Conference PaperPDF Available

Customers Like It Hot and Fast – Incorporating Customer Effects into the Meal Delivery Process

Authors:

Abstract

Delivering meal orders as fast as possible and the meal itself as hot as possible are the most important factors in the meal delivery process as they drive customer satisfaction. High customer satisfaction leads to loyal customers, implying a higher rate of recurring orders, in return. Existing approaches tackle the meal delivery process by taking a short-term perspective on a single optimization criterion (e.g. minimizing delivery costs). Still missing is an alternative perspective that also incorporates the long-term value contribution of individual customers. By neglecting this customer-centric perspective, frequent out-of-town located ordering customers might be disadvantaged as they are repeatedly served at the end of the route. To close this research gap, we propose a decision model (C2RG) that incorporates a long-term customer-centric view. Depending on different short-and long-term preferences, the model can be appropriately customized. We observe a significant increase in a long-term factor, such as customer fairness by only slightly reducing short-term route performance. We instantiated a software prototype of the C2RG and evaluated it with real-world data of a local platform-to-consumer delivery service located in Germany. The results show the importance of considering a customer-centric long-term perspective in the meal delivery process.
van Dun et al. /Customer-Centric Meal Delivery Processes
Twenty-Eigth European Conference on Information Systems (ECIS2020), Marrakesh, Morocco. 1
CUSTOMERS LIKE IT HOT AND FAST INCORPORATING
CUSTOMER EFFECTS INTO THE MEAL
DELIVERY PROCESS
Research paper
van Dun, Christopher, FIM Research Center, University of Bayreuth; Project Group Business
& Information Systems Engineering of the Fraunhofer FIT, Bayreuth, Germany
christopher.vandun@fim-rc.de
Fehrer, Tobias, FIM Research Center, University of Augsburg, Augsburg, Germany
tobias.fehrer@fim-rc.de
Kratsch, Wolfgang, FIM Research Center, University of Bayreuth; Project Group Business &
Information Systems Engineering of the Fraunhofer FIT, Bayreuth, Germany
wolfgang.kratsch@fim-rc.de
Wolf, Nicholas, FIM Research Center, University of Augsburg, Augsburg, Germany
nicholas.wolf@fim-rc.de
Abstract
Delivering meal orders as fast as possible and the meal itself as hot as possible are the most important
factors in the meal delivery process as they drive customer satisfaction. High customer satisfaction leads
to loyal customers, implying a higher rate of recurring orders, in return. Existing approaches tackle the
meal delivery process by taking a short-term perspective on a single optimization criterion (e.g. mini-
mizing delivery costs). Still missing is an alternative perspective that also incorporates the long-term
value contribution of individual customers. By neglecting this customer-centric perspective, frequent
out-of-town located ordering customers might be disadvantaged as they are repeatedly served at the
end of the route. To close this research gap, we propose a decision model (C2RG) that incorporates a
long-term customer-centric view. Depending on different short- and long-term preferences, the model
can be appropriately customized. We observe a significant increase in a long-term factor, such as cus-
tomer fairness by only slightly reducing short-term route performance. We instantiated a software pro-
totype of the C2RG and evaluated it with real-world data of a local platform-to-consumer delivery ser-
vice located in Germany. The results show the importance of considering a customer-centric long-term
perspective in the meal delivery process.
Keywords: Vehicle Routing Problem, Meal Delivery Routing Process, Customer-Centricity, Decision
Model, Routing Optimization.
1 Introduction
Online food delivery has become a billion-dollar business since Pizza Hut delivered its first online or-
dered pizza in 1994 (Schrage, 1994; Steiner, 1994; Statista, 2019). Due to the increasing ubiquity of
smartphones, the sector is growing at a fast pace worldwide, expected to increase by 11.4% until 2023
(Statista, 2019). In the dominating platform-to-consumer delivery” business model, logistics is oper-
ated by a service provider, whereas restaurants act as third-party suppliers. Organizations see themselves
challenged not only to withstand pricing pressure but also to perfect service quality to set themselves
apart from competitors in a consolidating market (Deliveryhero, 2018). Operational excellence, as well
as a focus on customer relationships, are necessary to handle future service sector growth (Statista, 2019;
Vakulenko et al., 2019).
van Dun et al. /Customer-Centric Meal Delivery Processes
Twenty-Eigth European Conference on Information Systems (ECIS2020), Marrakesh, Morocco. 2
This is not a trivial task as the meal delivery industry (as all last-mile delivery services) faces versatile
challenges: Operational challenges include the uncertainty of upcoming orders, the unpredictability of
meal preparation times, and the urgency to deliver fast when dealing with perishable goods. Environ-
