Using Lean Personas to the Description of UX-related Requirements: A
Study with Software Startup Professionals
Gabriel V. Teixeira and Luciana A. M. Zaina
Federal University of S
˜
ao Carlos (UFSCar), Sorocaba, S
˜
ao Paulo, Brazil
Keywords:
Lean Personas, User Experience, UX, Software Requirements, Software Practitioners, Software Startups,
Empirical Study.
Abstract:
User experience (UX) is a quality requirement widely discussed by software developers. Organizations have
targeted to offer software features that carry value to the audience. For software startups, UX-related re-
quirements can represent a competitive edge in their fast-paced environment with constant time pressures and
limited resources. However, software startup professionals often have little experience and lack knowledge
about UX techniques. Lean persona technique emerges as a slim form of constructing personas to allow the
description of end-users needs. In this paper, we investigated the use of the lean persona technique with 21
software professionals, 10 and 11 from software startups and established companies respectively. We carried
out a comparison to see whether the startup professionals use the technique in a different way from the estab-
lished company professionals. Our results revealed that the professionals of both groups used the technique
for similar purposes and wrote up UX-related requirements in different levels of abstraction. They also re-
ported positive feedback about the technique acceptance. We saw that the participants’ characteristics as years
of experience, prior knowledge about personas technique, or the fact of working in startups did not have an
influence on the technique acceptance.
1 INTRODUCTION
User experience (UX) is an important aspect of soft-
ware quality that affects software acceptance by the
users (Ohashi et al., 2018). The interaction needs of
the end-users with the software thus have to be taken
into account during the software production. Even
though there are different definitions of UX, most of
them state that UX encompasses both the software
functionalities and its quality characteristics that are
perceived by end-users during their interaction (Has-
senzahl, 2018). Usually, all software products deliver
some experience to the end-users that can result in a
positive or negative UX (Kashfi et al., 2017; Ohashi
et al., 2018; Zaina et al., 2021).
The topic of UX is not new in the software engi-
neering area. Practitioners and researchers have in-
vestigated challenges and problems in incorporating
UX in software development over the years (Kashfi
et al., 2017; Da Silva et al., 2018; Zaina et al., 2021).
The literature points out that there is a gap in stud-
ies about how tools can effectively help software pro-
fessionals to deal with UX in software development
practices (Da Silva et al., 2018). Recently, Zaina et
al. (2021) discuss the need to adopt techniques and
methods that support software professionals in iden-
tifying UX information that can aid in the software
development. Kashfi et al. (2017) reported that soft-
ware engineering professionals face barriers to incor-
porating UX into software processes. The literature
has pointed out that software professionals have dif-
ficulties in the elicitation activity for UX-related re-
quirements more than for functionality or other qual-
ity characteristics (Kashfi et al., 2017; Sch
¨
on et al.,
2017; Choma et al., 2016). UX-related requirements
describe and inform user needs (Hassenzahl, 2018).
For software startups, UX-related requirements
are relevant due to the experience with the product
being a factor that affects the user’s decision about
the use or not of products (Hokkanen et al., 2016).
Startups are companies with a focus on developing
innovative products or services (Paternoster et al.,
2014). These organizations differ from established
companies by searching for a scalable, repeatable,
and profitable business model with the aim of grow-
ing in the market (Paternoster et al., 2014). Startups
can work in different areas; however, they are char-
acterized by producing software or making intense
Teixeira, G. and Zaina, L.
Using Lean Personas to the Description of UX-related Requirements: A Study with Software Startup Professionals.
DOI: 10.5220/0011033200003179
In Proceedings of the 24th International Conference on Enterprise Information Systems (ICEIS 2022) - Volume 2, pages 211-222
ISBN: 978-989-758-569-2; ISSN: 2184-4992
Copyright
c
2022 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
211
use of software to manage their activities (Giardino
et al., 2014). These companies often operate with
a small team of professionals, exploring new tech-
nologies, being marked by rapid evolution, high un-
certainty about customers and market conditions, and
high failure rate (Paternoster et al., 2014). The liter-
ature has emphasized the relevance of studies about
software development practices to better support the
software practitioners in the fast-paced environment
where startups operate (Giardino et al., 2014). More-
over, startups software professionals usually are new-
comers with little professional experience and due to
this they struggle on using different software meth-
ods and techniques (Paternoster et al., 2014; Giardino
et al., 2014). These difficulties are extended to the
use of UX methods and techniques (Hokkanen et al.,
2016).
UX-related requirements elicitation has been con-
ducted with the support of different techniques, and
the Personas technique is one of them (Faily and
Lyle, 2013; Castro et al., 2008). The personas tech-
nique provides a process for creating personas arti-
facts from the analysis of end-user data. These arti-
facts present fictional characters that represent the dif-
ferent user types that can use a service or product and
help the software practitioners to understand users’
needs, experiences, behaviors, and goals (Billestrup
et al., 2014a). However, the traditional persona con-
struction is often seen as a costly technique in terms
of time, and effort (Billestrup et al., 2014b). Gothelf
proposes a leaner process for developing personas,
called proto-personas and also known as lean per-
sonas (Gothelf, 2012). The process allows the cre-
ation of personas artifacts based upon prior knowl-
edge that the practitioners have about the target audi-
ence (Gothelf, 2012).
Lean personas is a technique that can be suitable
for startup needs of adopting leaner practices that de-
mand few resources and time to be used (Paternos-
ter et al., 2014). However, little has been explored
about the use of these lean UX practices by startup
professionals. To address this, we investigate the fol-
lowing research questions: (RQ1) What type of UX-
related requirements do software startup profession-
als describe by using lean personas technique?; and
(RQ2) What are the participants’ feedback about the
use of lean personas technique?
To answer the RQs, we conducted a study with
21 software professionals, 11 from established com-
panies, and 10 from software startups, who built
lean personas artifacts to describe UX-related require-
ments. The participants worked in software compa-
nies from diverse segments (e.g., financial, health) in
Brazil. They occupied development-related positions
(e.g. product owners, software engineers, develop-
ers). From the data collected, we compared the re-
sults of the two groups to understand whether there
were differences in the use and acceptance of the
lean persona by professionals from startups and es-
tablished companies. Our decision of comparison is
due to the fact that the literature points out that startup
professionals have little experience and usually have
demands for techniques and tools that are more ad-
herent to the startup context. We examine the lean
personas produced by the participants looking for ev-
idence on descriptions of UX-related requirements to
analyze the use of the technique. We also collected
participant acceptance of the technique’s use from an
online questionnaire.
