literature on businesses’
relations with their suppliers [4; 5; 6], we classied the
businesses in our sample again, this time according
to the form of their supplier relations. We used four
criteria: (i) number and characteristics of suppliers;
(ii) frequency of exchanges and communications; (iii)
nancial operations; and (iv) level of trust between the
The results show that, just like for large formal
businesses [2], supplier relations in the informal sector
are not uniform. They evolve from a relationship
without trust and with little contact into one in which
entrepreneurs and suppliers maintain informal links
and are relatively trusting of each other. More precisely,
we identied four types of supplier relationships, the
characteristics of which are summarised in Table 1:
Table 1: types of supplier relationships
in Dakar’s informal sector
Independently, all segments of the informal economy
are engaged in simple transactional relationships.
The subsistence-level informal segment is statistically
overrepresented in the simple partnership category,
and has a considerable presence in the advanced
transactional and advanced partnership categories.
This can be explained by the characteristics of
entrepreneurs in the subsistence-level informal
segment. They are mostly rural migrants in insecure
circumstances, seeking new opportunities to attain a
more stable status in the city. As a result, they strive
to maintain a relationship of trust with their suppliers,
which, because these businesses have only one
activity, are the only suppliers they have. Because of
the small scale of their activity and their small storage
capacity, maintaining informal links with suppliers is,
to some extent, a way for them to procure products
without too many constraints or, if necessary, on
credit. As they move from transactional relationships
to partnerships, some of them manage to establish
solid, advanced partnerships with their suppliers.
The mobile phone is the most widespread device in
Dakar’s informal sector. Nearly all businesses own one
(cf. Figure 2). Access to touchscreen mobiles sets
connected businesses apart from unconnected ones
(see Table 2).
Which businesses in our sample have Internet access?
65% of those in a simple transactional relationship, 63%
of those with an advanced transactional relationship,
43% of those with a simple partnership, and 73% of
• Generally a single supplier,
located a long way away (even
abroad), usually formal
• Longstanding relations
• Occasional interactions and
• No work on credit
• No trust at all.
• The number of suppliers is
• Very close and informal
• Longstanding relations
• Very regular interactions and
Indierent attitude to work
on credit
• Low credit amounts
• Very short payment terms
• High level of trust.
• Generally a single supplier,
located a long way away (but
in the same country), usually
• Regular interactions and
• No work on credit
• Low level of trust.
• Several suppliers
• The type and location
of suppliers are of little
• Relations dating back more
than one year
• Regular interactions and
• Work on credit accepted
• High credit amounts
• Long credit payment terms
• Very high level of trust.
1- For the purposes of this study, a touchscreen phone is considered
equivalent to a smartphone. It enables users to connect to the Internet.
Departments limits
Number of IPU
Orange Labs
What kinds of digital usage practices do we nd in supplier relations in the informal sector?
those in an advanced partnership. The low connection
rates in the simple partnership category (compared
with the sample as a whole) can be explained by the
nature of the businesses in this kind of relationship.
As described above, they are businesses operating in
the subsistence-level informal segment. Their low level
of education may be a limiting factor for some uses
of the Internet (stock management applications, for
instance) which require a high level of education. Table
3 summarises all the uses made of digital technology
in supplier relations.
If we cross-reference each usage practice with the
dierent types of supplier relationship through a
binomial regression, we can see (cf. Table 4) that the
probability of using digital services in an advanced
partnership is generally high compared with the
probability of using them in a simple transactional
relationship (the reference relationship).
This result can be seen in the use of mobile voice calls,
video calls, SMS and instant messaging, credit cards
and bank transfers, weather apps and photo apps, and
is consistent with high levels of Internet access in the
advanced partnership conguration.
As regards the advanced transactional relationship
and the simple partnership, there is no signicant
dierence compared with the simple transactional
relationship, due to the prevalence of subsistence-
level informal activities in both segments. The usage
practices mentioned above are also inuenced by the
business’s revenues and the entrepreneur’s level of
education: two variables on which the subsistence-level
informal segment scores the lowest. Consequently,
we can suppose that, even though subsistence-
level informal businesses sometimes manage to
maintain a relationship of trust with their suppliers,
this relationship does not necessarily lead them to
use digital technology, given their limited nancial
resources and the costs using such technology can
represent. We have observed that, while using digital
technology improves the performance of a business
(in terms of revenue), this positive eect does not stem
from the use of such technology in supplier relations.
The type of relationship with suppliers has no impact
on use of xed-line telephony, email, Western Union/
Moneygram/Post Oce transfers, or document
archiving, stock management or weather apps. These
usage practices are not frequent across the whole of
Dakar’s informal sector. The use of calculator and of
geolocation and map apps is negatively impacted by
supplier relationship types.
Financial functions
Advanced telephone
and Internet use
Mobile voice (traditional/IP)
Fixed voice
Video call
Messaging (SMS/IM)
Credit card / Bank transfer
Mobile money
Western Union/Post Office
Document archiving application
Application for managing stock, accounts and transactions
Weather application
Geolocation and maps application
Photo application
Touchscreen phone only
Fixed line only
Total number of connected
284 90%
3 1%
30 9%
317 100%
Table 2: Internet access devices (Source: Author)
Table 3: description of usage practices in supplier relations
Figure 2: possession of digital devices in Dakar’s informal
sector (source: Author)
9% 6%
Mobile phone (Simple GSM or touchscreen)
Touchscreen mobile phone tactile
Computer (desktop or laptop)
Fixed line
January 2019uses and value 60
and Internet
Mobile voice (traditional/IP)
Fixed line
Video call
Messaging (SMS/IM)
Credit card / Bank transfers
Mobile money
Western Union/Post Office
Document archiving application
Application for managing stock,
accounts and transactions
Weather application
Geolocation and maps application
Photo application
What kinds of digital usage practices do we nd in supplier relations in the informal sector?
«West Africa Economic Outlook - Macroeconomic
developments and poverty, inequality and
employment - Labor markets and jobs»
[2] WORLD BANK, (2016),
«World Development Report: Digital Dividends»
T (2017), “Les TIC: Une réponse au dé du
développement des micro et petites entreprises
informelles en Afrique sub-sahariennes?”