mental challenges include continuing urbanization and its impact on city traffic (United Nations,
Department of Economic and Social Affairs, Population Division, 2019), internal migration from inner
cities to suburban or exurban areas (Sander, 2017; Henger and Oberst, 2019) and the high impact of
service quality in last-mile delivery on the overall customer experience (Vakulenko et al., 2019).
Consequently, meal delivery businesses are very complex to operate. Current academic literature intro-
duces different approaches to execute meal delivery processes (Xiang et al., 2008; Ahmadi-Javid et al.,
2018; Alan Eera, 2017). These processes are characterized by delivering meals from a single depot or a
sequence of depots to diversely located customers with a certain number of couriers. A commonly shared
goal of suggested solution algorithms is to minimize delivery costs and other efforts. The problem of
allocating couriers to routes can be solved quite well from a logistic point of view (Ioannou et al., 2001).
However, by minimizing costs, most of these algorithms focus on a short-term, efficiency-driven per-
spective. This may lead to repetitive patterns in decision-making that cause unwanted effects on cus-
tomer satisfaction for fragments of customers with unfavorable characteristics, e.g. comparatively long
distance to the restaurant or traffic density on the route. Eventually, repeated unfortunate decision-mak-
ing and decreasing customer satisfaction may cause disaffection and migration of these potentially val-
uable customers in the long run (Galbraith, 2005; Vakulenko et al., 2019). One might intuitively imagine
the unwanted loss of customer satisfaction of customers living in one of the rapidly growing, wealthy,
and therefore promising exurban areas (Sander, 2017), always receiving their order later than preferred
orders in the inner city. Extensive research into route scheduling algorithms of currently popular plat-
forms such as Delivery Hero, Foodora or Deliveroo and last-mile delivery software such as onfleet
1
,
vromo
2
, getswift
3
or route4me
4
does not suggest particular awareness towards this structural issue. Or-
ganizations that are not aware of their maladjusted process design may unintentionally lose customers
to competitors. Therefore, it is beneficial to incorporate a long-term perspective and a customer-centric
view when making decisions about the proceeding. Such an approach is necessary to establish a sustain-
able competitive advantage (van den Hemel and Rademakers, 2016).
In consideration of this, we formulate the following research question: How can the meal delivery rout-
ing process be enhanced by incorporating long-term customer-centricity?
In previous approaches, the share of customers located conveniently along dynamically calculated de-
livery routes is systematically preferred over others and receive their goods before all others. To estab-
lish an enhanced approach, we look at “pickup and delivery problems” (PDPs), where a courier picks
up a perishable commodity at one out of numerous depots and delivers it to a customer (Berbeglia et al.,
2010). To address our research question, we then develop the Customer-Centric Route Generation
(C2RG) model, integrating a long-term perspective in route bundling. We suggest a decision-making
algorithm for short-term route assignments considering an additional long-term customer-centric view
on the delivery process. Driven by a real-world evaluation case which aims at equal treatment of cus-
tomers over time, we implement the C2RG with countervailing delivery waiting times in mind, focusing
on the Meal Delivery Routing Problem (MDRP) (Reyes et al., 2018). Following real-world case require-
ments, this implementation covers pickup and delivery from a single depot to several customers. The
presented model overcomes systematic location-based discrimination or preference of certain customers
by considering their historic perceptions. In doing so, we detect disadvantaged customers and prioritize
distinct orders during optimization time to avoid poor customer perception.
1
https://support.onfleet.com/hc/en-us/articles/360023910351-Route-Optimization-Operating (accessed on March 23, 2020)
2
https://help.vromo.io/knowledge/how-to-auto-dispatch-jobs (accessed on March 23, 2020)
3
https://getswift.zendesk.com/hc/en-us/articles/360034981334-Route-Optimization (accessed on March 23, 2020)
4
https://route4me.com/industries/food-delivery (accessed on March 23, 2020)
van Dun et al. /Customer-Centric Meal Delivery Processes
Twenty-Eigth European Conference on Information Systems (ECIS2020), Marrakesh, Morocco. 3
Designing the C2RG as a valid design artifact (March and Smith, 1995), we adapt the design science
research (DSR) paradigm proposed by Gregor and Hevner (2013). Following the DSR reference process
(Peffers et al., 2007), we identify the research gap and motivate our research in this section. In Section
2, we derive design objectives to solve the problem using justificatory knowledge. In Section 3, we
present the design specification of our C2RG. In Section 4, we report our evaluation results, while we
