Practice Notes

IMPROVING THE IMPACT OF MICROFINANCE ON POVERTY: ACTION RESEARCH PROGRAMME ISSN: 1740-4711 • NUMBER THREE • 2004

Learning from

Client Exit
is a global action research
What do we mean by
programme designed to
improve the quality of
microfinance services and their
Client Exit?
impact on poverty.
Imp-Act promotes the
development of reliable social
A L L M I C R O F I N A N C E institutions
(MFIs) have clients who decide to
leave their programme. Clients may
group, or that institutional changes need
to take place.
Second, knowing who ex-clients are
performance management leave for positive reasons – for example and why they leave is an important part
systems, which include impact they might have outgrown the size of of market research; it helps MFIs to
assessment. These systems the loans that the MFI can offer and be monitor client satisfaction. If clients are
reflect and respond to client graduating to formal sector finance. leaving because they are unhappy with
needs, as well as the priorities They may also leave for negative some aspect of the programme,
of microfinance institutions reasons, such as business failure or a managers can use this knowledge to
(MFIs) and their stakeholders. bad experience with the MFI. Further- make changes and improve the
The programme is a more, some clients who leave may decide programme. If clients are drawn to the
to return at some stage in the future. competition, managers will benefit from
collaboration between 30 MFIs
knowing what the competitors offer that
in 20 countries and a team of
How can knowing about client their programme does not. This is vital
academics from the UK
exits help an MFI? information in an increasingly
universities of Bath and Client exit is often thought of as competitive market.
Sheffield, and the Institute of inevitable for MFIs. Poorer clients may
Development Studies, Sussex not be protected against crises or How can these Practice Notes
University. The Imp-Act shocks. Richer clients may choose to help you?
programme was initiated by leave a programme whose products and These Practice Notes provide a guide to
the Ford Foundation, which policies are not appropriate for them. No helping your MFI track and understand
funds all Imp-Act activities. MFI will retain every client it has. client exit, in a simple, low-cost way.
However, as this practice note shows, They start by looking closely at what
unchecked high levels of client exits can client exit can mean for your
Imp-Act seriously affect both the financial and programme. They go on to show how
Imp-Act Secretariat the social performance of a programme. you can define and measure your MFI’s
Institute of Development Studies An MFI which finds out and tracks who exit rate. The notes then cover a number
University of Sussex leaves their programme and why they of tools you can use for finding out who
Brighton, BN1 9RE leave can avoid those dangers. is leaving your programme, and how
Telephone: 01273 873733 First, knowing who is leaving can be these tools or other data can help you
Fax: 01273 621202/691647 an indication of whether MFIs are draw up profiles of the different types of
Email: [email protected] meeting their social mission goals. If leavers. The next step is to find out why
Web: www.imp-act.org they aim to target and retain poor people are exiting from the programme.
www.microfinancegateway.org/ clients but these poor clients tend to Finally, the notes explain how to use the
section/resourcecenters/ leave, this may be an indication that the information you collect in the most
impactassessment products need to be tailored to suit this effective way.

I M P - A C T P R A C T I C E N O T E S • N U M B E R T H R E E • 2 0 0 4 • PA G E O N E
Client Exit
1 Why is it important invaluable in terms of getting to know
the advantages your competitors may FIGURE 1
to know about client have over your MFI. What you learn can MICROFINANCE CLIENT
enable you to fine-tune your products
exits? and policies to the requirements of
DESERTION MAGNITUDE