[4] LACOSTE S. (2011),
“Segmentation fournisseurs et négociation : le
cas du fournisseur « stratégique »”, Management
& Avenir 2011/4 (n° 44), p. 202-218. DOI 10.3917/
[5] LEPRES X. (2003),
« Vers une nouvelle conceptualisation de la relation
d’échange fournisseurs – grands distributeurs »,
XIIe Conférence de l’Association Internationale de
Management Stratégique,
Les Côtes de Carthage
– 3, 4, 5 et 6 juin 2003
[6] MACNEIL, I.R. (1980),
“The New Social Contract: An Inquiry into Modern
Contractual Relations”. Yale University Press, New
Haven, 134-137.
To understand the role played by digital technologies
in boosting the performance of informal businesses in
the Dakar region, we analysed the links between their
digital usage practices and the types of relations these
businesses have with their suppliers. We can see that
when the two parties work together regularly and trust
each other, they step up the use of various services
within their relationship: mobile voice, video calls, SMS
and IM messaging, credit cards and bank transfers,
mobile money and geolocation applications. However,
this link does not translate into an increase in revenue.
This opens up other avenues to explore concerning
the adoption of digital technology by these informal
businesses, particularly the coordination mechanisms
and their translation into improved performance.
Other aspects related to usage practices in customer
relations, horizontal relations and internal relations are
also worth exploring in future research work.
Though we have shown that some digital usage
practices are quite widespread in the sector, it should
be emphasised that some usage practices, such as
email and account management apps, are so rare that
it was not possible to obtain signicant coecients
that could truly show the eects we were looking for. To
take things further, the next studies on the subject will
have to adopt an experimental approach (randomised
controlled trials or dierence in dierences) to measure
the impact of these relationships on the use of digital
technology. In future research, priority could be given to
businesses in the subsistence-level informal segment,
many of which are engaged in advanced partnerships,
but which have the lowest levels of equipment and
Internet access.
Table 4: summary of the eects of supplier relations on digital usage practices
(-) signicant
negative link,
(+) signicant
positive link
Orange Labs
The informal sector is predominant
in the economies of the developing
countries. African countries are no
exception, particularly Ivory Coast, with
the economic consequences of the civil
war. But what is informal economy? It is
entirely informal? How do the formal and
informal aspects and mechanisms of the
informal economy interact?
These are important questions to address
if we are to understand the use of digital
tools in relations among stakeholders in
the informal economy. Will these tools, or
should they, serve to formalise informal
relations with a view to modernising the
economy? Or will they make informal
interactions more efficient and thus act
as vectors of growth for the informal
Jean-Marc Josset
(Orange Labs)
Philippe N’Guessan N’da K.
(RITM, Université Paris-Saclay)
Alain Rallet
(UFR-SED, université de Bouaké)
Use of digital
technologies in
informal business
activities in Ivory
In development economics, the formal and the informal
are generally represented as separate, distinct sectors.
There is no relationship between them other than the
constant supply of labour from the informal sector to
the formal until the informal workforce no longer exists.
To argue this position, it is necessary to assume that
there is no other mobility between the two sectors:
i.e. that no individual works partly in both sectors,
and that no one switches back and forth between the
two. The same applies to business activities: under
the conventional view, they belong either to the formal
sector or to the informal sector.
This separation is highly questionable. An anthropo-
logical approach shows that the informal and the
formal are closely intertwined within given elds of
activity, be they considered formal or informal. This is
the assumption we preferred to adopt and verify here.
To this end, we conducted a qualitative study of the
coordination mechanisms at work in a given industry,
analysing their nature and specifying the instruments
(technological or otherwise) used by stakeholders to
coordinate with one another.
The sector chosen for the study conducted by Orange
and Paris-Sud University was the automotive spare parts
market, a buoyant part of the Ivorian economy. More than
Figure 1: Abobo-Anador scrapyard
January 2019uses and value 60
Use of digital technologies in informal business activities in Ivory Coast
two-thirds of cars in Ivory Coast are second-hand, meaning
frequent repairs are needed.
The parts sold at scrapyards may be new, second-hand or
recycled, with second-hand parts particularly prevalent
and thus more affordable. The most appreciated second-
hand parts are known as “France au revoir” parts, which
come from Europe, mainly from France. The automotive
spare parts workshops in Abidjan tend to cluster in
scrapyards. 6 garage owners and 42 vendors based at Ivory
Coast’s four largest scrapyards (Abobo-Anador, Abobo-
N’dotré, Adjamé Mirador and Yopougon) were interviewed
and observed. The semi-directive interviews, conducted in
the interviewees’ workplaces, focused on 5 topics:
the vendors’ career history and integration in their
their sales activities (suppliers, customers, partners,
relations between the vendor or association of
vendors and government authorities and banks;
the communication services used in the activity
(phone, computer, etc.) ;
the social organisation of vendors to tackle the
difficulties they face.
Due to the nature of the sector (second-hand parts) and
the context, we might have expected exchanges within this
business to be informal. However, many exchanges have
a formal basis: scrapyard vendors pay taxes to the local
authorities, and receipts are used in the payment circuits
(customs, advance payment, partial payment, private or
bank credit). The formalisation of the exchange does not
stem solely from relations with the players in the formal
economy (institutions, importers, shopkeepers, banks),
but also from relations among customers, automotive
workshop owners and vendors at the scrapyards. When
formalisation is accepted, it is always accepted for a
practical reason: the players are much more willing to
consent to formalisation (written records) when it makes
exchanges easier or more secure.
Figure 2 presents a simplified view of the players in the
Generally based in small shops with a clearly defined
speciality: a car brand (Peugeot, Toyota, etc.) or a type of
equipment (lighting, ventilation, etc.). This specialisation
leads to limited competition and considerable cooperation
with other vendors.
Apart from a few private individuals and some professionals
(taxis), they are mainly automotive workshop owners.
Billing methods vary according to constraints and trust
levels: if a workshop owner has not been paid in full in
advance, he will ask the scrap merchant for partial credit,
a request that will generally be granted if he is a regular
customer. Exchanging or replacing parts is also common
practice, albeit not appreciated by scrap merchants. Cash
is the predominant payment method between workshop
owners and scrap merchants. Payments between private
individuals and garage owners are more varied in nature
(Orange Money or an equivalent).