conclude our work in Section 5 by pointing to limitations and further research.
2 Theoretical Background & Design Objectives
2.1 Customer centricity
Not only in research but also in practice, a shift from a product-centric view towards focusing the cus-
tomer as the central starting point for all further corporate activities is increasingly gaining popularity
after being introduced to marketing literature decades ago (Gartner, 2019; Sheth et al., 2000). Marketing
research highlights the positive impact of customer-centricity on an organization’s firm value and mar-
ket success, as well as on customer satisfaction and loyalty that can be achieved by applying a set of
transformational activities to a firm (Fornell et al., 1996; Khan and Fasih, 2014; Shah et al., 2006).
Whereas product-centricity aims at selling as many products as possible (Shah et al., 2006; Rust et al.,
2010), customer-centric organizations focus on serving the customer. This change of paradigms re-
quires, amongst others, processes and systems to provide the best service to customers during the whole
customer life cycle (Shah et al., 2006). An important activity here is to learn from customer behavior
(Jayachandran et al., 2005). The development of customer-centric information systems (CCIS) focusses
on configuring four major components, i.e. customer, process, technology, and product/service, to learn
from customer behavior and to satisfy their needs (Liang and Tanniru, 2006). The configuration includes
the capture of customer needs, an on-demand configuration of service processes, and the customization
of services (Liang and Tanniru, 2006).
In the context of e-commerce and last-mile delivery, the evaluation of customer experience is strongly
related to customer satisfaction as well as customer loyalty (Vakulenko et al., 2019; Oliver, 1999; Liang
and Tanniru, 2006). As such, an integrated view of all stakeholders in the last-mile delivery service
network, including the logistics service provider, build the customer experience. From a CCIS point of
view, the passive role that customers take in meal delivery leaves it up to “the system to capture their
implicit preferences and needs (Liang and Tanniru, 2006). One important decreasing factor of customer
satisfaction is the customer’s perception of being treated fairly. Balancing re-prioritization of disadvan-
taged but valuable customers has a positive effect on their average satisfaction while at the same time,
a negative effect on the average satisfaction of other customers is not measurable (Homburg et al., 2008).
Hence, taking control of the service delivery framework to be able to shape the end-to-end service ex-
perience allows for a balancing re-prioritization of customers and thus to improve overall customer sat-
isfaction. Consequently, we define (DO.1) as follows:
(DO.1) The artifact must enable an operational process to incorporate customer prioritization.
Depending on the application case, specific measures are to be considered when aiming for an increase
in customer satisfaction. This follows general demands towards processes embedded in CCIS, that are
required to be easily configurable to incorporate customer demands (Liang and Tanniru, 2006). Con-
cerning the diversity of definitions of satisfaction-increasing measures we require:
(DO.2) To allow for different aspects of customer-centric service design, the model’s input parameters
must be parameterizable depending on the current customer perspective.
2.2 Vehicle Routing Problems
A crucial problem for organizations in last-mile delivery is the efficient delivery process from geograph-
ically dispersed pickup locations (restaurants) and customers, subject to limitations of varying capacities
van Dun et al. /Customer-Centric Meal Delivery Processes
Twenty-Eigth European Conference on Information Systems (ECIS2020), Marrakesh, Morocco. 4
(e.g. the number of available couriers, the uncertainty of upcoming orders). Approaches for efficient
solutions concerning work assignment and route scheduling can be found in OR literature. In OR, the
challenge of Vehicle Routing Problems (VRP) can be solved using linear optimization. Characteristics
and assumptions of VRP vary widely in OR literature (Braekers et al., 2016) and can be illustrated in a
hierarchy as depicted in Figure 1. Platform-to-consumer delivery services can be modeled as pickup and
delivery problems (PDP), a subclass of general VRP. PDP subdivide into three problem groups (Ber-
beglia et al., 2010): (1) many-to-many problems, in which any node represents a pickup or a destination
location for commodities, (2) one-to-many-to-one problems, where outbound commodities are available
at a single depot and need to be shipped to many customers which in return hand over an inbound
commodity sent back to the depot and lastly (3) one-to-one problems, covering all logistics processes
where a specific pickup and destination location is predefined for each commodity to be delivered.
Figure 1. A brief overview of Vehicle Routing Problems
Routing problems can furthermore be classified into static and dynamic problem groups (Berbeglia et