2 BACKGROUND
2.1 User Experience
In general, the literature states UX as the
user/customer interaction with the product and
company (Lallemand et al., 2015). In a different
perspective, Hassenzahl (2018) points out that the
user experience is related to the users’ motivation
when interacting with a product. According to the
author, users are motivated by goals they want to
achieve, and these goals, in turn, guide the user’s
interactions with a product.
Hassenzahl formalizes UX definition in a content-
oriented model which contains three levels of interac-
tion action, i.e., ‘why”, “what” and “how” and a goal,
that is to achieve the “wellbeing” of users (see Figure
1). The why level focuses on the meaning that the
interaction can provide for users; it discusses the mo-
tivations and needs that lead the user to use the prod-
uct. The what level defines the functionalities that
the product offers to the users to fulfill their needs.
The how level explores the concrete actions of the
users for interacting with the product, e.g. clicking
on a button or reading instructions. We decided to
adopt the content-oriented model of UX in our study
because our focus is on identifying the specifications
about UX that were described from the lean personas.
We consider that this model provides concrete ele-
ments to support our analysis instead of other abstract
definitions of UX.
2.2 Related Work
The personas technique has been used to support
software teams in the identification of end-users re-
quirements. An investigation with 60 software com-
ICEIS 2022 - 24th International Conference on Enterprise Information Systems
212
Figure 1: Content-oriented model of UX adapted from
(Hassenzahl, 2018).
panies showed that only 7 of them used this tech-
nique (Billestrup et al., 2014b). The companies that
did not use the technique recognized it as relevant for
identifying end-user requirements. However, they re-
ported not being able to use it in their daily work due
to problems such as lack of knowledge about how to
apply the technique, lack of time, and lack of finan-
cial resources (Billestrup et al., 2014b). In another
work, software engineers reported that the personas
technique not only allows a better understanding of
the user’s needs, but it also supports these require-
ments being elicited up-front to the design and coding
activities (Sim and Brouse, 2014).
The results about the effectiveness and efficiency
of adopting personas artifacts during requirements
identification was presented by Salmien et al. (2020) .
The authors compared the use of personas with re-
quirements elicitation made using available data in an
analytics tool (Salminen et al., 2020). The results re-
vealed that the developers spend more time on iden-
tifying the requirements which were caused by the
need of interpreting a great amount of data and graph-
ics available from the analytics tool. On the other
hand, the personas artifacts presented end-user data
in a straightway which make the developers work be-
come faster (Salminen et al., 2020).
PATHY technique was proposed to help develop-
ers identify requirements (Ferreira et al., 2017). Its fo-
cus is to help software engineers recognize user char-
acteristics and present an overview of product func-
tionality. PATHY is an adaptation of the empathy
map, which is a variation on the persona technique.
Its differential is a set of guideline questions that
guide the development team to build personas that ad-
here to the target audience (Ferreira et al., 2017).
Pinheiro et al. (2019) proposed the Proto-
Persona+, which is an extension of the proto-persona
technique proposed by Gothelf (2012). Similar to
the PATHY, the Proto-persona+ provides guideline
questions to support personas’ construction. The au-
thors conducted a study in which the use of proto-
personas proved to be effective for eliciting UX re-
quirements with the participation of software engi-
neers and domain experts. The results also showed
that the proposal contributed to both software engi-
neers and domain experts writing UX requirements
relevant to software development.
The literature above shows evidence of the rele-
vance of personas technique for requirements elici-
tation. However, as far as we know, no studies in-
vestigate the use of the personas technique by startup
software professionals. In particular, we argue that
the lean UX techniques such as lean persona are more
suitable for software startups where few resources are
available, and professionals often have little experi-
ence (see discussion in Section 1).
3 LEAN PERSONA
Gothelf (2012) proposes the lean persona technique,
also known as proto-persona, as an agile and low-cost
alternative for creating personas. The main difference
between the traditional persona and the lean persona
is the order of executing of the steps for its construc-
tion (Gothelf, 2012; Billestrup et al., 2014a). Rather
than building the personas based on a broad demo-
graphic and profile survey of end-users, lean personas
are built by taking into account the prior knowledge
that the software professionals have about the end-
users (Gothelf, 2012).
The lean personas are constructed during work-
shops that were conducted by a moderator who is an
expert on this technique application with the partici-
pation of software professionals. First, the lean per-
sonas can be built individually by the software profes-
sionals. Later, the workshop participants examine all
the lean personas and they together refine them to pro-
duce a reduced number of personas. In an extension
of Gothelfs proposal, Pinheiro et al. (2019) present
the Proto-Persona+, an quadrant-based approach pro-
posal to constructing lean personas. Proto-Persona+
features a four-quadrants template to enhance the de-
scription of users’ information whereas keeping the
characteristic of being a lean artifact.
The four quadrants have the following functions:
(Q1) Demographic data contains the users’ charac-
terization relevant to the development of the product,
including an image that represents the persona; (Q2)
Objectives and needs presents the users’ goals and
the needs to achieve these goals; (Q3) Behaviors and
preferences show details of how the user likes to carry
out tasks to achieve their goal, and their preferences
regarding contents and interaction modes; and (Q4)
Difficulties point out the user’s difficulties and frustra-
Using Lean Personas to the Description of UX-related Requirements: A Study with Software Startup Professionals
213
Figure 2: Lean persona tool.
tions the users can have during their interacting with
the product. To guide the personas elaboration, the
Proto-Persona+ approach provides a set of guideline
questions for each quadrant.
According to Pinheiro et al.(2019), there is a
dilemma about the use of personas technique by soft-
ware professionals. While the personas technique
is recognized as useful by these professionals, they
often struggle in putting this technique into prac-
tice (Billestrup et al., 2014b). Besides, sometimes the
personas present information that was not relevant to
the development scope (e.g. hobbies, personas’ pets)
which can cause a miscommunication about the core
requirements (Billestrup et al., 2014b). In this way,
the guideline questions help software professionals to
focus on descriptions that fulfill the goals of the quad-
rants and keep the attention on the important require-
ments. Pinheiro et al. (2019) state that their proposal
is flexible by allowing to the addition of other guide-
line questions under demand.
