Investigating Aha Moment Through Process Mining

Wan-Hsuan Chiang

1 a

, Usman Ahmad

2 b

, Shenghui Wang

1 c

and Faiza Allah Bukhsh

1 d

Faculty of Electrical Engineering, Mathematics and Computer Science (EEMCS), University of Twente,

Enschede, The Netherlands

Management Science Department, DHA Suffa University Karachi, Pakistan

Keywords:

Aha Moment, Process Mining, Customer Satisfaction.

Abstract:

Aha moment is when users realize the value of using the software product, which is a key to driving rev-

enue, mainly for B2B SaaS vendors. According to the Acquisition-Activation-Retention-Referral-Revenue

(AARRR) model, the aha moment can refer to activation, and the following customer phase is retention. This

research aims to ﬁnd customers’ ”aha moment” through process mining. Since the customers in retention are

obligated to experience activation before, the research ﬁrst identiﬁes milestone actions in terms of product

features and user roles to cluster the retention customers. Evaluation is performed through the data of the

clustered customers from the time before they move to the retention phase. The event log analysis is discussed

based on multiple dimensions: product solution, time, and user roles. The research mainly applies the process

mining technique, heuristic miner, to discover the customer’s behavior patterns. Apart from marketing fun-

nels, Moreover concept of human-computer interaction is focused on event classiﬁcation and data cleaning,

which is practical for cleaning UI logs. The discovered processes and aha moment can guide future product

development and value proposition re-stratigizing.

1 INTRODUCTION

Digital e-platforms, Amazon or Airbnb, have dis-

rupted the fundamental structure of the industries

through the change in essential customer interaction

structure (Parker et al., 2016). Particularly, the digital

platforms of the business-to-consumer (B2C) sector

have already established their potential. However, in

business-to-business (B2B) digital platforms are still

work in the process. In B2B the digital platforms en-

able the smooth interaction between supply and de-

mand sides through e-infrastructure (Rix et al., 2020).

In B2B e-models, the managers face a real challenge

of customer retention (Rix et al., 2020). For instance,

in Indonesia, the customer retention rate of BIZGO

is very low and customer relationship management is

recommended to avail the opportunities in digital plat-

forms (Situmorang and Harmawan, 2022).

The role of digital marketing is increasing in the

post-pandemic era and several B2B ﬁrms have started

using digital marketing to increase customer acqui-

https://orcid.org/0000-0002-6370-6356

https://orcid.org/0000-0001-8591-6904

https://orcid.org/0000-0003-0583-6969

https://orcid.org/0000-0001-5978-2754

sition (Assal, 2022). Digital marketing enables busi-

nesses to connect with speciﬁc customers through tar-

get marketing activities (Teixeira et al., 2023). Al-

though digital marketing beneﬁts almost every B2B

ﬁrm nevertheless, there is a dearth of literature on

this topic (Shaltoni, 2017; Pandey et al., 2020). In

fact, various businesses believe that digital platforms

work for B2C organizations only (Teixeira et al.,

2023; Lacka and Chong, 2016). However, the busi-

ness models of Cisco and IBM are considered the

success stories of the B2B digital platforms (Teix-

eira et al., 2023; Venkatesh et al., 2019). Today due

to digital media, the customers of B2B have ample

access to information of products and services and

this information aids B2B customers to make better

decisions. However, traditionally customer relation-

ship management remains a problem for B2B orga-

nizations, including online B2B ﬁrms (Teixeira et al.,

2023; Hochstein et al., 2020).

In order to cope with the challenges of customer

relationship management in B2B platforms, the con-

cept of user growth was developed by turning the

”passengers on digital website” into customers. This

practice helped the various start-ups to retain a large

number of users in a short period of time with very

low or even zero investment. Based on this practice

164

Chiang, W., Ahmad, U., Wang, S. and Bukhsh, F.

Investigating Aha Moment Through Process Mining.