al., 2010). In a static environment, the input data are known before the routes are built whereas in a
dynamic environment input data (e.g. upcoming orders) are revealed or modified during execution time
of the routing process, e.g. in the case of real-time decision systems. Platform-to-consumer meal deliv-
ery processes can be classified as dynamic one-to-one optimization problems, as customer orders are
not known ex-ante and a single customer orders his meal at a restaurant that corresponds to the specific
depot. An additional level of segregation distinguishes between three types of dynamic one-to-one PDP
(Berbeglia et al., 2010): The first problem group is the dynamic vehicle routing problem with pickups
and deliveries (Dynamic VRPPD), in which vehicles can fulfill more than one delivery of commodities,
also referred to as bundling. The second group, dynamic stacker crane problems (Dynamic SCP), defines
problems where vehicles can serve only one request at a time (i.e. the commodity exactly fits vehicle
capacity). In the third problem group, known as dynamic dial-a-ride problems (Dynamic DARP),
passengers are transported instead of commodities. Besides the fact, that in food delivery processes we
still transport commodities instead of passengers, our problem set has many similarities with the Dy-
namic DARP problem as the ordering customer is waiting immediately for his order, time constraints
concerning the freshness of the food as well as an existing maximum waiting time of the customer. In
academic literature, the problem tailored to the conditions of food logistic processes can be found under
the term Meal Delivery Routing Problem (MDRP) (Reyes et al., 2018; Yildiz and Savelsbergh, 2019).
Considering the various approaches to take on the meal-to-vehicle assignment and routing from an op-
timizing perspective, we define the requirement to efficiently deliver orders as follows:
(DO.3) The artifact must schedule and assign upcoming orders in a way that incorporates an efficiency-
driven perspective on the meal delivery process.
2.3 Customer-Centric Vehicle Routing Problems
Our research aims at combining and balancing a short-term efficiency-driven perspective together with
a long-term customer-centric view as discussed in the previous subsections. Related work has addressed
van Dun et al. /Customer-Centric Meal Delivery Processes
Twenty-Eigth European Conference on Information Systems (ECIS2020), Marrakesh, Morocco. 5
and demonstrated the relevance of an integrated view of CRM and problems in OR (Buhl et al., 2011).
Augmenting a traditional process-oriented route-optimization approach by incorporating a customer-
centric view on the process allows for being able to take over control over the customer service experi-
ence. Depending on the state of the organization, different criteria fulfill the requirement of customer-
centricity. Furthermore, the weight put on embracing the long-term perspective may vary over time.
Thus, we specify a design objective integrating both perspectives:
(DO.4) The artifact must be able to cater to different companies' short and long-term preferences.
3 Model Design
3.1 Conceptual architecture
The C2RG intends to assist platform-to-consumer organizations by grouping sets of orders into bun-
dles and recommending an optimal sequence of delivery to balance customers’ waiting time. On a high
level of abstraction, the C2RG offers a solution to allocating a set of orders to routes in a way that
introduces a customer-centric perspective to the route-generation process. This enhances customer sat-
isfaction via shortening meal delivery times for disadvantaged customers without neglecting delivery
routes’ efficiency, especially during peak times. Proposing a solving algorithm, we build on previous
work by Ioannou et al. (2001), and especially Reyes et al. (2018), reflecting the real-world requirements
where a courier picks up a bundle at a single restaurant.
Let   be a set of restaurants, where each restaurant has a location and let be a set
of customers. Each customer    has a performance indicator aggregating his previous interaction
with the organization as well as , his overall priority depending on the customer segment he is assigned
to. The set contains all revealed orders. After an order is placed, meal preparation is initiated, and a
ready time is determined (e.g. via restaurant feedback or estimation). Each order   belongs to a
restaurant  , has a placement time , a time at which the order can be picked up at the restaurant
by a courier (i.e., the ready time), and a drop-off location . The attribute refers to the customer
placing the order. After drop-off, each order gets assigned its delivery duration (i.e., the time from
pickup to drop-off), for later reference. An order is thereby described as the tuple  
. Furthermore, let be a set of couriers with each courier having
an initial position, , at which the courier will start his shift at the beginning time , and an off-time
  when his shift ends (  ). While information about , and is known in advance,
orders    are revealed sequentially at their placement time , yielding a dynamic problem.