In our study, we decided to adopt the proposal
of Pinheiro et al. (2019) due to it provides a tem-
plate and guideline questions which guide the soft-
ware professionals on the artifacts construction. Be-
sides, the authors proposal have been evaluated to the
UX requirments specifications previously. Based on
the Pinheiro et al. proposal, we developed the lean
persona tool (see Figure 2). The tool automates and
makes easy the use of the Proto-Persona+ technique.
The lean persona tool provides two views, one for the
moderator of the session to the personas construction
and the other for the participants, i.e. software pro-
fessionals, to fulfill the template and thus produce the
personas artifact.
The tool provides the moderator with a project-
based organization (see Figure 2(a)). For each project,
the tool delivers the default template which contains
the guideline questions (see Figure 2(b)). New guide-
line questions can be added or the available ones can
be modified. After preparing the template, the moder-
ICEIS 2022 - 24th International Conference on Enterprise Information Systems
214
ator makes it available for the participants of the ses-
sion of the lean persona creation from a link generated
by the tool (see sharing link field in Figure 2(a)). The
participants then create the lean personas, individu-
ally or in group ((see Figure 2(c)). The moderator can
access all the personas created by the participants (see
Figure 2(d)). The tool is available at the link
1
.
4 EXPERIMENTAL STUDY
Our study was conducted following the recommenda-
tion of Wohlin et al. (2012) . Our study was approved
by the Federal University of S
˜
ao Carlos ethical com-
mittee (CAAE process: 37663220.5.0000.5504).
4.1 Planning
Participants were invited to take part in the study vol-
untarily via our professional network. Our research
group has conducted different investigations about
UX in industrial settings, which established contacts
with software professionals. The participants were se-
lected by convenience, according to their availability
to participate in the study (Wohlin et al., 2012). Both
developers from startups and established companies
were invited. The participants were divided into two
distinct groups, i.e., one with software startup profes-
sionals and another composed of software profession-
als from established companies.
Personas elaboration is tightly associated with the
domain of the product; we first determined the soft-
ware focus in our study. We defined tourism mo-
bile apps as the domain of our study, being TripAd-
visor
2
and Google Maps
3
examples of these type of
apps. The tourism app was chosen because it rep-
resents a common-sense domain. Consequently, the
participants could concentrate their efforts on creat-
ing the personas artifacts, avoiding the need of learn-
ing about the app domain. Considering the app do-
main, we prepared the lean persona tool to support
our study. We kept the demographic questions (i.e.,
Q1 quadrant) suggest by the lean persona tool and de-
fined the quadrants guideline questions for the other
quadrants (see Table 1).
An online questionnaire was elaborated to gather
professionals’ profile data (e.g., years of experience,
company market segment, position in the company,
experience with tourism apps, and knowledge on per-
sonas techniques). The questionnaire included the
1
http://uxtools.uxleris.net/
2
https://www.tripadvisor.com.br/
3
https://www.google.com/maps/
Informed Consent Form to collect the participants
agreement in taking part in the study. We also cre-
ated an online feedback questionnaire based on the
Technology Acceptance Model (TAM) (Davis, 1989)
to collect the participants’ perceptions about the use
of lean persona technique. TAM questions are divided
into two dimensions The perceived usefulness dimen-
sion represents how much a person considers that the
use of a specific technology may improve their per-
formance. The ease-of-use dimension is related to
the perception that technology can be adopted with
no effort. We added three open questions about the
suitability of the lean persona on the creation of the
artifacts in the feedback questionnaire.
Finally, we wrote up a scenario of users’ interac-
tion with the tourism app domain. This scenario in-
tended to overview this kind of app to support the par-
ticipants during the personas elaboration. However,
the participants were free to use all their knowledge
about this kind of apps.The scenario is described as
follow: “A tourist usually uses a mobile application
(mobile, tablet, etc.) to plan and guide their trips. The
app displays places in a city or region according to
the interest and searches made by users. These places
can be hotels, monuments, museums, parks, restau-
rants, among others. For each location, the apps
show details such as the location name, photos, ad-
dresses, and feedback from other individuals that of-
ten is ranked by a score from 0 to 5. Moreover, these
apps can present estimates of expenses to spend in the
places and also comments to assist the user to decide
and plan visits to places that have good ratings”.
A senior expert with 10+ years of experience in
UX area evaluated our study reviewing and refining
the profile questionnaire and the other artifacts. Be-
sides, a pilot test was carried out in our research group
with five UX researchers and we concluded that no
changes were necessary and the study could be run.
Table 1: Guideline questions for the quadrants Q2 to Q4.
Quadrant Guideline questions
Q2
(GQ1) What does he/she want to achieve?
(GQ2) What does he/she need to accomplish his/her goal?
Q3
(GQ3) What does he/she like?
(GQ4) What is he/she outstanding to do?
(GQ5) How does he/she like to do it?
Q4
(GQ6) What are the difficulties he/she has?
(GQ7) What barriers/obstacle can he/she face in a
task/an interaction?
(GQ8) What can disappoint him/her?
4.2 Execution
The study was conducted with 21 software profes-
sionals
4
who work in different companies in Brazil,
4
Professionals in software development positions as
such as developers, interface designers
Using Lean Personas to the Description of UX-related Requirements: A Study with Software Startup Professionals
215
being 10 professionals from software startups and 11
from established companies. In our previous contact
with the participants, we observed that most had little
knowledge about the personas techniques. To level off
the participants’ knowledge, we recommended they
watch two videos about persona and lean persona be-
fore their participation in the study.
Due to the COVID-19 pandemic, the study took
place in online meetings using the Google Meet tool
5
.
To attend to the different schedule availability of the
participants, we ran 6 sessions distributed on different
days and each one lasted about 1 hour and a half. In
each session, we had the participation of 2 to 4 pro-
fessionals. Each participant took part in only one ses-
sion. A researcher with 1+ year of experience in UX
was responsible for running the study. In all sessions,
the same script was followed, thus avoiding the inser-
tion of biases in the study conduction.