DOI: 10.5220/0011848800003467

In Proceedings of the 25th International Conference on Enterprise Information Systems (ICEIS 2023) - Volume 1, pages 164-172

ISBN: 978-989-758-648-4; ISSN: 2184-4992

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

of increasing users, a few frameworks have gradually

formed in the market that effectively grows the users

on digital platforms. In these models, the AARRR

model is the most frequently mentioned and used

framework in user growth. AARRR is a set of in-

dicators developed by the Dave McClure, the founder

of 500 Start-ups, that refers to ﬁve stages; acquisi-

tion, activation, retention, revenue and referral (Mc-

Clure, 2022; Zhang, 2021; Lin and Chaomin, 2021).

The activation begins with the ﬁrst happy visit of the

user which is also called as ”Aha! moment” Initially,

the term is deﬁned as an insight that represents a sud-

den cognitive change and was adapted to product de-

velopment, indicating the time when the users realize

the value of the product (Carpenter, 2019). The fol-

lowing customer phase of activation is retention. In

this phase, businesses consider whether users contin-

uously engage with the product, and the frequency of

usage is calculated to understand the repeated behav-

iors. This repeated behavior leads to the retention of

customers and plays a vital role in customer relation-

ship management. Through process mining, we can

explore users’ repeated behavior patterns which may

lead to improvement in customer relationship man-

agement.

Software companies, especially startups, are

struggling to develop product strategies. Finding the

aha moment is the key to growth, which guide the

companies in deﬁning strategies. The aha moment

is a set of actions the users realize the value of the

product (Stancil, 2015; Balboni, 2022). The idea is

often bound with the customer lifecycle where the

customer funnel consists of ﬁve steps: acquisition,

activation, retention, referral, and revenue (McClure,

2022; Zhang, 2021; Lin and Chaomin, 2021). The

aha moment usually associates with activation. Users

who ﬁnd the value of the products are more likely to

be retained.

The objective of this paper is to study the case

of K, a B2B company with a Software-as-a-Service

(SaaS) deployment model, and investigate the ”Aha!

moment”. The study intends to use the event logs gen-

erated by clients of K to ﬁnd the ”Aha! moment,”

based on customers’ repeated behavior. Therefore,

the objectives of this study(OS) are as follows:

OS1 - Distinguish the repeated behavior pattern

from retained customers.

OS2 - Finding the aha moment to understand cus-

tomer’s delight.

The research sheds light on applying process min-

ing to customer lifecycle analysis. Applying process

mining techniques to discover the aha moment is an

innovative approach among previous studies. Intro-

ducing human-computer interaction to data cleaning

provides a brand-new perspective to re-organize event

logs. The results provide the business with a guideline

for future product development. The research frame-

work can be further applied to other B2B SaaS cus-

tomer analyses.

This paper ﬁrst introduces the background knowl-

edge and methodology used in this research. Section

3 introduces the research questions associated with

the objective statement, as well as the framework used

in the following experiment. Section 4 demonstrates

the process models of the empirical study. We wrap-

up the paper in Section 6 and Section 7 focusing on

conclusions and future work.

2 BACKGROUND

2.1 The Aha! Moment

Companies, especially software service providers,

struggle to deﬁne the business strategies that cus-

tomers love, which can be referred to as the aha mo-

ment (Aha!, 2022). The Aha! moment is a set of

actions the users realize the product’s value (Stancil,

2015; Balboni, 2022). Service providers, especially

start-ups, change their product strategies dynamically

to test the product market ﬁt or value propositions.

Discovering the aha moment guides companies to de-

velop product roadmaps based on customers’ experi-

ences rather than companies’ claims.

2.2 Process Mining

Businesses pay attention to analyzing their customers

in order to improve their products and services. How-

ever, the existing solutions cannot solve the problems

of developing product strategies in companies with-

out clear value propositions because the vendors are

still discovering.

There are many commercial analytic tools, such

as Google Analytics, Mixpanel, and Tableau, are

adopted for customer analysis. However, researchers

mentioned the limitation (Vinod et al., 2013; Terragni

and Hassani, 2018). Using these tool to analyze cus-

tomer journeys require a clear understanding of the

primary processes, which is not suitable for discover-

ing customer patterns.