different types of clients in response to Exit rate All MFIs

Client exit can affect the success
market demand. If MFIs put their clients
of your MFI
first, they need to know if their products 60%
A rising exit rate may indicate major
are those that clients need and want.
problems for an MFI and even threaten 50%
its survival. Users may be unhappy with
Achieving your social goals 40%
terms and conditions, or relations with
Some clients’ departure from
staff. They may be switching to 30%
programmes will be for positive reasons.
competitors, or overall demand may be
It may be due to an improvement in 20%
falling due to a change in the economic
their financial situation, enabling them to
climate. Therefore, if you do not 10%
reduce their debts or ‘graduate’ to larger
research and address exit rates, this can
sums elsewhere, such as the formal 0%
have a serious effect on your MFI’s
banking sector. They may also be Africa Asia Latin CEE and
financial performance. America NIS*
benefiting from flexible services that
* Central and Eastern Europe and the Newly Independent States
allow a break from the loan cycle.
Understanding client exit is Source: MicroBanking Bulletin, April 2001
However, high or rising exit rates may
cost-effective for your MFI
also indicate that your clients are
The cost of attracting new clients is
dissatisfied with the quality of the
relatively high in microfinance, and you
services you are providing, or that they 2 How can you define
lose part of that investment with each
have been negatively affected or are in
client who leaves your programme after
a short time. If you are constantly
debt because of their participation. This and measure the exit
should concern MFIs, as the well-being
having to recruit new clients, staff time
of clients is usually their main priority. In
rate for your
will be taken up with administration,
which adds to the overall cost.
particular, if clients leave early, they may programme?
not have a chance to benefit from other
Furthermore, you will not benefit from
services that are built into the design of Exit rates need to be relevant
the faster portfolio growth from longer-
the loan programme, such as the to your organisation
term users who might take larger loans.
transfer of healthcare knowledge or As Figure 1 shows, exit rates vary
The longer-term financial dangers of
business skills. considerably across continents, and
exits are even more significant. High
across MFIs. This is partly due to the
drop-out rates can affect the reputation
Assessing the quality of your service different economic environments, design
of your MFI for potential new clients. It
Exit rates can be an indicator of how of products, levels of competition etc. It
can also affect clients’ loyalty and
good your targeting of clients has been, is also because exit rate
willingness to repay. In addition,
and which clients you are able to retain calculations are often inconsistent and
investors might be scared away (see
long enough for them to see positive based on badly-designed surveys. In
Case study 1).
impacts. This is particularly important for order to gain a clear and informative
programmes seeking to reach the picture of exits, a first concrete step is
Knowing your competitors
poorest people. to establish a well-defined exit status for
Learning from ex-clients can be
your clients. Below are some important
questions to ask in determining your
clients’ exit status.
CASE STUDY 1
At what stage does a member
become an ex-member?
SEF, the Small Enterprise Foundation monitoring. SEF combines on-going
Exit and retention (its opposite) rates
in South Africa, sees client drop-outs monitoring with in-depth drop-out
vary according to the time interval
as a critical indicator of both impact studies. In this way, it is able to
adopted for their measurement. For
and operational performance. They understand the reasons behind drop- example, if a village bank provides loans
found that drop-outs from their outs, and has been successfully on a routine 16-week cycle, it seems
Tshomisano project, launched taking immediate corrective obvious to define the exit rate as the
specifically to target the poorest, measures. As a result of exit proportion of borrowers in one cycle not
were often clients who had been in monitoring, SEF decreased its drop- taking a loan in the next cycle. However,
arrears or with portfolio at risk. SEF out rate from 35 per cent for 1998 to in reality these cycles are more flexible,
realised that exits are expensive to 14 per cent in 1999. because the struggle to repay one loan
reverse and decided to invest in exit to the village bank often delays receipt
of the next loan. It is therefore better,
where possible, to measure exit rates