3. SUPPLIERS, with 4 main sectors:
parts manufactured in Asia (often counterfeits)
parts salvaged locally from vehicles that have
broken down or been involved in accidents (3.4);
parts from the European second-hand parts market
(3.2): agents are generally sent to collect enough
of the desired parts to fill a container, which is then
forwarded to Abidjan or other markets, such as
Guinea and Nigeria;
markets which, in turn, will become regional
suppliers (3.3).
Trade unions represent the scrap merchants to other
players in their environment in order to:
find a place for them to do business, which requires
negotiating with the local or government authorities,
with owners of shops and urban developments, and
with local residents;
discuss the terms of tax payment with the local
council (request for a tax stamp proving the payment).
They also make it possible to make collective purchases
from service providers such as telecom carriers (8) (fleet
contracts). Trade unions also have a role in protecting
against theft (security guards) or against bad customer
behaviour. They help maintain solidarity among
members (e.g. organising a collection in the event of a
member falling ill or suffering an accident). Finally, they
play a role in resolving disputes between stakeholders.
The local council issues certificates that enable scrap
merchants to ply their trade. These certificates are
negotiated through trade unions. They are accepted by the
scrap merchants, who see them as a way of proving their
legitimacy to their customers and of preventing illegal
practices (taking apart stolen cars, for example) that are
detrimental to the image of the profession. However, the
authorities are more generally viewed with mistrust, with
considerable suspicions of embezzlement and corruption.
Orange Labs
Use of digital technologies in informal business activities in Ivory Coast
Religion and traditions sometimes play a clear role in scrap
merchants’ activities. Loans are viewed with suspicion and,
when a member of the community has a problem or needs
help, religious personalities (imams) and traditional moral
leaders (sages, griots) are consulted.
Vendors have access to 3 types of finance:
funds possessed by the vendor (1) or supplied by
their family or friends (6) and, more rarely, from bank
loans (7);
pooling of money by several vendors to make a joint
purchase (e.g. a broken-down car, a container);
money from the sales cycle: selling parts makes it
possible to buy more, with profits used to gradually
increase stock.
The formalisation of exchanges between these players is
based around horizontal relationships. The players carry
out their activities in an environment in which trust is
often absent. In particular, it is difficult for them to rely on
legal rules without suspicion (relations with government
authorities, dispute resolution, etc.).
This is why a large part of the necessary regulation of the
sector’s activity is endogenous: the players generate their
own regulating norms. Restricted to the community of
players in the sector, these self-generated rules, arising
from the repetition of transactions, create the necessary
trust between players. Trust is thus generated not only
through social markers (family, ethnic group, etc.), but also,
and above all, through the establishment of pragmatic
shared rules based on experience.
The uses of communication tools by the workshop owners
and vendors interviewed: (i) are not systematic, (ii) require
little prior investment, and (iii) are substitutable (i.e.
easily replaceable with other services). They generally
use the technical solutions that are the simplest, the least
expensive, and the fastest to implement (though not
always for their intended purpose).
Voice exchangers : as most of the interviewees
emphatically confirm, the phone is the key working
tool (“if you haven’t got your phone, you might as well
go home”). Many conversations will be held between
people who know each other, which has led to the
negotiation of ‘fleet’ contracts at specially adapted
Text messages : Several interviewees said they
did not use text messages because they have not
been formally educated, which leads them (and, by
contagion, their interlocutors) to prefer voice-based
and visual methods (photos).
Photos : taking and exchanging photos is the second
most common use of telephones. These photos serve
various functions: (i) to identify cars and parts; (ii) to
prove possession of cars and parts; and (iii) to replace
written messages.
Figure 2: main players in the industry (illustration by the authors)
87 6
5 4
1 2
Influence Certificates + taxes
Procurement CUSTOMERS
January 2019uses and value 60
Use of digital technologies in informal business activities in Ivory Coast
Benjamin, N. et A. A. Mbaye (2012),
“Informal businesses in French-speaking West Africa:
size, productivity and institutions”, World Bank, Africa in
Development collection, Pearson.
Berrou J.P., Combarnous F., Eekhout T., 2017,
«Les TIC : une réponse au dé du développement des
micro et petites entreprises informelles », Blog recherche
Chambre Nationale des Métiers de Ivory Coast (2013),
Étude nationale : « La formation professionnelle et le
secteur informel en Ivory Coast ».
Lewis, W. A., 1954. “Economic Development with
Unlimited Supplies of Labour,” Manchester School, 22,
Union Économique et Monétaire Ouest Africaine
(2003), « Le secteur informel dans les principales
agglomérations de sept Etats membres de l’UEMOA :
Performances, insertion, perspectives ».
Computers, fixed solutions: though some
application needs were mentioned (for order taking,
billing and stock management, for instance) computers
are conspicuous by their absence. For most players, it is
a problem of training or space.
Digital social networks: the notion of community
is mentioned at various points, but not as a higher
structure encompassing all the rest (trust, exchange
rules, financing rules, etc.). Like the other points, it is
only referred to when it makes sense to do so. That
would appear to exclude structured digital social
networks (Facebook, LinkedIn, etc.), which all require
prior investment and entail considerable operating
costs, not least due to the social externalities generated.
Payment solutions: the dematerialisation of
monetary exchanges seems not to bother the inter-
viewees, who often make limited use of such solutions
already. This kind of solution could meet several needs:
joint financing, credits, advance payment, PayPal-style
services, joint payments, etc.
Observing the means of communication used in the spare
parts sector in Abidjan reveals constant interweaving of the
formal and the informal, which should lead us to reconsider
the cardinal distinction economics makes between formal
and informal. This exchange network operates according
to various pragmatic principles that only a usage-based
approach can truly reveal in full. These principles include
endogenous formalisation of exchanges (not driven by an
authority), the establishment of relationships of trust based
on repeated sales, and the use of photographs to fulfil
multiple functions. Lastly, the appetite for new solutions
among players in the sector suggests that innovations
could be adopted rapidly, provided they meet the criteria
highlighted on several occasions: i.e. they must (i) be non-
systematic, (ii) require little prior investment, and (iii) be
Orange Labs
In Ivory Coast, as in many other African
countries, 99% of businesses are very
small enterprises. Often without business
premises or desktop computers for
accessing the Internet, they are very
difficult to identify from a commercial
point of view due to the informal nature
of their activities. They do, however, use
mobile technology for their business
operations and are embracing value-
added digital services to help grow
their business. Recent experiments are
testament to this.inspired by social media,
via an integrated on-mobile customer
journey (including communication,
transaction/negotiation and payment).