The MDRP builds upon four assumptions. (1) A predominant assumption is that instead of delivering
each order individually, several orders with different drop-off locations but assigned to one restaurant
may be combined to bundles . By utilizing bundles, we can vastly improve the total
delivery time. Although a target bundle size is considered, we do not limit the theoretically possible
bundle size via an additional constraint. The sequence of orders within a bundle vector determines a
bundle’s delivery route. The ready time of a bundle is the latest ready time of its orders. (2) Following
Reyes et al. (2018), we assume travel times between two locations to be invariant over time for every
courier (i.e. traffic situation will never justify postponement of an order). (3) Invariant restaurant service
times model the time required to pick up a bundle after arriving at pickup location. Besides, half of
the invariable customer service time determines the time required to drop off an order at and
another half of is required until the route can be continued after drop-off or the courier can start a
new assignment after finishing a bundle. (4) Lastly, couriers    earn a fixed salary independent of
their number of deliveries. In return, couriers follow any instructions given by the algorithm and do not
make decisions on their own. They can be assigned to their first order after and accept new assign-
ments before . If they do not immediately receive a new assignment after finishing a bundle, they
remain at their position. In our evaluation (Section 4), we show that these assumptions comply with real-
world use cases and discuss them with a domain expert.
van Dun et al. /Customer-Centric Meal Delivery Processes
Twenty-Eigth European Conference on Information Systems (ECIS2020), Marrakesh, Morocco. 6
3.2 Proposed Artifact
Regarding the structural assumptions, commonly applied performance metrics in existing solution algo-
rithms focus on short-term efficiency, e.g. number of orders delivered or click-to-door time (the differ-
ence between placement time and drop-off time) or cost-per-order measures. Additionally, C2RG con-
siders a long-term perspective. Section 3.1 depicts the incorporation of customer information into
C2RG. This allows for extended performance metrics, which enable a broadened view on the delivery
process as a key enabler for customer satisfaction. The goal of the C2RG is to establish a customer-
centric view on the delivery process since the overall goal of each organization is to enhance its long-
term firm value.
Figure 2. Schematic comparison of the Basic MDRP Solution to our proposed C2RG Solution
Considering a scenario, we schematically introduce the functionality of the C2RG. We do so by com-
paring a basic bundle generation to the C2RG approach and highlight differences. Figure 2 illustrates
the decision-making. Five customers , place an order at restaurant . The customers are
known to our algorithm and have a track record of previous orders. We can, therefore, retrieve a per-
ception of (i.e. for illustrative reasons we rank-order the customers with decreasing importance ac-
cording to the treatment they have experienced in the past, in terms of waiting times). This indicates that
and have enjoyed the best customer experience to date (e.g. living close to their favorite restaurant
may repeatedly result in low waiting times). Customers has made a medium experience,  
denoting also the average in this setting. Customers and have the worst experiences  
, as these customers live further away from the restaurant and are, therefore likely to experience higher
waiting times. As an efficiency-driven process, the MDRP optimization algorithm does not take histor-
ical customers’ experiences into account, is ignored and the most efficient route is generated for this
set of orders. Examining this solution, two bundles are created for the available couriers. From a cus-
tomer-centric perspective, the execution of this strategy would impair the experiences of and while
other customers are either unaffected or excessively benefit from the strategy. Although this is the most
efficient route, it may cause , , or both to leave the customer base. In contrast, the C2RG considers
long-term effects in his decision-making. The algorithm detects urgency to prioritize and aiming
for generation of bundles that enhance the perception of customer and . Hence the orders of
and will be preponed in this solution strategy. In our algorithm, we also allow for a complete reor-
ganization of bundles. We demonstrate this reorganization capability switching the order of from
bundle 1 to bundle 2 and additionally postponed the order to the last position on the route decreasing
the perception of as this customer mostly received his order first.
Comprising this scenario, we introduce the objective function as depicted in Eq. 1 aggregating the -
weighted short- and long-term perspective scores expressed by 
 and 
 in the objective function:
van Dun et al. /Customer-Centric Meal Delivery Processes
Twenty-Eigth European Conference on Information Systems (ECIS2020), Marrakesh, Morocco. 7
   