PA G E T W O • I M P - A C T P R A C T I C E N O T E S • N U M B E R T H R E E • 2 0 0 4
over a standard period, such as one
year, or to at least specify the time CASE STUDY 2
interval to which a particular statistic
refers. PARTNER, in Bosnia and Herzegovina, come back to the programme at all,
found that 75 per cent of all of its or came back after one year. In the
How do you account for exiting clients, if they returned, came dynamically changing Bosnian
temporary exits? back after no longer than 250 days, market, annual changes within the
Some users may have only temporarily
while 50 per cent came back after no institution and outside are significant
left your programme, and plan or hope
longer than 150 days. After discussion enough to consider a client returning
to return at a later time. These people
with the staff it was decided to define after one year as a new client.
are known as ‘resters’. One reason for
a drop-out as a client who did not
resting might be the seasonality of many
businesses, especially those dependent
on a particular agricultural crop. The
should refer to the particular service When?
freedom to ‘rest’, i.e. not to borrow for a
clients have stopped using. You also need to know when clients are
cycle or two, allows clients with such
leaving. How many months or cycles
constraints to remain in the programme. What is an acceptable exit rate? have they been within the programme?
When you are calculating exits you need There is no universal figure below which Are they anticipating their repayment so
to think about whether or not you want you must get your MFI’s exit rate. You as to leave before a cycle is finished?
to include resters, or distinguish need to decide what is right for your MFI Are exits concentrated at specific times
between them and permanent leavers. and the environment in which it of the year?
operates. What is acceptable to an MFI
Have the clients really left? in terms of social performance will Why?
Be careful you do not exclude clients depend on what the programme is trying Even though you may be able to produce
who are still within your system. Take to achieve, who is leaving the statistics that tell you who is leaving and
into account clients who have switched programme, when and why, and what when, they do not tell you why clients
groups or dropped out of one type of impact both the participating and leaving leave or for how long. The best way to
loan programme within your MFI in order has had. For example, high exit rates know about clients’ past experiences,
to move to another. Your Management might be accepted by an organisation and whether they intend to rejoin, is to
Information System (MIS) might show such as SHARE in India, if it helps to ask them (see section 6 for more
these clients as being exits, but in fact sustain zero portfolio at risk. information).
they are not. These are issues you might
need to build into your exit rate
Building ex-client profiles
calculation, so as not to produce an 3 Profiling ex-clients: One way to start building profiles of
artificially high exit rate.
who, when and why? exiting clients is to sort them into
categories. This will enable you to
Over what time period should you
Who? establish patterns and make
calculate client exits?
It is important to know who is leaving – comparisons between categories,
There is no blueprint for choosing a time
is it older or younger clients? Is it first- groups, branches or different MFIs (see
interval for identifying exits. Instead
time borrowers or mature clients? Is it Case study 3). The categories you
MFIs need to take cultural and
men or women? Are those leaving choose will differ according to the MFI,
organisational factors into account (see
concentrated in a particular location or but where possible it is useful to clarify
Case study 2). Look at the type of
business sector? whether client exits are due to negative
clients you target, as well as data held
within your MIS. This can be a good first
step in assessing the most sensitive and CASE STUDY 3
appropriate time interval to adopt for
your MFI’s definition of a drop-out.
Prizma, an MFC (Poland) partner This allows Prizma to arrange focus
organisation also registered in Bosnia group discussions with people who
Which services are less popular
and Herzegovina, classifies its exit have left for similar reasons. Using
with clients?
clients into five categories: information from their MIS for further
Identify which services clients are
1. Voluntary, satisfied clients leaving segmentation, Prizma has been able
continuing to use and those they are
opting out of. For example, members for external reasons to find out more details about the
may take a loan holiday but continue to 2. Sleepers characteristics of leavers in each
make savings into the village bank, 3. Voluntary, dissatisfied clients who category. For example, it found that
attend meetings and participate in might return subject to changes in the voluntary, dissatisfied leavers
internal loan activities. If so, then it policy were most likely to be Serb nationals,
could also be argued that cycle-to-cycle 4. Voluntary, dissatisfied clients lost among whom older women were the
membership exit rates are zero between to the competition most likely to be lost to competition.
each cycle as well as over the whole 5. Forced out clients.
period. Therefore any exit rate statistic