Servane Crave
(Orange Labs)
Bruno Conquet
(Orange Labs)
"Mon business
avec Orange"
and relations
between formal
and informal
businesses in
Ivory Coast
This article presents the results of the experiments
conducted with "Mon Business avec Orange" (MBAO)
in Ivory Coast in 2017. The MBAO service is aimed
at our B2B customers and is an example of the kind
of application that could eventually be developed
by partners or large companies for their distribution
networks. The solution oers a unied customer
experience spanning communication, management
and transactions. It enables the creation of structured
B2B value chains via an inter-business transaction
services platform and oers an application ecosystem
inspired by social media, via an integrated on-
mobile customer journey (including communication,
transaction/negotiation and payment).
Like their Western counterparts, professionals in the
MEA zone are seeking to improve their operational
eciency, sales performance and customer
experience, but they are doing so in a highly specic
context. Often informal and engaged in multiple
activities, very small businesses rely primarily on
smartphones for their B2B and B2B2C operations,
and the use of such tools is growing exponentially.
Equipped with prepaid SIM cards, these businesses
switch from one carrier to another to get the best prices
and the highest-quality coverage. They already use
WhatsApp, Facebook and Instagram to promote their
products online. In hyper-competitive markets where
Over-The-Top (OTT) services abound, there is a huge
risk of telecom carriers being cut out of the equation.
Their goal - for telecom carriers - must be to attract
these entrepreneurs and keep their custom through
paid bundles that inject new life into their business
through a combination of connectivity and business
process digitalisation services (catalogues, orders,
payments). In this respect, mobile money services –
such as Orange Money, which has some 20 million
users – have had a huge impact in MEA due to the
frequent lack of available cash, the low proportion of
people with bank accounts, and the security problems
involved in transporting funds.
Our application enables vendors to create an online
store and get their customers talking through
Instagram-style posts. It also oers customers the
possibility of negotiating their online orders directly
through several communication channels. Payment is
secured through Orange Money, which is tailored to
the businesses’ needs.
Developed as part of the “Digital Emerging Countries”
research eld and presented at the Orange Labs
research event in December 2015, MBAO was greeted
with considerable enthusiasm due to its simplicity
and the needs it meets. To take things further and
make the tool operational, we contacted Orange Ivory
1- Sofrecom-Orange Study 2015 “Marché des services numériques en
January 2019uses and value 60
Figure 1: an integrated customer experience
"Mon Business avec Orange" and relations between formal and informal businesses in Ivory Coast
Flexible Data Sync (WebRTC technology) developed by
Orange Labs.
Based on a commercially oriented social media model,
a wholesaler would thus be able to communicate with
his customers about new products and promotions by
synchronising stores and sending notifications.
Customers would be able to contact their wholesaler
easily to request information, via a built-in chat function
or an audio/video call, for instance.
Use new B2B Orange Money
Based on the prepaid model, the
aim is to enable businesses to set up
payments of several types (payment
on due date; payment on delivery,
handled on the delivery person’s
terminal, etc.). Such solutions meet
a strong demand in emerging countries to avoid
transporting cash for security reasons.
A partnership was signed with the Abidjan-based
association My Way Network, which aims to develop the
work of street vendors in various neighbourhoods of
the city. The head of the association offered his retailers
the chance to participate on a voluntary basis. 10 testers
(jewellery vendors) aged between 22 and 55 agreed to take
part: all sell jewellery for the association as a secondary
activity, have smartphones, are Orange customers and
are part of Ivory Coast’s middle class. These testers were
given gift cards worth €30, usable in supermarkets. The
testers could not all be trained to use the application due
to difficulties in travelling from one neighbourhood of
Abidjan to another, and two people were ultimately unable
to take part in the test, having had problems installing the
application and failed to reply to the various reminders
Coast to set up a pilot scheme. Firstly, thanks to Azao
Consulting, a Paris rm specialising in developing
Social Business initiatives in Africa, we got in touch
with the association “My Way Network”. In July
2016, focus groups were organised in Abidjan, where
the app was presented and users’ rst impressions
noted. This user feedback enabled improvements to
the app’s customer journey and the simplication of
certain interactions.
The core context of the pilot
scheme launched in Ivory Coast
was the interaction between formal
and informal businesses, illustrated
by a wholesale jewellery vendor and
retailers charged with distributing
the collections. The retailers work
in the informal sector, while the
wholesaler, by the very nature of his activities, which
we could classify as Social Business (promoting the
work of street vendors, for instance) operates in the
formal economy.
By using MBAO, the wholesaler would be able to
accomplish the following:
Manage his business
From his smartphone, he can create and manage his
store, add and modify products, manage stock, and
track his business activities.
The application is highly visual and particularly suited
to smartphone use, as Figure 1 shows.
Manage his customer relations and
MBAO should enable wholesalers to manage customer
relations through the integration of a real-time data
communication and synchronisation environment:
Our applications must
be designed to suit
business people
with multiple activities
Orange Labs
"Mon Business avec Orange" and relations between formal and informal businesses in Ivory Coast
we sent them. The head of the association, who works as
a jewellery wholesaler, also took part by creating the first
online store, taking the total number of testers to nine.
Two systems were implemented to analyse the results
of the pilot scheme. First, Google Analytics was used to
observe all the connections to the service, their duration,
and customer journeys (from browsing the store to paying
via Orange Money). Second, an interview guide was
written to help us collect feedback from testers through
telephone interviews, which were
particularly difficult to schedule, as
the testers often had little availability
and the connection was often of poor
During the experiment held in Abidjan in the second
quarter of 2017, we observed regular use of the
application over a three-month period, confirmed by 179
sessions lasting an average of 6.5 minutes. These sessions
generated 125 exchanged messages, 39 shopping baskets,
and 9 orders placed with the wholesaler, four of them paid
for with Orange Money. Once the testers had explored
the application and connected a considerable number
of times, an obstacle quickly appeared: the products
offered did not meet vendors’ expectations, which greatly
diminished the attractiveness of the service. The first store
entirely devoted to jewellery did not live up to vendors’
hopes (insufficient quality, models that did not appeal to
their customers).
This obstacle was quickly identified by the wholesaler, and
he promptly created two cosmetics-focused stores, which
met with greater success. The simplicity of creating online
stores thus offered considerable commercial adaptability.