Why Process Mining: Quantitative research is

powerful during the discovery phase. However, user

researchers still need to capture the primary scenar-

ios before recruiting users. However, quantitative

methodologies are often costly and time-consuming,

which is not practical for small businesses.

Investigating Aha Moment Through Process Mining

165

Thanks to the widely spread data-driven concept,

most companies are used to storing and leveraging

data for decision-making. Therefore, process min-

ing discovery techniques become the most sufﬁcient

tools for discovering customer behaviors. Process

mining adopts event logs to analyze sequential pat-

terns (van der Aalst, 2016). Process discovery is part

of process mining that can build models without a

priori knowledge, making it the most powerful and

commonly-used technique in process mining research

(Zerbino et al., 2021; Guzzo et al., 2022).

Previous process mining papers, however, mainly

focus on workﬂow and operational management

(Corallo et al., 2020; Zerbino et al., 2021). The lim-

ited papers discovering customer behaviors mainly

focus on navigation and learning behaviors (Vinod

et al., 2013; Zaim et al., 2018; Husin and Ismail,

2021; Taub et al., 2022). None of the studies adopt

process mining to discover an aha moment for devel-

oping product strategies.

Meanwhile, process mining also shows the po-

tential to integrate with other customer research dis-

ciplines, such as customer journey mapping (Chan

et al., 2021; Bernard and Andritsos, 2017; Terragni

and Hassani, 2018; Terragni and Hassani, 2019) and

human-computer interaction (Theis and Darabi, 2019;

Else et al., 2019; Liu et al., 2018; Cerone, 2015). Pre-

vious papers show the possibility of process mining to

support analyzing customer behavior in the evidence

base.

Figure 1: Initial Process Model (Part), Showing the Com-

plexity of Uncleaned Data.

Process Discovery Techniques: This research

applies the miners that are capable of addressing nois-

iness because the real world is full of noise, as seen

in Figure 1. Heuristic miner (Weijters et al., 2006)

and fuzzy miner (G

unther and van der Aalst, 2007) are

the most widely used miners among previous experi-

mental process mining papers (Zerbino et al., 2021;

Corallo et al., 2020). In this study, fuzzy miner is

used to produce the test models in order to consult

stakeholders due to the intuitive output process map;

heuristic miner becomes the primary algorithm be-

cause the miner takes frequency into account to cap-

ture the primary behaviors.

We use ProM Lite 1.3 to build process models;

Python and Google BigQuery to process data.

2.3 Domain Object

In this study, we adapt the fundamental concept

of human-computer interaction to restructure the UI

logs. The domain object, from the set of potential

objects of interest for the user of a given applica-

tion, forms the basis of the interaction and its purpose

(Beaudouin-Lafon, 2000). Each domain object can be

viewed as a virtual object on the software for users to

create, edit, and change attributes.

The domain object is derived from the object of

interest, which is associated with the ﬁrst proposed vi-

sual interaction style, direct manipulation, to replace

the command-line interface in the 1980s. The ob-

ject of interest is a virtual representation that can be

directly manipulated (Shneiderman, 1982; Shneider-

man, 1983).

The concept of virtual objects and actions plays a

vital role in data cleaning. We adapt the concept to

reduce the number of events, preserving the semantic

meanings. More details can be seen in Section 4.2.

3 METHODOLOGY

In order to discover the aha moment, our research

questions are as follows:

RQ1 - What are the behavior patterns of the

clients in the retention phase for B2B platforms?

RQ2 - What is the “Aha! moment” of clients for

B2B platforms, leading to retention?

Figure 2: Research Plan.

Adapted from CRISP-DM (Wirth and Hipp,

2000), Figure 2 shows the research plan of this paper.

There are ﬁve steps:

Step 1 - Understanding: This step contains busi-

ness and data understanding in CRISP-DM.

ICEIS 2023 - 25th International Conference on Enterprise Information Systems

166

Step 2 - Initial Data Preparation: This step ex-

tracts and cleans the data from the data warehouse.

Step 3 - Retention Clustering: The retention

clustering answers RQ1as discussed in Section 4.3.