At the beginning of each session, the researcher
welcomed participants and briefly pointed out the
study’s objective. The participants accepted the term
of consent to data and images for academic ends and
thus completed the profile questionnaire. Before start-
ing the construction of the lean personas, the partic-
ipants took part in a warm-up session where the re-
searcher presented a quick review covering the per-
sonas and lean personas concepts (i.e., based on the
videos watched by the participants previously). The
researcher thus carried out hands-on training using
the lean persona tool to familiarize participants with
its resources. After the hands-on training, the partic-
ipants built the lean personas considering the tourism
app scenario. They were oriented to create at least one
and up to three personas due to the limitations of the
study time slot. In the end, 21 participants produced
24 personas in total. The participants answered the
feedback questionnaire (i.e., TAM questions) at the
end of the study.
4.3 Analysis
We examined the 24 lean personas produced by the
participants and the answers from the feedback ques-
tionnaire. We carried out a qualitative analysis on the
learn personas artifacts to examine the presence of the
different types of UX-related requirements. We ex-
plored the Q2, Q3, and Q4 quadrants to search for ev-
idence of UX descriptions. These three quadrants pro-
vide descriptions about goals and needs, behavior and
preferences, and difficulties, respectively. In our anal-
ysis, We did not consider the Q1 quadrant because
it only informed personas demographic data and we
were interested in the requirements’ descriptions.
5
https://meet.google.com/
Figure 3: Example of labeling of lean personas artifact -
why in red and delimitated with circles; what in green and
delimitated with squares; how in purple and delimitated
with triangles.
Our analysis on the lean personas was carried out
in three steps. First, one junior researcher (i.e., 1+
year of experience) explored all the lean personas by
applying a closed coding technique (Corbin, 1998).
Closed coding guides researchers in identifying ex-
cerpts of text and labeling them to pre-established
codes which belong to a codebook (Corbin, 1998).
In our study, the codebook was composed of the UX
levels proposed by Hassenzahl (2018), i.e., the codes
why”, “what and “how”. The respective definitions
that each code carries guided the junior researcher to
look for evidence about UX specifications that were
presented in the learn personas. Figure 3 illustrates
our labeling activity in a lean persona that was pro-
duced by a startup professional. A senior UX re-
searcher revised and refined all the labels assigned to
the 24 lean personas as a second step. After, the two
researchers worked together to map all the labeling
(“why”, “what and “how”) to the guideline questions
(see Table 1). Consequently, we could see the rela-
tionship between the labeling and the lean personas
quadrants. Finally, we explored the participants’ re-
sponses regarding the acceptance of the lean personas
technique. We also examined whether the partici-
pants’ profiles had influenced the technique accep-
tance.
4.4 Threats to Validity
We followed the recommendations of Wohlin et al.
(2012) to discuss the four threats to validity (i.e., con-
clusion, construct, internal, and external) and how we
proceeded to mitigate them.
We use different data sources to give reliability to
our conclusions from the results. The adoption of UX
levels and their definitions to explore the lean per-
ICEIS 2022 - 24th International Conference on Enterprise Information Systems
216
sonas avoided the possible bias that could be intro-
duced by the researchers on the qualitative analysis.
Moreover, the closed coding process was conducted
in three steps with the participation of two different
researchers. All the codes assigned to the learn per-
sonas have been revised by researchers with experi-
ence in qualitative analysis. We also examined the
feedback questionnaire to give conclusions about our
study. To mitigate construct problems, we followed
the same script to conduct the study sessions. We used
the same artifacts (i.e., the lean persona tool and the
tourism app scenario) with all the participants to sup-
port the construction of the lean personas. We also
conducted a warm-up session providing the concepts
about lean personas to the participants. A hands-on
exercise helped the participants to handle the lean
persona tool. These two activities allowed us to miti-
gate the impact of little knowledge of the participants
about personas. The lean persona tool is fully adher-
ent to the proposed use of the template and the guid-
ing questions for building lean personas.
To mitigate the participants’ fatigue, an internal
validity threat, we conducted short sessions that lasted
no more than 1 hour and a half. By using the lean
persona tool, the participants employed less effort in
handling different artifacts during the creation of the
personas. All the information to carry out the activity
was available in the tool, i.e., the lean persona tem-
plate and the guideline questions to fulfill the quad-
rants. Concerning the external validity, the partic-
ipants who took part in the study had positions in
startups and established companies which led us to
consider we had a representative sample to our study
goals.
5 FINDINGS
We present our findings in four sections: the partici-
pants’ profile, the answers to the RQ1 and RQ2 (see
in Section 1), the influence of the participants profile
in our results.
5.1 Participants’ Profile
Table 2 presents an overview of the participants’ pro-
file. The profile questionnaire answers reveal that
from 10 participants of startups, half of them had <
1 year of professional experience and only one had
4-6 years. Considering the personas and lean per-
sonas techniques, we see that more than half had only
theoretical knowledge or never heard about the tech-
niques. We also notice that the more experienced pro-
fessional, i.e. 4-6 years, reported having only theo-
retical knowledge about these techniques. Looking
at the established company participants’ information,
we notice that 7 from 11 had theoretical and practical
knowledge about the personas technique; however, a
smaller number of participants had the same knowl-
edge about the lean personas. All the participants, i.e.
from startups and established companies, have used
some kind of tourism app which represents that they
had knowledge of the domain as end-users.
5.2 RQ1 - Type of UX-related
Requirements
To answer RQ1 (What type of UX-related require-
ments do software startup professionals describe by
using lean personas technique?), we examined the 24
personas produced by the 21 participants. We looked
for evidence of descriptions related to the three lev-
els of the Hassenzahl model, i.e., why”, what and
how (see Figure 3). We did two granularity levels
of mapping, one more general at the level of lean per-
sonas quadrants and another more specific at the level
of guideline questions (see Table 1).
Figure 4 shows a heat-map chart with the distri-
bution of occurrences of the three levels of the Has-
senzahl’s model (see Figure 1) by the lean personas
quadrants. The results show a concentration on the
level “what in the three quadrants (see Figure 4 (b))
in which product features are described. In Figure 4
(b), we see this results independently of the partici-
pants were professionals from startups or established
companies. From the level why (see Figure 4 (a)),
we observe that startup professionals did not give de-
tails on the end-users difficulties (i.e., Q4 quadrant).