For the retailers, the service saved time (“I don’t have to
travel around any more. I can see the prices directly, not like
on Facebook”), offered greater efficiency (“I can compare
the prices and negotiate better”) and enhanced traceability
(“it helps me track my business better”).
The service improves the image the retailers have of the
wholesaler, and the retailers spontaneously envisage
making symmetrical use of the application with their
customers, as a way of saving time (“I show the application
to my customers and they can tell me what they like”) and
gaining in professionalism.
For the wholesaler, the service makes two aspects of his
work easier: stock management (“It does me a huge
favour. Often, customers come in for a piece I haven’t got
any more... When I realise, I take the photo off Facebook”)
and traceability (“I know where I’m up to and where I need
to go to get money. The written records are useful after a
negotiation”). In the current version, the wholesaler does
not see the time saving, and a tablet or PC version would
be more appropriate for adding products to the stores.
Analysis of our panel of testers shows that it is extremely
common for people to be engaged in multiple business
activities, be it habitually or on a seasonal basis, and that
this phenomenon should be taken into account in service
design, despite the relative lack of literature on such
working arrangements. All the testers see selling jewellery
as a source of additional revenue on top of one or more
other activities. In order to optimise return on investment,
develop turnover and meet customer needs and desires
as closely as possible, the digitalisation
of processes must be applied to these
activities, too. Great care must be taken
when designing the service to ensure
that users are able to get to grips with
it quickly and intuitively. People stop
using an application even more quickly when they use it
for a secondary activity.
An informal activity is not incompatible with formal
payment methods such as Orange Money. On the contrary,
Orange Money is viewed positively for several reasons: it
saves time and offers traceability and payment security.
The scarcity of cash is another factor that contributes to
the appeal of Mobile Banking.
In conclusion, this experiment shows that digital
technology is used for sales in the Africa and the Middle
East zone. Be it via Facebook (wholesaler pages, segmented
community groups) or WhatsApp (groups, discussions),
the means exist to initiate sales online with retailers or end
The nine users all confirmed an interest in the service.
They see it as a tool for structuring their business and a
way for vendors to develop their turnover. Making the
application more stable to eliminate bugs and enhancing
data connectivity (to display photos where it is currently
impossible due to insufficient coverage, for instance)
are the two main areas for improvement reported by
the testers. A new version has been developed and an
experiment with the “Boutique Paysanne” is under way in
For the wholesaler to understand the added value an
application of this kind can bring, greater commercial
activity and additional features will be required. Payment
via Orange Money does not seem to be an obstacle to
the transaction. Interestingly, there is no reluctance to
use Visio, which is considered a useful tool for making an
initial contact and presenting a product in greater detail.
This enabled the wholesaler to show his vendors jewellery
without the vendors having to travel.
MBAO makes small
more responsive
External communications site for the project:;language=english
January 2019uses and value 60
When taking the point of view of a telecom
operator, who markets communication
services (and more recently money
transfer and payment services), it really
is uneasy to address small informal
businesses– also known as Small offices
Home offices (SoHos). These companies
are not registered, so there is no official
way to reach and target them. They resort
to mass market services which are in line
with their small size and their finances–
while B2B offers typically target large or
medium sized formal companies. Yet their
digital uses, and needs, are specific, and
quite often intensive; hence, dedicated
services and offers could be sent to them.
But how can we identify them?
Romain Trinquard
Ismaïl Rebaï
small informal
businesses with
machine learning:
a challenge for
Orange MEA
This paper describes the steps, and promising results, of a
machine learning approach to improve the identification
of informal businesses among Orange existing
customers. It illustrates how ‘raw’ electronic payment
and telecommunication usage data can be used to score
individuals among our customer base according to their
similarity with known businesses. High scores can then
be used to spot potentially unknown businesses and
hence to drive a dedicated marketing campaign or push a
customized offer.
This work has been performed through a close collaboration
between “Emerg Data” Research project, “SoHo In Retail
Acquisition” Program (SIRA) and Middle East and Africa
(MEA) Orange subsidiaries in Ivory Coast, Mali, Congo
Democratic Republic and other African countries.
Using call records data as well as transactions data from
Orange Money, we show that it is possible to build a model
from the behavior of known professional customers and
that this model can effectively discriminate previously
unknown businesses within our customer base. The
burden of mining both telco and Orange Money data
was kept to a minimum thanks to the use of Orange Lab’s
Machine Learning solution, namely Khiops.
The relative importance of businesses in the economy is
well known: their revenues are highly unequally distribu-
ted. Large companies are only a small minority in MEA
countries (only 0.2% in number) but they stand on average
for almost half of the domestic product (44% of GDP). On
the opposite end of the spectrum, the SoHo segment,
which gathers companies with 1 to 9 employees, contains
94.5% of companies, and stands for 26% of GDP. Moreover,
another measure is crucial to understand the difference
between B2B segments: the ratio between registered
and unregistered businesses. All large companies are
registered, and identified as such by telecom operators,
whereas SoHos are mostly unregistered, and thus invisible.
Hence SoHos are a specific target for marketing: They are
an unaddressed audience, which may need specific offers
and many customers among Orange B2C base are actually
SoHos (but which among them?).
Based on these observations, Orange MEA’s marketing
unit implemented two streams of actions. They designed
new offers for professionals with the aim to upsell
existing professional customers as well as to acquire new
B2B customers. And they developed a Customer Based
Management (CBM) approach to identify B2B prospects
within their B2C customer bases. Note that the second
stream is under the B2B marketing team responsibility, but
requires close coordination with the B2C marketing team
in charge of the first stream in order to push consistent
Orange Labs
Identifying small informal businesses with machine learning: a challenge for Orange MEA
Figure 1: the Sira Program: addressing the SoHo segment in Orange MEA
offers to the clients. The hidden threat here is to create
conflicts between different marketing teams (ie. B2C and
B2B), which could target the same customers.
We did contribute to this second stream by providing the
tool and methodology to actually identify B2B prospects
within our customer base and push scores that could be
used to drive active marketing campaigns. We provide
details on the methodology and results in the following
The ground idea to identify B2B prospects within the
existing customer base is to mine usage data to learn
behavior profiles: customers whose profiles are similar to
those of known professionals are assimilated into potential
This is crucial to the success of the whole program: based on
that, the marketing team has to define an active campaign
to verify and exploit the identified potential businesses.