Step 4 - Discover Aha! Moment: The output

from the previous step is the input of this step to build

a process model for discovering the user activation.

Step 5 - Stakeholder Feedback: Aligned with the

evaluation phase in CRISP-DM, is the ﬁnal step of

this project, discussed in Section 5.

Step 2, Step 3, and Step 4 represents the loops data

preparation and modeling. Each step contains more

than 10 small and big iterations. The following dis-

cussions only write down the big iterations.

Step 5 includes stakeholders’ views on the mod-

els. However, the involvement of the domain knowl-

edge does not limit to the ﬁnal stage.

This study adopts the business context from K,

a B2B SaaS vendor providing performance tracking

service for customer support (CRM) teams. The com-

pany actively collected customer behavior data in the

past. We select the customers that installed the soft-

ware between December 27, 2021, and June 26, 2022.

The records start from December 27, 2021, to July 3,

2022. We make sure the records cover the accounts’

actions for more than a week. The accounts for inter-

nal use or uninstalled within a week are ﬁltered out.

4 ANALYSIS

4.1 Dimensions

We recognize several dimensions to structure the

analysis during the business understanding phase in

Step 1. These are the main factors we consider during

the research. Besides the time dimension, the study

can be discussed under 6 scenarios, with 3 pillars and

2 role types. Each scenario has its own process model

as it has its particular behavior pattern.

4.1.1 Pillar and Domain Object

The study can be divided into 3 pillars, representing

the solutions provided by the product: quality assur-

ance, missions, and performance coaching. Each so-

lution aims to solve a particular pain point from target

customers. We also recognize the domain object as-

sociated with each pillar.

4.1.2 Role Type

The role in B2B software products for operational

needs often refers to permission, reﬂecting the real-

world position in the organization. In our study, the

product has 5 different roles. The lowest is the agent,

while the highest is the account owner. If we consider

performance tracking as a game requiring two parties

to interact, the roles can be classiﬁed into manager

and agent, and the former supervises the latter.

4.1.3 Time

The business customers follow the weekly operation

cycle from Monday to Sunday. Customers who have

repeated behaviors week over a week are considered

as retention, or actively using such features.

4.2 Data Cleaning

The job in this subsection is done during the data

preparation in Step 3, retention clustering. However,

since the event classiﬁcation inﬂuence most reshapes

the implementation at the very beginning, this sub-

section can be viewed as part of Step 2, initial data

preparation.

The diverse event naming leads to spaghetti-like

models (Figure 1). The original event logs are named

by UI components, design purpose, user actions, and

domain objects. The UI components change over time

and versions; the design purpose frequently changes

in order to test different user ﬂows.

Therefore, we adopt the concept of domain object

as discussed in Section 2.3 to diminish the interfer-

ence of user interface and design purpose. By focus-

ing on the domain object and the actions that the user

act on, the event naming reﬂects the motivation be-

hind it. For instance,

performance_journey_coachable_moment

_comment_created

event can be divided into Create Comment and

Coaching card. The latter is the associated domain

object of one of the pillars. The former reﬂects that

users act on the attribute Comment of the virtual ob-

ject.

We further identiﬁed the duplicated events and

nodes without semantic meanings. The number of

nodes, in the end, is reduced by 62%.

4.3 Retention Clustering

We ﬁrst consult stakeholders from customer success

team about which behaviors can be recognized as re-

tained customers. The ﬁndings can be extracted in

Figure 4 that retained customers in a particular pillar

should contain milestone actions from both parties,

inspired by identifying key actions of business goals

Investigating Aha Moment Through Process Mining

167

for preprocessing (Peng and Cheng, 2009). In ad-

dition, the behaviors should repeat week over week,

according to the deﬁnition of retention and business

understanding. However the frequency is not clearly

identiﬁed within the business.

Figure 4: Pillar, Domain Object, Role Types, and Milestone

Actions.

After the interview, we extracted 4 scenarios with

clear milestone actions. We select the users’ work-

ing days with milestone actions as cases to build the

process models.