   
 }
(1)
Both, 
 and 
 incorporate different views on a route-to-bundle combination and contribute to the
valuation of an order ’s assignment to a bundle at position    on the courier’s route. The
first component 
 expresses the impact on route efficiency: Compared to the short-term optimal
route, any other constellation of orders within this bundle will, at least, have no impact on route effi-
ciency but is expected to increase the total time required to deliver the orders due to indirect routes. The
same applies when considering a multitude of bundles and the option to re-bundle all orders. Therefore,

 is defined to be the amount of route efficiency loss. The latter component 
 quantifies the
improvement of a customer-centric performance metric. The specific implementation varies, depending
on the organization’s customer relationship management and which target variable should be optimized
(e.g. prioritization of most valuable customers; equal treatment, equating location-based disadvantages).
The sum of both partial scores leads to , the overall score. The derived score of an order for each
constellation of route and position will find its optimum in a minimal value of 
 .
The customer-centric preference factor (C2P-factor),  , serves as a weighting factor that allows
each organization to moderate its strategic orientation, i.e. by focusing on short-term oriented ( )
or long-term oriented steering ( ). The extreme cases of lead to organizations either fully
short-term oriented ( ) or fully long-term aligned ( ).
At each time of optimization , the C2RG selects only those combinations of orders to a bundle, that
results in a minimal long- and short-term integrated route score. We use information from an interval
possibly different from the “assignment” horizon (i.e., the window of orders to include in routes) to
determine how intensely the long-term perspective should be prioritized. At or before busy periods, like
dinner time, when many orders become ready within a short time, the C2RG enforces route efficiency,
while in relatively calm periods, it allows for a bigger weighting of customer-centricity. Therefore, we
define to auto-adjust over time according to the processing workload , limited by parameterizable
fixed boundaries  and  . The boundaries express strategic goals for customer-centricity. To
induce such a target measure dynamically, we consider the direct relation
 