The customer business management (CBM) approach to
identify B2B prospects consists of several steps, each one
involving different actors:
1) first the Business Intelligence (BI) and Information
Technology (DSI) teams of the subsidiary (country) should
work with marketing team to select available and relevant
data :
a) to select a population of known professional and
standard (non-professional) customers within our
customer base;
b) to gather usage data;
2) the data-science team conducts the analysis to build a
discriminant model of professional customers;
3) the model can then be used to score the whole customer
base, hence identifying those customers whose behavior is
similar to professionals;
4) the data science team transfers an interpretation of the
model to the marketing team, together with the top scored
customers (i.e. customers most likely to be professional);
5) the marketing team can then decide if the model should
be exploited and then designs a campaign, including
a specific script to check if the prospect is actually a
professional and which offer should be pushed.
The success of the process depends on several aspects:
the quality of the input data (relevance of the examples,
availability of detailed usage data, and availability of a
sufficiently large set of identified SoHos), the performance
of the learning algorithm and finally the design of the
campaign script. In order to assess the success of the
whole process, it is important to keep track of different
metrics. Especially, one should measure the rate of good
detections (prospects who are actually professional), the
rate of successful calls (prospects who actually subscribe
to an offer) and finally the increase in sales (in term of cash).
One last thing that we should stress out is the importance
of the initial set-up. The learning algorithm takes as input
a selection of basic customers and known professionals
from the customer base. The output model is expected
to discriminate the customers whose usage is similar to
the professionals provided as examples. These examples
should be picked with great care because the model will
not be able to detect professionals whose behaviors differ
completely from the initial set.
We now discuss the datamining part of identifying B2B
prospects in the customer base. We first present the data
we consider and then the tool we used to swiftly build
models for up to five countries of the MEA zone.
January 2019uses and value 60
Identifying small informal businesses with machine learning: a challenge for Orange MEA
Usage data encompasses both telco data and transactional
data from Orange Money. Telco data, namely Call Detailed
Records (CDR), sums up the activity of the customer in
term of voice calls, SMS, data sessions as well as recharges
for prepaid account. In the same manner, Orange Money
raw data consists in detailed transactions of all types
(Cash-in, cash-out, Peer-to-Peer transactions, Merchant
Payment, etc.) In both cases, marketing teams got access
to aggregated variables describing the activities of each
We also had to provide a solution that could deal as quickly
as possible with various datasets, since each country could
provide its own specific dataset: detailed or monthly-
aggregated data for both telco and banking dataset.
Dealing with detailed data is both an opportunity and a
potential burden. It is an opportunity because instead of
trying to fit hard-coded aggregates with the target, one
can explore plenty of variables to select the most relevant
ones. It is a burden because specifying which aggregates
to explore can be extremely costly in terms of engineering
and analysis time.
We used Khiops, a machine learning solution developed by
Orange Lab. This tool has several strengths which make it a
key factor of success in our work:
first it is parameter-free, delivering first-of-class
results in terms of performance and robustness to
noise in the data;
second, this tool has the unique ability to handle
raw detailed data such as CDRs instead of aggregated
values: the tool can cope with a star-like relational
schema and dynamically generate and evaluate
aggregation formula. This is a by-pass to the otherwise
cumbersome task of specifying manually how to
summarize usage data;
third, it makes possible model interpretation and
variation: one can easily explore the selected variables
as well as their correlation to the target;
last but not least, Khiops handles easily large
datasets, with no required pretreatment for missing or
abnormal values.
For each country, the objective was to deliver two kinds
of results. First, the model itself, with both performance
measurement and a sketch of the main discriminant
variables. Second, after explaining the results to the
datamining and/or marketing team in the country, the
scoring of all customers in the database.
As an example, Figure 3 illustrates the lift curve of the
model learnt from the data from Ivory Coast. This curve
draws the rates of businesses that the model has correctly
ranked along all customers. A marketing campaign will
focus on those customer registered as non-professional
and yet with high scores.
Figure 2: relational diagram of Supervised Learning
Figure 3: performance of the model
Figure 4 illustrates how the model learnt by Khiops can
be interpreted: the most informative variables can be
picked and described in terms of standard customer
behaviors. This interpretation should not be mistaken with
Performan indicators :
accuracy, AuC, success rate by quantiles
On test data, the first 4,5 % of top scores
gather 80 % of the target.
Figure 4 interpreting SoHos’ behavior
Pros do less rells but with a much greater
Pros have incoming calls from a greater number
of dierent contacts.
There is a greater concentration of Pros
within the receiver of incoming P2P transaction.
A small niche of Pros makes intense use of Data.
Optimal: test
Naive Bayes:
MAP Naive
Bayes: test
25 50 75 100
Target modality %
Orange Labs
Identifying small informal businesses with machine learning: a challenge for Orange MEA
a full-fledged segmentation. Yet it provides glimpses of
understanding that can be shared with marketing teams
and thus facilitates the adoption of the proposed top
In 2017, 11 countries were contacted. The methodology
was presented to 9 of these countries, out of which 6
countries expressed strong interest. Among the 6 countries
that expressed interest, Senegal opted for implementing
and testing the methodology on its own. We thus worked
on the data of the 5 remaining countries, listed here in
chronological order: Democratic Republic of Congo, Ivory
Coast, Mali, Jordan and Sierra Leone. All these countries
were able to provide us with a dataset covering a period
of three months of collected data. Figure 5 summarizes the
current status of the program all over the countries.
In the remaining of this section, we will illustrate the value
we did provide to our partners by detailing our results with
Orange Democratic Republic of Congo (ORDC) and Orange
Ivory Coast (OCIT).
ORDC did provide a very rich dataset, with both raw CDRs
data and raw Orange Money transactions. We shared the
model and the top scores with the subsidiary. They then
launched a campaign and contacted 5.000 customers
identified as potential professionals. The campaigned
revealed that 28% of the successful calls were very small
businesses whereas 11% of the successful calls were
middle businesses. 90% of these identified professional
prospects lead to successful sales, thus generating a 9%
increase of the monthly B2B results (while the campaign
only run over 9 days).