The remaining 2 scenarios consider the domain

object as cases. The node combines action and

role type that who acts on the objects can be clearly

shown. The process models uncover the actions on

each domain object, as well as the interaction between

managers and agents. Figure 3a shows the process

model of the domain object of performance coach-

ing. The nodes highlighted in red are the actions from

managers; the nodes highlighted in green are the ac-

tions from agents. The process starts from Create by

managers and then Open by agents. Before the end,

managers Open the domain object again for review-

ing.

Based on the retention modeling, milestone ac-

tions of all scenarios are veriﬁed and recognized. Fig-

ure 5 are the dotted charts in terms of pillar. The logs

are colored by the milestone actions of both parties.

The y-axis is the business customers, and the x-axis

is week to capture the weekly patterns. The milestone

actions, however, ﬁltered out the test actions, such as

self-creating domain objects. The companies contain-

ing the milestone actions from both role types week

over week are considered as retained customers. As

a result, three sets of retained customers are clearly

identiﬁed.

4.4 Discover Aha! Moment

This subsection addresses RQ2 what the aha moment

looks like in terms of pillar and user role? Based

on the previous retention clustering, the retained cus-

tomers are selected in terms of pillars.

The event sequences of the business customers are

sliced into two periods: activation and retention. We

select the time period before the clustered customers

start the retention patterns in order to associate the ac-

tivation period. That is, the output models start from

Installation and end at Start Retention.

Meanwhile, the events that are not ”constructive”

are ﬁltered out to simplify the model. Such events

(a) Retention Pattern, Discussed in Section 4.3. (b) Discovering Aha Moment, Discussed in Section 4.4.

Figure 3: Example Process Models (Focus on this ﬁgure is cluster but node information).

ICEIS 2023 - 25th International Conference on Enterprise Information Systems

168

Figure 5: Identify Retained Customers by Dotted Chart

(The row shows anonymous customer name).

include the visit events that occur too frequently and

the events not triggered by users. Furthermore, the

follow-up actions such as Edit, Disable, and Delete

are ﬁltered out. Since we are capturing the primary

interest, for instance, Create, which is the prerequisite

of the follow-up actions, can represent the interest in

the models.

Figure 3b shows the retained companies using

quality assurance in activation phase. The process

highlighted with the orange circle shows the account’s

interest in the associated feature conﬁguration; the

process with a yellow circle indicates the nodes with

the interest of team management.

The models successfully show the prerequisites

of adopting the features, including feature conﬁgura-

tions. Apart from feature settings, retained customers

of quality assurance actively modify the qualify stan-

dard and criterion; missions model shows strong in-

terest in monitoring performance since the function

is meant to provide goals for agents to improve. The

models for performance coaching contain a mixed be-

havior pattern, positioning the function as an assisted

solution among the other two features.

Abstracting the aha moment from the models

involves graphical analysis and domain knowledge.

The understanding of each event log and the relation-

ship with the primary pillar functions plays an essen-

tial role.

5 EVALUATION

During the experiments, we continuously involve do-

main knowledge and stakeholders’ feedback on im-

proving the modeling. The small model adjustments

rely on authors’ work experience and domain knowl-

edge in K; the iterations in larger scale, involving the

change of dataset request feedback from sales, cus-

tomer success, and analyst team. The feedbacks can

be viewed as the evaluation of the results, which is the

last step, Step 5.

Half of the stakeholders with business and strat-

egy mindsets involves in the evaluation or consult-

ing. Employees with the functions of marketing,

sales, customer support, product planning, and anal-

ysis are selected because they have the potential to

get in touch with customers. In the whole project, the

colleague from the customer support team in the case

company is highly involved due to their deep under-

standing of customer behavior.

The results reﬂect the primary paths which are rea-

sonable for stakeholders. The behavior patterns meet

the company’s understanding of customers. The iden-

tiﬁed frequent behaviors, creating teams, even met the

product development strategy at the same time, veri-

fying the company’s observation of customers. How-

ever, the project failed to discuss the interaction be-

tween pillars, which was the current focus of the com-

pany at that time. The gap might due to the fact

that the business strategy changes rapidly during the

project conducted.