I M P - A C T P R A C T I C E N O T E S • N U M B E R T H R E E • 2 0 0 4 • PA G E T H R E E
 Client Exit

or positive/neutral experiences of be doing two things: Decide who should collect data
services, and whether they are Again, there is no right or wrong answer.
1 Routinely monitoring how many
permanent or temporary losses. You may If you decide to use field staff to collect
people are exiting and what sort of
also want to distinguish between data, they will learn more about clients’
people these are. This needs to be an
voluntary leavers, whether these are lives and will feel a greater sense of
on-going process conducted quarterly or
satisfied or dissatisfied clients, and those involvement. On the other hand, clients
every six months at least. This will
who are forced out by the MFIs or their may not be as honest and open with
enable MFIs to identify worrying trends
fellow-borrowers. Another issue is to find people they know. Some MFIs such as
early and act before they grow into
out whether ex-clients have been lost to ODEF in Honduras have addressed this
significant problems.
competition or not. problem by switching staff from one
2 Following-up this monitoring branch to another for exit studies. It is
information with in-depth studies to also important to choose staff with the
4 Getting information address why they are leaving. Such time and experience to conduct in-depth
information can be gathered from exit interviews.
on exits routine exit forms, sample surveys,
focus group discussions or QUIP-type Decide when to collect data
Think about how you will get studies (see Resources), and can be Again, there is no set rule on when to
the data broad-sweeping or narrower in focus. collect data. If you collect data at the
Do not rely on staff or remaining clients’
A good compromise can be to combine point when clients leave the programme,
informal explanations for client exit; they
different types of approaches. SEF, in you will have fewer problems in tracing
may be biased. For example, clients who
South Africa, undertakes drop-out them or motivating them to gather for a
have left because they are unhappy with
monitoring on an ongoing basis through group discussion. On the other hand,
programme staff may be uncomfortable
group discussions. It follows up with their responses might be biased if they
saying so. You need to develop ways to
individual interviews of the ex-clients, are interviewed at exit, particularly if
systematically access reliable information
combined with data from their files and they are keen to keep their options open
about who is leaving the programme,
interviews with the field worker. and not displease the MFI with negative
and why.
opinions. Furthermore, a delay may help
In-depth information or a survey? Decide who you will interview them to think more about the reasons
The two main types of research methods If you have a large client base, you will why they left.
are quantitative, such as a survey, and not be able to interview everyone or
qualitative, where clients are asked to indeed want to, as this would be too
describe their situation in more detail. costly and time-consuming. This leaves 5 Choosing methods
You need to decide whether to use one or the problem of how you make sure of data collection
the other, or a mixture. One of the main different views are included in your data.
advantages of quantitative measures, Sampling is one way to solve this Decide which method suits
whether routine exit forms or surveys problem; you can reflect the diverse your needs
such as the SEEP-AIMS exit survey experiences of clients by randomly A range of available methods is available
questionnaire, is that they can give a selecting a fairly large sample of to help you collect information on client
good picture of drop-out patterns. This individual leavers or groups for exits, some of which are detailed below.
can reveal a lot to an MFI about both interview. The size of sample will depend You need to weigh up the benefits and
operational and impact issues. The on the total number and diversity of draw-backs of each and decide on one
drawback of a survey approach to exit clients. Even a very small sample is approach or a mix of them that will best
research is that it can fail to provide the better than relying on anecdotal meet your MFI’s needs, resources and
necessary detailed information on why information only. You can then increase time available. You also need to make
something has happened and how the sample size in later rounds, to see sure you do not spend time collecting
programmes can be changed to avoid the whether the extra effort results in information that already exists in your
same thing happening again. greater insights. MIS. Finally, the method you choose
Qualitative approaches, such as should be easy to repeat at a later stage.
in-depth case studies or focus group Encourage ex-clients to participate
discussions, can provide a fuller picture You may find that some ex-clients do not Focus group discussion
(see Imp-Act QUIP Practice Note for more wish to participate in surveys or studies Focus group discussions (FGDs) are
information). They allow the MFI to build because they left for negative reasons, useful because they enable you to collect
up a complex picture of the situation. or are simply not motivated to do so. in-depth qualitative information on
However, tools such as these need to be The only real way to avoid this situation particular issues. The discussions can be
carefully managed, through participant is to routinely collect information from fairly open-ended, meaning that
selection and skilled facilitation, to clients at the moment of exit. However, participants might bring up reasons for
enable all voices to be heard, not only if this is not possible, MFIs need to be exit that you had not previously
the loudest. creative. Promujer, in Peru, approached identified. However, FGDs are not always
the problem by proposing a raffle to easy to manage. For a group discussion
Aim to see both patterns and depth survey respondents, with household to really work, the participants need to
To be able to usefully monitor and goods as prizes. be from similar backgrounds and to trust
understand exits, MFIs should ideally each other.