to express the load factor . A parametric definition of
the workload at optimization time is:
  
     
        
where 
  is the ready time of a bundle and is the time when a courier becomes available
for a new assignment. It is possible, that no courier is available before   , in which case is set to
. Specific values for  are set through a tuning procedure but cannot exceed the foreseeable
horizon of known orders. To balance within the bounds of  and , we define  as:
       
We suggest the concept of a combined short- and long-term perspective as it allows for establishing an
abstract model independent of the specific input variables. Depending on the information available on
customer experience, different approaches can be used to either incorporate a proxy for customer satis-
faction or customer satisfaction itself. In extension to performance measures introduced by Reyes et al.
(2018) (e.g. click-to-door time, ready-to-door time), we introduce additional performance metrics, ex-
pressing individual customer experiences and therefore serve as proxy values for customer satisfaction:
Customer average click-to-door time: the average of previous click-to-door times of a customer.
Customer average ready-to-door time: the average of previous ready-to-door times of a customer.
van Dun et al. /Customer-Centric Meal Delivery Processes
Twenty-Eigth European Conference on Information Systems (ECIS2020), Marrakesh, Morocco. 8
Customer average pickup-to-door time: the average time from the pickup location to drop-off loca-
tion per customer.
4 Demonstration & Evaluation
4.1 Evaluation Strategy
To evaluate the C2RG, we base our evaluation strategy on the evaluation framework for DSR by Son-
nenberg and vom Brocke (2012). We complete the justification of our research topic in the Introduction
and Theoretical Background. Furthermore, we derive design objectives from relevant literature in Sec-
tion 2. We construct our C2RG as a software prototype with respect to the previously defined design
objectives in Section 4.2. Testing the prototype with synthetic data provides us with a proof of concept
of our artifact. In Section 4.3, we finish the evaluation of the artifact by demonstrating its performance
using real-world data as input for our prototype and therefore providing indications of the artifact’s
usefulness and applicability in realistic settings.
4.2 Prototype Construction
To allow for application in realistic settings and to provide a proof of concept, we instantiated the C2RG
as a software prototype. To put C2RG fully into effect, we adopted the proposal for solving the MDRP
in a three-step “double horizon algorithm” as introduced by Reyes et al. (2018) and implemented our
enhancement into the first step of their algorithm. We can, therefore, validate the C2RG by benchmark-
ing the decision-making and results of our enhancement against the initial solution.
Delivery time and meal freshness are the two influencing factors of customer satisfaction in the meal
delivery domain (Liu and Florkowski, 2018) and can thus serve as a proxy for customer satisfaction. In
the evaluation case, the controllable part of the delivery service process is the time starting the bundle
pickup at the restaurant (pickup time) and the time at which the courier arrives at the customers door
(door time). We refer to this measure as the pickup-to-door time. In our prototypical instantiation, the
C2RG aims towards the fair treatment of all customers regardless of their ease of reachability along a
route. Defining and implementing this precise customer-centric measure, we confirm (DO.2) as the user
can decide on the manifestation of long-term effects. Considering this measure is expected to result in
decreasing variance within the resulting pickup-to-delivery times. After generating short-term optimal
bundles using parallel insertion and a remove-reinsert search (Reyes et al., 2018), we decrease short-
term route performance by preponing orders marked as critical. Doing so, we model the short-term
component of the C2RG 
 , and the long-term component 
 , combined with the total evaluation
score of a bundle-to-order constellation , as follows. The short-term component 
 comprises
the decrease of route efficiency between an optimal route and the constructed route by placing order
at position of bundle . The long-term component 
 includes the effect on customer satisfaction
(i.e. pickup-to-door time) due to the bundling of order on bundle regardless of their ease of reacha-
bility along a route. Since the aggregation of short-term order values 
 and thus the implicit effi-
ciency of the route is an important factor in the allocation of incoming orders, (DO.3) is fulfilled by our
proposed artifact. The long-term effect is calculated with the ratio of pickup-to-door time  and
the daily average pickup-to-door time  in relation to
, the customer’s previous experience.
The variable
 is calculated by all his historic pickup-to-door times compared to the average pickup-
to-door time of all customers. The C2P-factor can be fine-tuned to balance the importance of short-
and long-term components. In considering the C2P-factor, we see (DO.4) as achieved by design, since
the model can align to organizations diverse strategic short- and long-term orientation.
At each time point of optimization , the C2RG selects only those combinations of orders to a bundle
that results in the lowest long- and short-term integrated route score. To identify a global optimum for
van Dun et al. /Customer-Centric Meal Delivery Processes
Twenty-Eigth European Conference on Information Systems (ECIS2020), Marrakesh, Morocco. 9
the order-to-bundle allocation incorporating long-term customer effects, a complete enumeration is nec-
essary. Numerous possible mappings of orders to bundles of variable length with each order possibly
being placed at every but the restaurant's position of a bundle’s route results in an np-hard problem. For
practical feasibility, we, therefore, applied a greedy heuristic depicted in the pseudo-code below.
Input: , set of upcoming orders at restaurant ,
sensitivity threshold for long-term score,
, C2P-factor,
 set of previous customer order pickup-to-door durations
, the set of pre-determined bundles from restaurant with short-term-optimal routes
Output: , the re-organized set of bundles from restaurant to be assigned to couriers.
/* Integrate long-term perspective in pre-determined bundles */
Define  
     as the set of orders with poor historical customer experiences
where