The occurrences in Q4 are low even for established
company’ professionals. The results also demonstrate
that the level how (see Figure 4 (c)) has the lowest
number of occurrences independently of the quadrant.
By exploring the relationship between the descrip-
tions available in all quadrants, we notice that Q2
presents the most information. It predominantly de-
scribes more abstract UX-related requirements which
report why and what the users need to achieve their
objective. Q3 provides descriptions of how (i.e., be-
havior) to do related to why (i.e., preferences). In
Q4, UX-related requirements appear describing prod-
uct functionalities (i.e.,“what”) in a high level of ab-
straction to provide information on how to mitigate
the difficulties that the user could have. The descrip-
tions concretely explain “how the interaction should
occur to minimize issues regarding the use of the soft-
ware.
We also examined the descriptions in relation to
the guideline questions per participants’ group, i.e.,
Using Lean Personas to the Description of UX-related Requirements: A Study with Software Startup Professionals
217
Table 2: Participants profile - Knowledge about the techniques is classified as (1) I’ve never heard of; (2) I only have
theoretical knowledge; (3) I used it a few times; (4) I used it a many times.
(a) Identification (b) Domain
(c) Knowledge
about the techniques
ID Group
Company
market-segment
Position
Professional
Experience
Tourism
apps often used
Personas Lean Persona
P1 Established Company Software Factory Product Owner 8+ years
Google Maps, TripAdvisor
and Youtube
2 2
P2 Established Company Finances UX Research 4-6 years Google Maps and Youtube 3 2
P3 Established Company R&D IT Analyst < 1 year Youtube 3 4
P4 Established Company Education IT Analyst 8+ years Google Maps and Youtube 1 1
P5 Established Company Financial Institution
Software
Engineer
8+ years TripAdvisor and Youtube 3 1
P6 Established Company Electronics Assembler IT Analyst 6-8 years Google Maps and Youtube 2 2
P7 Established Company Software Factory
Full-stack
Developer
1-3 years Google Maps and Youtube 3 3
P8 Established Company Software Factory Internship < 1 year Google Maps and Youtube 3 3
P9 Established Company IT IT Analyst 1-3 years Google Maps 3 3
P10 Established Company Software Factory Internship < 1 year Google Maps 3 3
P11 Established Company Health
Software
Engineer
8+ years
Google Maps, TripAdvisor
and Youtube
2 2
P12 Startup Web Development Internship < 1 year Google Maps 2 3
P13 Startup Education
Software
Engineer
1-3 years Google Maps and Youtube 1 1
P14 Startup Software House
Front-end
Developer
1-3 years Google Maps 1 1
P15 Startup
Mobile Device
Management
Back-end
Developer
< 1 year Google Maps 4 4
P16 Startup
3D Solutions
for real estate
Developer 4-6 years Google Maps and Youtube 3 2
P17 Startup Process Mining IT Analyst 1-3 years Google Maps and Youtube 4 4
P18 Startup IT Consulting Developer < 1 year Google Maps 1 1
P19 Startup Software Factory Internship < 1 year Google Maps 2 2
P20 Startup Software Factory Developer 1-3 years Google Maps 3 3
P21 Startup Shop Streaming
Front-end
Developer
< 1 year Google Maps and TripAdvisor 2 2
Figure 4: UX-related requirement occurrences mapped by quadrants and Hassenzahl’s levels - (Q2) objectives and needs;
(Q3) behavior and preferences; (Q4) difficulties.
startups and established companies. However, we saw
that the difference in the number of occurrences in
each group for each level (i.e., why”, what and
how”) has no impact on our results. Figure 5 presents
a Sankey diagram which lists the influence of the
guideline questions on the results. Sankey diagram
is a flow diagram in which the width of the arrows is
shown proportionally to the flow quantity. It helps
locate dominant contributions to an overall flow
6
.
We can see the type of UX-related requirements (i.e.,
based on UX levels) versus the guideline questions.
Figure 5 reflects a greater concentration on the use
of guideline questions to describe UX-related require-
ments of the what level regardless of the quadrant.
Examples of descriptions found out in the results re-
ferring to this level are: Need to find reviews, infor-
mation, prices, and locations of places (referring to
question GQ2), and “[...] the lack of filtering for per-
forming searches (referring to question GQ8). This
result confirms what had already been evidenced by
6
https://www.r-graph-gallery.com/sankey-
diagram.html
the heat-map chart (see Figure 4). Specifically, we see
that guideline questions GQ1 and GQ3 of the quad-
rants Q2 and Q3, respectively, like the ones that most
guided the participants in describing the why”. In
GQ1, we found out answers as for instance: Plan
and guide your [the persona] next trip with friends to
an ecological city”; while in GQ3, we also found out
description to explain the why”, for example: Driv-
ing on beautiful roads with many landscapes”. Fi-
nally, questions GQ6, GQ7, and GQ8 (Q4) were the
ones that were most linked to how level, in which
we found descriptions as the following “”[...]there is
limitation of characters to the post my contributions”.
5.3 RQ2 - Feedback about the Use of
Lean Persona
Taking into account the responses collected from the
online questionnaire, we could answer the RQ2 (What
are the participants’ feedback about the use of lean
personas technique?). We analyzed the participants’
feedback regarding the perceived usefulness and ease-
ICEIS 2022 - 24th International Conference on Enterprise Information Systems
218
Figure 5: Relation of the guideline questions and the type
of UX-related requirements.
of-use of the lean persona technique (see Table 3).
We used a four-point Likert scale as answer options,
ranging from strongly disagree to totally agree. Fig-
ures 6 and 7 illustrate the results of the acceptance
feedback for the two groups, i.e., startups profession-
als and established companies professionals. We see
that the lean persona had a good acceptance for both
groups. In particular, P4 and P14 assigned low ratings
of acceptance in most of the answers regarding easy
of use perception. These participants were from es-
tablished company and startup, respectively (see Ta-
ble 2). Looking at their profiles in Table 2, we notice
that both declared to “have never heard about the per-
sona and lean persona techniques which can explain
their ratings. However, P4 and P14 did not give any
suggestion of improvements to technique.