I M P - A C T P R A C T I C E N O T E S • N U M B E R T H R E E • 2 0 0 4 • PA G E F O U R
Client Exit CASE STUDY 5

At SHARE in India, a first exit

study in one branch concluded
that the main reason for client
exit was migration out of the
Those leading the discussion need to from the MIS, so the interviews can be
area, a finding primarily based on
be skilled facilitators and experienced in short. The drawback is that a structured
staff opinion. A repeat exit study
this kind of research. Even when a group interview is not flexible enough to pick
is going well, it may be difficult to bring up on unexpected responses. It also two years later looked more
up certain sensitive issues. Finally, requires someone with quantitative closely at the matter and found
convincing ex-clients to participate in the analytical skills. that migration was not a main
group may be difficult; they may not be reason for leaving. It found a
motivated, have limited time to do so, or Self-completion survey more complex set of reasons for
even feel hostile towards the programme The main benefit here is the low cost of client exit, including local market
if they left for negative reasons. data collection because there is no need saturation, poor health,
for interviewers. It is a good way to unavailability of work, and the
Individual unstructured and avoid bias or embarrassment, as
closure of groups due to poor
semi-structured interviews respondents may remain anonymous. In
meeting attendance by some
Holding unstructured, one-to-one addition, this type of self-administered
members. Some of these reasons
interviews with ex-clients is a way to survey lends itself to ongoing monitoring
collect detailed descriptive information processes. However, there is no way to had led to clients migrating from
on particular issues, including more monitor who is actually filling out the the area.
sensitive ones. The drawback is that forms. In addition, the forms need to be
these interviews take time and are thus carefully piloted so that they are clear
usually limited in number, making it less for clients to understand and use. business has failed or that they
likely that you will identify a wide range experienced problems as a direct result
of reasons for exit. Using the management of breaking an MFI’s policies over loan
Semi-structured interviews have a information system utilisation, for example. Asking the direct
basic structure but the questions leave Centralised organisations keep question: ‘Why did you leave the
some room for the interviewee to give a computerised records of services being programme?’ will not result in accurate
longer, more descriptive answer. This used by individuals from which exit rates information, nor does it expose the
kind of interview can provide rich, can be generated. These are known as underlying reasons for exit (see Case
unexpected information and is less Management Information Systems study 4). Only with patience and
difficult to administer than the (MIS). This information may also be growing trust can these problems be
unstructured individual interview supplemented with information from exit overcome. Extensive piloting of
described above. The interview can be forms that leavers are required to questions is essential, as is careful
kept short, as much of the information answer before having savings returned training of interviewers. Particular care is
you need, such as credit history, can be to them. In addition to providing data needed when questions have to be
found in the MIS. for one-off surveys, access to translated, to ensure that the meaning is
In both cases, interviewers need to be information from an MIS facilitates the not lost. Other situations, such as cases
skilled in qualitative interviewing, and process of ongoing monitoring.
follow a standard process, so that the
interview data can be compared. Skill is CASE STUDY 6
also needed for analysing the
6 Finding out why
information. clients leave In Nigeria, LAPO’s first
organisation-wide exit research
Structured interviews Framing and analysing questions and client satisfaction studies
A structured interview is based on a set Even when interviewed alone, clients
seemed to indicate that clients
of pre-written questions to which only a may be unwilling to give the full story of
were leaving mainly because loan
short answer is expected. As with the why they left, particularly if the
sizes were too small for any sort
semi-structured interview, much of the questioner is a member of staff. They
client information can probably be taken may feel bad about admitting that their of productive use, as this is what
47 per cent of leavers claimed.
However, when a more in-depth
CASE STUDY 4 analysis was conducted, it
appeared that it was in fact
SEF, in South Africa, tries to build up many cases, the reasons they gave mainly richer, urban clients with
a picture of the ex-client’s experience, indicated deeper problems. They strong businesses, who were
starting with the least sensitive issues pointed to the fact that participation leaving after just one cycle. The
and building slowly to the reasons he in the programme had changed power study found that loan size was
or she left the programme. SEF found relations within the family, or they not a concern for the poorer,
that ex-clients’ standard responses to indicated financial difficulties, and longer-term exiting clients, who
direct questions about leaving were associated shame, due to business instead reported difficulty in
that leavers were ‘resting’, had ‘family failure. making payments and a decline in
problems’ or had ‘found a job’. In their businesses.