The representation is another factor that many re-

searchers ignored. The process map from fuzzy miner

is intuitive but not simple enough, due to the fact that

the algorithm does not consider frequency to simplify

the model. However, heuristic miner does not guar-

antee an understandable model. While applying all-

tasks-connected heuristic, the models become more

explainable, verifying the statement that many depen-

dency relations can be hidden without records (Wei-

jters and Ribeiro, 2011).

It is also important to know the stakeholders’ re-

sponsibility. It is natural that people only focus on

what they are responsible for and care of. For in-

stance, in this case, the opinions of stakeholders who

focus on customer relationships are more valid than

Investigating Aha Moment Through Process Mining

169

the reviews from the sales team.

6 DISCUSSION

This study applies process mining to discover cus-

tomer behavior on retention and activation, which is

a brand-new ﬁeld that no previous researchers have

done. The involvement of customer funnel (McClure,

2022) provides a framework to analyze customers, es-

pecially for software industries. Companies can eas-

ily adapt the framework to their customer journey to

discover more processes.

Understanding how customers experience the

value of the product is the key to growth. Discov-

ering the aha moment provides companies with a

new approach to deﬁning a bottom-up and evidence-

based value proposition, which is more accurate than

a top-down approach. During the discovery phase,

decision-makers in the business are able to develop

better business strategies and product roadmaps. This

insights can help develop a better onboarding ﬂow, re-

ducing the churn rate at the beginning.

The integration of human-computer interaction to

process mining is also another newly discovered ﬁeld.

Adopting the concept of domain object to classify the

event logs is a practical method to clean the data be-

cause it reduces 62% of the nodes in this research.

The concept transforms the event logs into actual in-

centives. This approach is especially suitable for

cleaning weblog and UI logs because these logs are

usually noisy and dynamically changing.

Although structuring the research by-product so-

lutions is a practical framework, it failed to discover

the interaction between each pillar since, in general,

the users act on the same platform. The result shows

that one of the pillars is the supportive function of the

other two, and the retained customers of this pillar

have more proportion of old installations, which is out

of scope in this research. It would be interesting to see

the interaction between these pillars.

7 CONCLUSION

During the business understanding, we structure the

research into pillars and role type, with 6 scenarios.

Meanwhile, time is also taken into account as it plays

an important role on identifying repeated patterns.

Inspired by the domain object concept, the domain

object in terms of pillar, and associated milestone ac-

tions by role type are recognized. Furthermore, the

same concept is used to reduce the event nodes and

the data size by diminishing the interference of dy-

namically changed user interfaces.

After data cleaning, we address RQ1 to cluster re-

tained accounts. Customers with repeated milestone

actions from each role type over three weeks are iden-

tiﬁed as retained accounts in terms of pillar. Sce-

narios with pre-recognized milestone actions are veri-

ﬁed; scenarios missing milestone actions are modeled

by the domain object to capture the actions from both

parties.

We further slice the event sequences into activa-

tion and retention from retention clusters to address

RQ2. The logs before the accounts start retention

are added as an input to build the models. Based on

the graphical analysis and the involvement of domain

knowledge, a set of actions and the motivations be-

hind them are recognized.

This research framework can be applied to B2B

SaaS industry to analyze their customers. The study

can also apply to a bigger dataset because the event

classiﬁcation can minimize the number of nodes.

However, the data cleaning itself requires a lot of

manual work, which should be re-consider in the fu-

ture for scalability.

Some of the limitations should be noted. This

project stopped at the discovery phase but failed to ap-

ply the ﬁndings to real business contexts. The evalua-

tion only relies on domain knowledge and stakehold-

ers’ eye. The limitation is due to the time limitation

and project scale. It would be interesting to see the

following inﬂuence on the product and the possible

process enhancement.

The research only focuses on the speciﬁc time pe-

riod and customer phases. We do not discuss the en-

tire customer lifecycle, from acquisition to revenue.