We also considered the comments added by the
participants in the three open question of the ques-
tionnaire. We asked to the participants to give a justi-
fication for their point of view. First, the participants
answered about the support of the guideline ques-
tions to the lean persona construction. Most of the
participants reported that the questions cover the im-
portant points to the lean personas construction; they
also mentioned the lean characteristics that the tem-
plate carries as a positive feedback. Unlike, four par-
ticipants, P7, P9, P14 and P15 reported they wanted
to have other questions options to built the lean per-
sonas with more details. Looking at to their profile
(see Table 2), we see that they all have practical expe-
rience with personas techniques which explain their
demands of having space for giving more detail about
the persona.
The next two questions asked whether questions
should be added or removed from the template. The
most of suggestions for addition of new questions
came from the established company participants. In
general, they mentioned that questions about the
product brand, more details on the product features,
technologies that the personas have expertise, could
become the lean persona description more complete.
P2 mentioned the wish of having a space on the tem-
plate to add [...]a statement that represents the per-
Figure 6: Startup professionals’ feedback.
Figure 7: Established company professionals’ feedback.
sona to provide more empathy to it”; P2’s answer
could be explained by the fact that s/he had more
knowledge on persona in which is usual to provide
a statement about the persona (see Table 2). For the
question of removing guideline questions, we see that
the most of suggestions were given by startups’ par-
ticipants. We observed that the quadrant Q3 was cited
by 3 participants from startups. Some the comments
are: I think the quadrant Q3 makes me to be lit-
tle confused about the first and third questions (P15)
and [...] when I was filling the quadrant Q3, I have
the impression that 2 questions have the same goal
(P11)”.
Table 3: Feedback questionnaire based on TAM (Davis,
1989).
Dimension ID Question
Perceived
of
Usefulness
U1 I find lean persona useful for supporting the identification of
UX-related requirements.
U2 Using the lean persona technique allowed me to achieve the
result I want.
U3 Using lean personas technique supported me for quickly
specifying details about end-users.
U4 Using lean personas technique allowed me to write charac-
teristics of representative end-users.
U5 Using lean personas technique made me to getting better in
describing end-users’ characteristics.
Perceived
of
Ease-of-use
F1 It was easy to learn how to use the lean persona technique.
F2 I find the lean persona technique easy to remember.
F3 I find it easy to understand the quadrant goals.
F4 I could use the lean persona technique as I want.
F5 It was it easy to get ability to use the lean persona technique.
Using Lean Personas to the Description of UX-related Requirements: A Study with Software Startup Professionals
219
5.4 Influence of Professional’ Profile in
the Acceptance
We decided to run the Fisher’s exact test
7
to check
whether the participants’ profile could have influence
on the acceptance of the lean persona technique. We
took the Fisher’s exact test because it allows com-
paring categorical data collected from small samples.
It calculates the exact significance of the deviation
from a null hypothesis using the p-value, while other
methods use an approximation. In addition to provid-
ing a p-value with greater accuracy in small samples,
the exact significance tests do not require a balanced
or well-distributed sample (Mehta and Patel, 1996)
which matches our sample’s characteristics. A 95%
(i.e., 0.05) confidence interval was considered to mit-
igate errors in the results
8
. We compiled the data for
running three different Fisher exact’s testing which
are available in Table 4 (a), (b) and (c).
First, we checked whether the company type af-
fected the acceptance results (see data in Table 4 (a)).
We thus defined the null and alternative hypothesis,
taking into account the answers of all TAM’s ques-
tions (see Table 3), as follows: H0 - The company
type where the participant work has no influence in
the acceptance, and H1 - The company type where
the participant work has influence in the acceptance.
After running the test, we got the results available in
column p-value in Table 4 (a). The results suggest
that there is no statistical evidence to support that the
fact of working in startups or established companies
influence on the acceptance of the lean personas use.
On our second testing, we defined the null and
alternative hypotheses as follow: H0 - The previous
knowledge of the participant on the Gothelf s lean
persona technique has no influence in the acceptance,
and H1 - he previous knowledge of the participant
on the Gothelfs lean persona technique has influence
in the acceptance. Prior knowledge of the technique
was as yes for those participants that declared to
have already used the technique. We collected this
data in the profile questionnaire from the question
Have you ever know the lean persona technique be-
fore this study?”. Table 4 (b) supported our testings
from which we see that all the p-values indicate that
there is no statistical evidence of influencing of the
previous knowledge on the technique upon the accep-
tance.
Finally, we carried out another testing to see
whether the professional’s experience could impact
on the acceptance. We used the data displayed in Ta-
7
https://www.statstest.com/fischers-exact-test/
8
We run tests from this website https://astatsa.com/
FisherTest/
ble 4 (c). The null and alternative hypotheses were de-
fined as follows: H0 - The years of professional’s ex-
perience have any influence on the technique accep-
tance, and H1 - The years of professional’s experience
have influence on the technique acceptance. We ob-
tained the results of the p-values available in Table 4
(c) which lead the rejection of null hypotheses. The
results suggest that there is no statistical evidence that
professional’s experience affect the acceptance on the
lean personas technique.
6 DISCUSSION
Taking into account our RQ1 (What type of UX-
related requirements do software startup profession-
als describe by using lean personas technique?), the
results reveal that the lean persona technique moti-
vated the participants of the two groups, i.e., startup
and established company professionals, to the writ-
ing up of UX-related requirements which cover all
the UX levels, i.e. why”, what”, and how (see
the results in Section 5.2). Looking at Figure 4, we
see a similar distribution in the four quadrants of the
lean personas considering the levels of UX. In our un-
derstanding, the use of the lean personas technique
can stimulate the professionals to describe end-users
needs even in cases where the professionals have little
experience and little knowledge in the personas tech-
nique, which are common in startups.
Our results also showed evidence that the guide-
line questions encourage the participants in describ-
ing “why the user has a given need in different quad-
rants (see Figure 4). According to Hassenzahl (2018),
the discussion about the reasons for transforming a
set of needs into requirements is essential for the soft-
ware developers to have better insights about the soft-
ware solutions. This conversation can also lead the
software teams to be more conscious in conceiving
features that bring value to the audience (i.e., “wellbe-
ing level of the Hassenzahl’s model - see Figure 1).