OCIT provided a dataset with raw Orange Money
transactions and monthly aggregates for telco usage. We
were able to provide a model with theoretical performance
similar to that of the model computed for ORDC. The model
also yielded insights into average professionals’ profiles:
average monthly top up (pre-paid credit recharge)
frequency for a Pro is less than a Non-Pro. However, Pro
ARPU is higher than Non-Pro’s;
a Pro tends to recharge a higher amount each time;
a Pro is called by a larger number of persons;
a Pro has a higher probability to receive P2P
pros are keen to use certain VAS;
some Pros (niche) have a higher usage of data.
On the research side, two streams of work should be further
investigated to make marketing campaigns even more
efficient. The first one would be to spread the deployment
of scores not only to individuals already in the customer
base but also to those that may appear as contacts either in
the CDRs or in electronic money transactions. The second
one would be a change in the learning set-up: instead of
applying a straight supervised approach, with the implicit
assumption that we are provided with accurate tagging
of pros versus non-pros in the learning sample, we should
try to analyse the distribution of the provided tagging and
somehow “correct” the sample distribution.
Within this program, eLob and MEA marketing units are
pursuing their efforts to spread the approach to other
countries. Negotiations have begun with the Jordan
Technocenter on the opportunity to embed such a
methodology into the Loyalty Management System
product, which is provided by Technocenter in Amman,
adopted throughout the Orange subsidiaries in the Africa.
© Anton Ivanov / Shutterstock
January 2019uses and value 60
Sharing of methodology
and modeling tests with
Sonatel teams.
Implementation by
Modeling and learning on the
whole base. Interpretation and
recommandations results.
Questionnaire under
Potential application in the next
Tests with Amman.
Technocenter on small and partial data
Business expectations under
Modeling and learning on the
whole base. Interpretation and
recommandations results. Campaign
launched on September.
51 % of targeted customers are
registered as businesses
Modeling and learning on the
whole base. Interpretation
and recommandations results.
Questionnaire completed and sent.
40 % of successfull calls are small and
medium enterprises.
Figure 5: overall deployment status
© Nancy Haggarty / Shutterstock
© Sarine Arslanian / Shutterstock
Orange Labs
In Africa, socioeconomic development
has long been hindered by insufficient
access to modern and sustainable energy
sources. However, with the production
costs of decentralised off-grid solutions
falling sharply, and mobile payment now
widespread, the time seems riper than
ever to bring electricity to millions of
households and businesses that do not
currently have it. As traditional players
develop strategies to seize this new
opportunity, what are the challenges for
a telecom carrier, and how could such an
operator position itself in the value chain?
To answer these questions, we have
compiled the most recent statistics on
the state of the sector, conducted an
original economic study spanning around
forty countries in Sub-Saharan Africa,
and examined case studies of effective
business models.
Erwan Le Quentrec
(Orange Labs)
Georges Vivien Houngbonon
(Orange Labs)
Access to
challenges and
for telecom
Despite its enormous potential, Sub-Saharan Africa has
thus far suffered from a lack of access to electricity. In 2016,
600 million Africans had no access to electrical power:
more than half of the world’s non-electrified population.
Those who do have access continue to suffer from limited
generating capacity and untimely power cuts. This results
in low electricity consumption per capita, estimated at
180 kWh, as against 1300 kWh in North Africa. The deficit
can be explained, in part, by low population density (45
inhabitants/km², compared with 365 in South Asia), which
makes extending the grid outside of urban areas more
costly. The preponderance of the informal economy in the
region, and the low living standards that result from it,
slows down growth in electricity consumption. According
to World Bank estimates, the deficit in access to costs the
region between 1 and 2 percentage points in GDP growth.
However, the technological progress made with
decentralised off-grid solutions (mini/micro-grids, solar
kits, etc.), combined with innovations in business models
through mobile payment, opens up new prospects for
electrification in Africa. Driven by product innovation
and large-scale production in China, the cost of off-grid
solutions, which are better suited to low-population-
density areas, has fallen considerably over the past decade.
According to the International Renewable Energy Agency,
the cost of photovoltaic panels fell by 80% between 2010
and 2016, while the cost of wind turbines dropped by 40%
over the same period. Forecasts for 2019 predict a unit cost
of 4 cents per kWh for solar energy, which would make it as
competitive as hydroelectric power. The spread of mobile
payment has also brought a rise in “pay-as-you-go” models,
which are better suited to users with a low, uncertain
income and build greater trust between the customer and
their energy supplier.
Both public and private stakeholders are striving to take
advantage of the business opportunities these innovations
offer. For instance, the African Development Bank launched
a New Deal for Energy in Africa in 2016, with the highly
ambitious goal of achieving universal access by 2025,
including 75 million new off-grid connections. However,
the financial instruments offered by international donors
Solar farm in Cameroon (Zoussi road)
© Erwan Le Quentrec, 2017
January 2019uses and value 60
Access to electricity: challenges and opportunities for telecom carriers
However, as Figure 1 shows, there is a positive correlation
between access to electricity and internet use. In the
Democratic Republic of Congo and in Burkina Faso,
where annual growth in electricity access is below 1
percentage point (pp), the growth
rate of internet penetration is also
below 1pp. Conversely, in Kenya and
Botswana, where electricity access
is growing by more than 2pp a year,
the internet penetration rate is rising
by over 2pp. Thus, greater expansion
of electricity access comes with a
larger increase in internet use. This positive correlation is
confirmed by more rigorous estimates which exclude the
effects of confounding factors such as economic growth,
urbanisation, education, the price of internet access, and
investment in mobile networks. More precisely, we found
that a 1pp rise in the electricity access rate was accompanied
by a 0.34pp rise in the internet penetration rate. These
results are corroborated by the fact that electricity access
has a significant positive impact on smartphone adoption.
The smartphone penetration rate rises by 0.16pp following
a 1pp increase in electricity access. That impact comes
over and above the positive effects of economic growth,
urbanisation, education, investment and competition.
are generally unsuited to micro-projects, limiting, for the
time being, their capacity to finance decentralised off-grid
solutions. At national level, some states, particularly in East
Africa, have stepped up investment in the energy sector,
with encouraging results. However,
despite the positive impact access to
electricity has on economic growth,
such efforts remain difficult to sustain
in light of high levels of public debt and
the fact that countries’ priorities often lie
elsewhere: in health and education, for
Consequently, public spending in the energy sector is
increasingly being supplemented by private investment,
with the annual average rising from $1 billion between
2000 and 2011 to $5 billion between 2012 and 2015. What
role could a telecom carrier play in this climate? Are there
some links between its core business and energy access?