We only focus on the newly installed customers and

do not discuss the longer customer relationship. It

could be interesting to discover the aha moment not

only from the retention but also from the revenue

phase. Meanwhile, applying process mining to other

customer phases remains a gap in the research ﬁeld.

REFERENCES

Aha! (2022). Aha! company history — aha! software.

Online, accessed September 14, 2022.

Assal, S. (2022). Digital marketing in business: A system-

atic review.

Balboni, K. (2022). The aha moment guide: How to ﬁnd,

optimize, and design for your product. Online, ac-

cessed September 14, 2022.

Beaudouin-Lafon, M. (2000). Instrumental interaction: An

interaction model for designing post-wimp user inter-

faces. In Proceedings of the SIGCHI Conference on

ICEIS 2023 - 25th International Conference on Enterprise Information Systems

170

Human Factors in Computing Systems, CHI ’00, page

446–453, New York, NY, USA. Association for Com-

puting Machinery.

Bernard, G. and Andritsos, P. (2017). A process mining

based model for customer journey mapping. vol-

ume 1848, pages 49–56. Clarify CJM components

and convert to XES format. This is a proposed modle

withou empirical study. Further clustering algorithm

and the prediction of the next touch point is required

in this ﬁeld.

Carpenter, W. (2019). The aha! moment: The science be-

hind creative insights. In Toward Super-Creativity-

Improving Creativity in Humans, Machines, and

Human-Machine Collaborations. IntechOpen.

Cerone, A. (2015). Process mining as a modelling tool:

Beyond the domain of business process management.

pages 139–144. Springer Berlin Heidelberg.

Chan, N. N., Vo, D. L. N., Pham-Nguyen, C., Dinh, T. L.,

Dam, N. A. K., Thi, T. T. P., and Thi, M. H. V. (2021).

Design and deployment of a customer journey man-

agement system: the cjma approach. pages 8–16.

ACM.

Corallo, A., Lazoi, M., and Striani, F. (2020). Process min-

ing and industrial applications: A systematic litera-

ture review. Knowledge and Process Management,

27:225–233.

Else, Natalia, S., Angelique, B.-K., van der Aalst Wil M

P van Eck Maikel L., and Markslag (2019). Data-

driven usability test scenario creation. pages 88–108.

Springer International Publishing.

Guzzo, A., Rullo, A., and Vocaturo, E. (2022). Process min-

ing applications in the healthcare domain: A compre-

hensive review. WIREs Data Mining and Knowledge

Discovery, 12.

unther, C. W. and van der Aalst, W. M. P. (2007). Fuzzy

mining – adaptive process simpliﬁcation based on

multi-perspective metrics.

Hochstein, B., Rangarajan, D., Mehta, N., and Kocher, D.

(2020). An industry/academic perspective on cus-

tomer success management.

Husin, H. S. and Ismail, S. (2021). Process mining approach

to analyze user navigation behavior of a news website.

pages 7–12. ACM.

Lacka, E. and Chong, A. (2016). Usability perspective on

social media sites’ adoption in the b2b context. Indus-

trial Marketing Management, 54:80–91.

Lin, Z. and Chaomin, G. (2021). Research on acquisi-

tion on authors sources of sci-tech journals based on

aarrr model. In Proceedings of the 2021 International

Conference on Control and Intelligent Robotics, pages

193–196.

Liu, C., van Dongen, B., Assy, N., and van der Aalst, W.

M. P. (2018). Component interface identiﬁcation and

behavioral model discovery from software execution

data. pages 97–107. ACM.

McClure, D. (2022). Startup metrics for pirates. Retrieved

from website ”Master of 500 Hats.”.

Pandey, N., Nayal, P., and Rathore, A. S. (2020). Digital

marketing for b2b organizations: structured literature

review and future research directions. Journal of Busi-

ness & Industrial Marketing.

Parker, G. G., Van Alstyne, M. W., and Choudary, S. P.

(2016). Platform revolution: How networked markets

are transforming the economy and how to make them

work for you. WW Norton & Company.