The delivery of valuable products is even more vital in
a startups environment where the professionals need
to quickly react to new target audience demands in
a highly competitive market (Hokkanen et al., 2016;
Paternoster et al., 2014). Therefore, we conclude that
lean persona can be a helpful technique to support the
low experienced software startup professionals on the
UX-related requirements specification.
Our results showed a positive acceptance of the
technique, i.e., the answer of our RQ2 ( What are the
participants’ feedback about the use of lean personas
technique?). Regardless of the group, i.e., profession-
als from startups or established companies, the partic-
ICEIS 2022 - 24th International Conference on Enterprise Information Systems
220
Table 4: Results of Fisher exact’s test - (1) Strongly disagree; (2) Partially disagree; (3) Partially agree; (4) Totally agree;—
Group: (G1) Startup; (G2) Established Company.
Influence
of the type of company (a)
TAM Group (1) (2) (3) (4) p-value
U1
G1 0 0 3 7
0.387
G2 0 0 6 5
U2
G1 0 1 6 3
0.659
G2 0 0 6 5
U3
G1 0 0 1 9
0.098
G2 0 1 5 5
U4
G1 0 1 4 5
0.380
G2 0 2 1 8
U5
G1 0 0 4 6
0.362
G2 0 0 2 9
F1
G1 0 1 2 7
0.183
G2 0 0 6 5
F2
G1 0 1 5 4
0.505
G2 0 0 4 7
F3
G1 1 2 4 3
0.524
G2 0 3 2 6
F4
G1 0 1 4 5
1.000
G2 0 2 4 5
F5
G1 0 0 4 6
0.565
G2 0 2 4 5
Influence of
the professionals experience (b)
TAM
Professional
experience
(1) (2) (3) (4) p-value
U1
3+ years 0 0 3 4
1.000
<=3 years 0 0 6 8
U2
3+ years 0 0 4 3
1.000
<=3 years 0 1 8 5
U3
3+ years 0 0 3 4
0.742
<=3 years 0 1 3 10
U4
3+ years 0 1 1 5
0.816
<=3 years 0 2 4 8
U5
3+ years 0 0 1 6
0.613
<=3 years 0 0 5 9
F1
3+ years 0 0 3 4
1.000
<=3 years 0 1 5 8
F2
3+ years 0 0 3 4
1.000
<=3 years 0 1 6 7
F3
3+ years 0 3 0 4
0.117
<=3 years 1 2 6 5
F4
3+ years 0 1 2 4
0.827
<=3 years 0 2 6 6
F5
3+ years 0 2 1 4
0.058
<=3 years 0 0 7 7
Influence of the knowledge
on lean personas technique (c)
TAM
Knowledge
on personas techniques?
(1) (2) (3) (4) p-value
U1
Yes 0 0 2 7
0.184
No 0 0 7 5
U2
Yes 0 0 4 5
0.367
No 0 1 8 3
U3
Yes 0 0 1 8
0.159
No 0 1 5 6
U4
Yes 0 0 3 6
0.314
No 0 3 2 7
U5
Yes 0 0 2 7
0.659
No 0 0 4 8
F1
Yes 0 0 4 5
1.000
No 0 1 4 7
F2
Yes 0 0 3 6
0.519
No 0 1 6 5
F3
Yes 0 3 2 4
0.914
No 1 2 4 5
F4
Yes 0 0 4 5
0.360
No 0 3 4 5
F5
Yes 0 0 3 6
0.557
No 0 2 5 5
ipants considered the technique easy to use and use-
fulness to writing UX-related requirements (see Sec-
tion 5.3). Billestrup et al. (2014) had pointed out
that software professionals had difficulties in using
the personas technique. Our study restates the results
of Pinheiro et al. (2019) about the positive acceptance
of the lean persona technique. Our positive results can
be influenced by the guideline questions, which aided
the participants to keep the focus on describing help-
ful information on the target audience. Ferreira et al.
(2017) and Pinheiro et al. (2019) had already high-
lighted that the guideline questions support develop-
ers in a better use of the personas technique.
Nonetheless, our findings allowed us to identify
problems with the guideline questions in the Q3 quad-
rant (Behavior and preferences - see Table 1). We
conclude that we need to sift through these questions
to thus decide about their changes. Considering dif-
ferent profile characteristics, we did different Fisher
exact testings to see the influence of the profile par-
ticipants in the acceptance results (see the results in
Section 5.4). We noticed that the participants’ char-
acteristics did not have an impact on the acceptance
of the lean persona technique.
Based on the related work (see Section 2.2), we
see that our study presents important contributions
discussing the use of the lean personas technique to
support the startup professionals in the writing of UX-
related requirements. As far as we know, no other
work had carried out a similar investigation. We can
state that the works of Pinheiro et al. (2019) and Fer-
reira et al. (2017) are the closest to ours. Pinheiro
et al. (2019) had different purposes in their research:
to assess whether non-technical stakeholders, i.e., do-
main experts, could write UX requirements using the
lean persona technique, they called Proto-Personas+.
The study conducted by Ferreira et al. (2017) ver-
ified the writing of requirements in general without
the focus on UX requirements. Although Pinheiro et
al. (2019) looked at UX requirements, they did not
examine them through theoretical lenses of the UX
models proposed by Hassenzahl (2018) .
Our results show evidence about the type of UX-
related requirement that the technique encourages
participants to write up. Besides, our findings can
provide insights for future improvements on the qual-
ity of UX-related requirements specification. More-
over, software professionals took part in our study,
which differed from the studies of Pinheiro et al.
(2019) and Ferreira et al. (2017) that carried out eval-
uations with students.
7 CONCLUSION AND FUTURE
WORK
In this paper, we investigated with 21 software pro-
fessionals the support of lean persona technique in
describing UX-related requirements. We developed a
web tool to make the creation of lean persona artifacts
easier and more automated. The results showed that
our lean persona proposal aided both groups of pro-
fessionals, i.e., software startups and established com-
panies, to focus on writing up end-user needs in dif-
ferent levels of abstraction. The needs described de-
scriptions of features, ways to interact with the prod-
uct, and the reasons for those specifications existed.
Finally, the professionals of both groups showed pos-
itive perceptions regarding the ease of use and useful-
ness of the lean persona technique.