In general, telecom subscribers with no access to electricity
have to pay to recharge their devices. In Uganda, these costs
are estimated at 1-3 US dollars per month. With the advent
of heavy energy-consuming devices such as smartphones,
charging costs are likely to increase, thus
restricting internet use and potentially
slowing the revenue growth of telecom
carriers. On top of charging costs, users also
have to pay transportation costs involved
in travelling to the nearest charging point,
another obstacle to telephone use. The
electricity access deficit could, therefore,
hinder use of telecom services.
To test this hypothesis, we conducted a
study on the impact of access to electricity
on telecom usage practices in around
forty countries in Sub-Saharan Africa. For
each country, we collected data on access
to electricity; mobile phone, smartphone and internet
penetration; and average revenue per user (ARPU) for
the period 2000-2016. First, we estimated the impact of
access to electricity on mobile phone subscriptions and on
internet and smartphone use.
As might have been expected, we did not find any
statistical link between access to electricity and mobile
phone subscriptions. Indeed, mobile phones remain
widespread in Sub-Saharan Africa despite the lack of
access to electricity. The results of our analysis suggest
that economic growth, urbanisation, education and
competition among telecom carriers are the decisive
factors influencing mobile phone subscriptions.
Electrifying Africa
through o-grid
power and mobile
Second, we focused on how greater use of telecom services
due to electricity access translated into ARPU gains for the
carrier. Figure 2 suggests a positive relationship between
electricity access and ARPU. It shows that in Burkina-Faso
and Madagascar, where growth in electricity access does
not exceed 1pp per year, ARPU is falling by more than 5%
a year. Conversely, in Kenya and Ghana, where access to
electricity is growing by more than 2 percentage points
a year, the annual decrease in ARPU is less than 2%. Thus,
while ARPU is falling in all countries, it is falling less sharply
in those countries where access to electricity has developed
the fastest. Estimates taking account of the impact of
economic growth, investment in the grid and competition
Figure 1: electricity access and internet use (xed and mobile)
1 2 3
Orange Labs
Access to electricity: challenges and opportunities for telecom carriers
among carriers confirm this positive relationship. The
resulting ARPU gain is 43 cents per month for each user
connected to an electricity supply.
While the lack of access to energy has
not, thus far, proved an obstacle to
the deployment of telecom networks,
these results suggest it is becoming an
important factor in expanding the use
of technology, and in particularly of
the internet. In addition, some carriers
already have expertise in deploying off-grid solutions to
power their towers in rural areas. In such conditions, how
could they position themselves in the energy access value
irrigation or the sale of fresh products. In light of these
findings, should loan services not be offered in addition
to mobile payment, to help people buy the equipment
to take advantage of their energy supply? The company
SunErgy is currently pursuing such an
initiative in Cameroon, with the sale of
household appliances and industrial
equipment on credit, and providing
training sessions on entrepreneurship
for its customers.
To promote new technologies such as decentralised off-grid
energy solutions, it is often necessary to raise communities’
awareness of the costs involved in the alternatives. For
example, the lack of access to modern
energy sources leads to the use of
more expensive or even harmful
sources, such as biomass and coal.
Such sources account for more than
75% of energy consumption in Sub-
Saharan Africa, compared with 6% in
OECD countries. Households and, in
particular, women, spend between
one and five hours a day collecting
firewood: time that could be spent on
more productive activities. According
to the World Health Organisation,
exposure to toxic fumes due to the
use of firewood costs 600,000 lives
a year in Africa. The lack of lighting
at night reduces school children’s
learning time, with long-term
consequences on economic and
social development. As far as businesses are concerned,
the absence of a quality energy supply causes revenue
losses of around 5% in the formal sector and 20% in the
informal sector.
In short, access to electricity in Africa presents challenges
and opportunities for the core business of a telecom
operator, particularly when it comes to expanding internet
use. At the same time, sustainable business models and
technologies for mass electrification of Africa are still in
their infancy, providing ample scope for mobile carriers to
explore innovative solutions and opportunities within this
new electricity supply paradigm.
43-cent increase
per connected user
Figure 2: electricity access and average revenue per user (ARPU)
Telecom carriers are already involved in distributing
innovative energy supply solutions, including off-grid
facilities for domestic purposes, or “Solar Home Systems”
(SHS). These systems use mobile payment platforms for
billing and customer relationship management. However,
growing demand for electrical power is calling into
question the sustainability of this model. According to
a study by the University of Cambridge, only 30% of the
315 million Africans in rural areas who will have access to
electricity by 2040 will be connected to national grids, the
majority getting their electricity from SHS (25%) or mini-
grids (45%). In the same vein, a study by the NGO Practical
Action shows that a hybrid system combining mini-grids
and SHS is the least costly when we take account of long-
term changes in demand.
As regards business model, several studies have
demonstrated that growing energy consumption,
particularly in rural areas, depends on the concomitant
development of productive activities such as pumping,
0 1 2 3
Access to electricity and ICT usage: A country-level
assessment on Sub-Saharan Africa, Working Paper, 2018.
G. V. Houngbonon & E. Le Quentrec
Infrastructures Africaines : Une transformation impéra-
tive, Rapport, 2010. Banque Mondiale & AFD
African Village Entrepreneurship, Livre, 2016. Stein
Publication director:
Chantal Maugin
Editorial comittee:
Thomas Beauvisage
Anca Boboc
Laurence Dhaleine
Hélène Jeannin
Erwan Le Quentrec
Kevin Mellet
Joseph Messina
Technocentre / xdlab / SENSE
Valérie Peugeot
Anne-Sylvie Pharabod
Maryse Piart
Jean-Marc Raibaud
Mathieu Sannié
Christian Warocquier
Moustafa Zouinar
Design: Joëlle Paris
Find Uses and Value:
Informal businesses and digital technology
in Sub-Saharan Africa
• On the intranet:
Or on the page of departement SENSE:
uses and value
A newsletter about research in economic and social sciences
Informal businesses and digital technology in Sub-Saharan Africa
Orange SA au capital de 10 595 541 532 €
78, rue Olivier de Serres
75505 Paris cedex 15
350 129 866 RCS Paris
© Xavier Boulenger / Shutterstock
ResearchGate has not been able to resolve any citations for this publication.
ResearchGate has not been able to resolve any references for this publication.