Peng, S. and Cheng, Q. (2009). Research on data prepro-

cessing process in the web log mining. pages 942–

945. IEEE.

Rix, C., Horst, C., Autenrieth, P., Paproth, Y., Frank, J., and

Gudergan, G. (2020). Typology of digital platforms

in the mechanical engineering. In 2020 IEEE Inter-

national Conference on Engineering, Technology and

Innovation (ICE/ITMC), pages 1–9. IEEE.

Shaltoni, A. M. (2017). From websites to social media: ex-

ploring the adoption of internet marketing in emerging

industrial markets. Journal of Business & Industrial

Marketing.

Shneiderman, B. (1982). The future of interactive systems

and the emergence of direct manipulation. Behaviour

& Information Technology, 1:237–256.

Shneiderman, B. (1983). Direct manipulation: A step be-

yond programming languages. Computer, 16:57–69.

Situmorang, S. H. and Harmawan, W. A. (2022). Customer

value, brand experience, and user experience on cus-

tomer satisfaction and loyalty in digital streaming ser-

vices. In Handbook of Research on Smart Manage-

ment for Digital Transformation, pages 233–255. IGI

Global.

Stancil, B. (2015). Facebook’s ”aha” moment was simpler

than you think. Online, accessed September 14, 2022.

Taub, M., Banzon, A. M., Zhang, T., and Chen, Z. (2022).

Tracking changes in students’ online self-regulated

learning behaviors and achievement goals using trace

clustering and process mining. Frontiers in Psychol-

ogy, 13.

Teixeira, A., Dominique-Ferreira, S., Martins, N., and

Sylla, C. (2023). Developing an integrated commu-

nication plan in the digital age. In Perspectives on De-

sign and Digital Communication III, pages 313–325.

Springer.

Terragni, A. and Hassani, M. (2018). Analyzing customer

journey with process mining: From discovery to rec-

ommendations. pages 224–229. IEEE.

Terragni, A. and Hassani, M. (2019). Optimizing cus-

tomer journey using process mining and sequence-

aware recommendation. pages 57–65. ACM. Filter

out the consequent events are necessary.

Theis, J. and Darabi, H. (2019). Behavioral petri net min-

ing and automated analysis for human-computer inter-

action recommendations in multi-application environ-

ments. Proceedings of the ACM on Human-Computer

Interaction, 3:1–16.

van der Aalst, W. (2016). Process mining : data science in

action. Springer Berlin Heidelberg.

Venkatesh, R., Mathew, L., and Singhal, T. K. (2019). Im-

peratives of business models and digital transforma-

tion for digital services providers. International Jour-

nal of Business Data Communications and Network-

ing (IJBDCN), 15(1):105–124.

Investigating Aha Moment Through Process Mining

171

Vinod, David, C., Nicolas, K. R. P., and Muthusamy (2013).

Business process mining from e-commerce web logs.

pages 65–80. Springer Berlin Heidelberg. Structure

user behaviour model intend to provide insights.

Weijters, A. and Ribeiro, J. (2011). Flexible heuristics

miner (fhm). pages 310–317. IEEE.

Weijters, A., van Der Aalst, W. M. P., and Medeiros, A.

K. A. D. (2006). Process mining with the heuristics

miner-algorithm. Technische Universiteit Eindhoven,

Tech. Rep. WP, 166:1–34.

Wirth, R. and Hipp, J. (2000). Crisp-dm: Towards a stan-

dard process model for data mining. volume 1, pages

29–39.

Zaim, H., Ramdani, M., and Haddi, A. (2018). Fuzzy-based

mining framework of browsing behavior to enhance

e-commerce website performance: Case study from

kelkoo.com. pages 1–6. ACM.

Zerbino, P., Stefanini, A., and Aloini, D. (2021). Process

science in action: A literature review on process min-

ing in business management. Technological Forecast-

ing and Social Change, 172:121021.

Zhang, L. (2021). Research on multi-user growth strategy

of pinduoduo based on aarrr model. In 7th Annual

International Conference on Social Science and Con-

temporary Humanity Development (SSCHD 2021),

pages 271–276. Atlantis Press.

ICEIS 2023 - 25th International Conference on Enterprise Information Systems

172