PA G E F I V E • I M P - A C T P R A C T I C E N O T E S • N U M B E R T H R E E • 2 0 0 4
Client Exit

of client dissatisfaction with field staff,

will also require delicate handling. CASE STUDY 7

There is usually more than one CARD, a Philippines MFI with a very who in turn were unwilling to let the
reason for client exit
advanced exit monitoring system, has poorest enter (or re-enter) their
An important point is that exit is never
investigated and experimented with a solidarity groups. As a response to
due to one single cause. However, for
number of approaches since it first this situation, CARD has tackled the
operational purposes, your MFI will need
to know which factors were particularly started monitoring exits in 1996. One problem in two main ways. First, it
important in pushing clients to leave of its most striking findings, after has introduced the BalikCARD
(see Case study 5). You can do this with undertaking regular exit surveys and programme for returning, vulnerable
ranking or scoring methods. However, if ‘progress tracking’ action-research, clients, who form their own groups
your information is from a focus group was that a large number of its exiting and receive special training and
discussion, the participants need to clients were the poorest members. support from staff. Second, it has
reach a collective decision on the main They were either being pushed out by addressed organisational structure
reasons. other group members, or were and staff development by updating
dropping out as the result of crises operational guidelines and training
Reasons for exit are different
and shocks. CARD found that, as a staff in approaches with vulnerable
for everyone
result, its staff were increasingly clients.
One of the challenges of investigating
exits is that for each client or group of tempted to attract less poor clients,
clients there may be a different
explanation. In-depth studies to follow-
up routine exit monitoring can provide larger database. For example, by rankings of reasons for exit by staff and
more details of each individual case (see merging exit and entry data for the user groups may highlight important
Case study 6). same sample of clients, the differences in their points of view.
characteristics of leavers may be
examined in more depth. Organisational learning
7 Collecting and and institutionalisation
• If you use the same data collection
using exit data methods in several areas, you will be
A main goal of your exits work should be
to record, analyse and present the
better able to make comparisons.
Link exit monitoring to other data information to management in a
If your MFI is committed to tracking and • The use of more than one method can digestible form with clear implications for
researching exits, it will need to be complementary. For example, the policy, products and services (see Case
institutionalise the way in which the data sample survey may indicate the relative Study 7). For this to be possible, you will
is both collected and used. A first step is importance of different reasons for exit, need to make sure that senior
to look at how you can make the most while in-depth interviews indicate the management understand and support
of your information. Below are some reasons why things happened. what you are doing, that operational
ways in which this can be done: staff are motivated to take on the work,
• Differences revealed in overlapping and that there are clear guidelines for
• If your MIS is effective, it may be findings may raise issues for further decision-taking.
possible to merge data to create a single investigation. For example, independent

Resources

Copestake, J., 2002, Working Paper, Kenya: selected Zambian Pawlak, K. and Michal M.,
‘Unfinished Business: the MicroSave microfinance institutions: 2004, ‘Client desertion – a
need for more effective causes and potential impact microfinance plague: how to
microfinance exit monitoring’, Imp-Act, 2004, ‘QUIP: on product design’, MicroSave diagnose it successfully?’ MFC
Imp-Act Working Paper 6, Understanding clients through Working Paper, Kenya: Spotlight Note 11, Poland:
Brighton: Institute of in-depth interviews’, Imp-Act MicroSave MFC
Development Studies Practice Notes No 2, Brighton:
Institute of Development Nelson, C. (ed.), 2000, Simanowitz, A., 2000, ‘Client
Hulme, D., 1999, ‘Client Studies Learning from Clients, exits: a tool for under-
exits (drop-outs) from Assessment Tools for standing client drop-out’,
East African microfinance Musona, D. and Coetzee, G., Microfinance Practitioners, Journal of Microfinance, Vol. 2
institutions’, MicroSave 2001, ‘Drop-outs among Washington: SEEP/AIMS No 1

I M P - A C T P R A C T I C E N O T E S • N U M B E R T H R E E • 2 0 0 4 • PA G E S I X