As contributions our study presents evidence that
the lean personas technique can help professionals
Using Lean Personas to the Description of UX-related Requirements: A Study with Software Startup Professionals
221
who have little experience in describing UX-related
requirements, which are commonly found in software
startups. The use of slim and easy-to-learn techniques
is suitable to the low-resource and fast-paced environ-
ment of startups. We also contributed by developing
the web lean persona tool which automated the pro-
cess of persona construction. As future work, we in-
tend to conduct case studies in software startups set-
tings to see the usefulness of the lean persona tech-
nique in the daily work of small software teams.
ACKNOWLEDGMENTS
We thank the support of grant #2020/00615-9, S
˜
ao
Paulo Research Foundation (FAPESP). We also
National Council for Scientific and Technological
Development CNPq, grant 313312/2019-2 and
149650/2021-3 for supporting this work.
REFERENCES
Billestrup, J., Stage, J., Bruun, A., Nielsen, L., and Nielsen,
K. S. (2014a). Creating and using personas in soft-
ware development: experiences from practice. In In-
ternational Conference on Human-Centred Software
Engineering, pages 251–258. Springer.
Billestrup, J., Stage, J., Nielsen, L., and Hansen, K. S.
(2014b). Persona usage in software development: ad-
vantages and obstacles. In The Seventh International
Conference on Advances in Computer-Human Inter-
actions, ACHI, pages 359–364. Citeseer.
Castro, J. W., Acu
˜
na, S. T., and Juristo, N. (2008). Inte-
grating the personas technique into the requirements
analysis activity. In 2008 Mexican International Con-
ference on Computer Science, pages 104–112. IEEE.
Choma, J., Zaina, L. A. M., and Beraldo, D. (2016).
Userx story: Incorporating ux aspects into user stories
elaboration. In International Conference on Human-
Computer Interaction, pages 131–140. Springer.
Corbin, A. S. (1998). Basics of qualitative research: Tech-
niques and procedures for developing grounded the-
ory. volume 4, CA: Sage. Thousand Oaks.
Da Silva, T. S., Silveira, M. S., Maurer, F., and Silveira, F. F.
(2018). The evolution of agile uxd. Information and
Software Technology, 102:1–5.
Davis, F. D. (1989). Perceived usefulness, perceived ease of
use, and user acceptance of information technology.
MIS Q., 13(3):319–340.
Faily, S. and Lyle, J. (2013). Guidelines for integrating per-
sonas into software engineering tools. In Proceedings
of the 5th ACM SIGCHI symposium on Engineering
interactive computing systems, pages 69–74.
Ferreira, B., Santos, G., and Conte., T. (2017). Identifying
possible requirements using personas - a qualitative
study. In Proceedings of the 19th International Con-
ference on Enterprise Information Systems - Volume
2: ICEIS,, pages 64–75. INSTICC, SciTePress.
Giardino, C., Unterkalmsteiner, M., Paternoster, N.,
Gorschek, T., and Abrahamsson, P. (2014). What
do we know about software development in startups?
IEEE Software, 31(5):28–32.
Gothelf, J. (2012). Using proto-personas for executive
alignment. UX Magazine, 1.
Hassenzahl, M. (2018). The thing and i (summer of ’17
remix): From usability to enjoyment. pages 17–31.
Hokkanen, L., Kuusinen, K., and V
¨
a
¨
an
¨
anen, K. (2016).
Minimum viable user experience: A framework for
supporting product design in startups. In Sharp, H.
and Hall, T., editors, Agile Processes, in Software En-
gineering, and Extreme Programming, pages 66–78,
Cham. Springer International Publishing.
Kashfi, P., Nilsson, A., and Feldt, R. (2017). Integrat-
ing user experience practices into software develop-
ment processes: implications of the ux characteristics.
PeerJ Comput. Sci., 3:e130.
Lallemand, C., Gronier, G., and Koenig, V. (2015). User
experience: A concept without consensus? exploring
practitioners’ perspectives through an international
survey. Computers in Human Behavior, 43:35–48.
Mehta, C. and Patel, N. (1996). SPSS exact tests.
Ohashi, K., Katayama, A., Hasegawa, N., Kurihara, H., Ya-
mamoto, R., Doerr, J., and Magin, D. P. (2018). Fo-
cusing requirements elicitation by using a ux measure-
ment method. In 2018 IEEE 26th International Re-
quirements Engineering Conference (RE), pages 347–
357.
Paternoster, N., Giardino, C., Unterkalmsteiner, M.,
Gorschek, T., and Abrahamsson, P. (2014). Software
development in startup companies: A systematic map-
ping study. Information and Software Technology,
56(10):1200–1218.
Pinheiro, E., Lopes, L., Conte, T., and Zaina, L. (2019). On
the contributions of non-technical stakeholders to de-
scribing ux requirements by applying proto-persona.
Journal of Software Engineering Research and Devel-
opment, 7:8:1–8:19.
Salminen, J., Jung, S.-G., Chowdhury, S., Seng
¨
un, S., and
Jansen, B. J. (2020). Personas and analytics: a com-
parative user study of efficiency and effectiveness for
a user identification task. In Proceedings of the 2020
CHI Conference on Human Factors in Computing Sys-
tems, pages 1–13.
Sch
¨
on, E.-M., Winter, D., Escalona, M. J., and
Thomaschewski, J. (2017). Key challenges in agile
requirements engineering. In Baumeister, H., Lichter,
H., and Riebisch, M., editors, Agile Processes in Soft-
ware Engineering and Extreme Programming, pages
37–51, Cham. Springer International Publishing.
Sim, W. W. and Brouse, P. S. (2014). Empowering require-
ments engineering activities with personas. Procedia
Computer Science, 28:237–246.
Wohlin, C., Runeson, P., H
¨
ost, M., Ohlsson, M. C., Reg-
nell, B., and Wessl
´
en, A. (2012). Experimentation in
software engineering. Springer Science & Business
Media.
Zaina, L. A., Sharp, H., and Barroca, L. (2021). Ux in-
formation in the daily work of an agile team: A dis-
tributed cognition analysis. International Journal of
Human-Computer Studies, 147:102574.
ICEIS 2022 - 24th International Conference on Enterprise Information Systems
222