;
For    do
Remove from its current bundle   ;
Find bundle    and insertion position to re-insert at a minimum score  rating where
the score for in bundle and position , is composed of    weighted “relative route
performance” 
  
 

and weighted “long-term delivery experience”,
defined as 
 


 with the current pickup-to-door time 
 and the
pickup-to-door time in the optimal route 
:     
    
;
Re-insert into the bundle at position ;
End
Integrating the C2RG into the rolling horizon algorithm, there are two more steps left to complete in
every iteration. In the second step, the previously generated bundles are assigned to couriers by solving
a linear optimization problem. The set of optimal assignments is then actually allocated to the different
couriers. If a courier can pick up the bundle within , he is committed to the bundle. If no courier can
pick up the bundle at the restaurant or the bundle will not be ready to pick up within , a courier receives
the assignment to travel to the restaurant and receive his full commitment in one of the next iterations.
Using the prototype requires data from a platform-to-consumer purchase and delivery process. Real-
world process data from a fleet management service can be loaded into the prototype. In evaluation
mode, the prototype can be enacted to simulate real-time decision-making and virtually solve the rolling
horizon algorithm every minutes and determine a suitable assignment of upcoming “ready” orders in
within the assignment horizon of length to the couriers on duty.
As there is no reference implementation of the initial MDRP rolling horizon algorithm available, we
decided to implement the prototype using a software stack based on the python programming language,
as it allows for good readability of code, which makes it easier to follow the three-step decision-making
process. Aiming towards realistic and comparable results of the simulation, we integrated the open-
source routing machine (OSRM) to calculate route durations ridden by cyclists (Luxen and Vetter,
2011).
To validate the prototype with the implemented C2RG, we conducted an analysis based on modified
labeled data. For the analysis, we make use of a set of sample input data and modify input parameters.
The subset of data is labeled as such that the top 20% of customers who are experiencing the highest
van Dun et al. /Customer-Centric Meal Delivery Processes
Twenty-Eigth European Conference on Information Systems (ECIS2020), Marrakesh, Morocco. 10
delivery times are marked segment “A”. The rest is marked segment “B”. By performing the optimiza-
tion, we observe, that both positive and negative effects occur as expected (see Table 1). The C2RG
reduces the average ready-to-door time of customer segment “A” by almost a minute (−3.72%). How-
ever, the negative effects occur in segment “B” as their average ready-to-door time increases by more
than half a minute (+4.41%). Besides that, by prioritizing individual orders, the average total route du-
ration slightly increases. Additionally, modifications targeting order prioritization have a two-fold effect
on process times. First, prioritizing an order usually leads to increasing route duration due to accepted
detours. Second, an increase in route time also causes a decrease in total delivery throughput as couriers
require more time before committing to a new bundle. The overall goal of realizing fairness in the rout-
ing process is achieved as both segments are aligned to the total mean of 15.2 minutes. This confirms
the functionality of the prototype and marks (DO.1) as achieved since it incorporates customer prioriti-
zation in the operational delivery process and produces favorable results.
Table 1. Comparison of MDRP versus C2RG algorithm.
4.3 Demonstration
To show that the C2RG and the software prototype are applicable in realistic settings, that required data
can be gathered, and that analyses can be conducted, we present a case that builds on pseudonymized
event data collected at a platform-to-consumer delivery service (SERVICE, name redacted for review)
operating in a German city with about 300,000 inhabitants. Within their meal delivery unit, SERVICE
offers a marketplace offering food from local restaurants which cyclist couriers then collect at the res-
taurant and deliver within the city area. Regarding data collection, SERVICE provided us with process
data and order information from the last two years containing about 30,000 orders after a data cleaning
step, placed by 7,000 customers a share of 55% of which are recurring customers. These recurring cus-
tomers account for 86% of the orders. SERVICE’s delivery process corresponds to the structural as-
sumptions of the MDRP: (1) management assigns shifts to couriers; (2) shifts have a defined start; (3) a
clerk assigns bundles to the couriers and (4) even though restaurant performance cannot be controlled,
the clerk requests the ready-time of prepared meals. All these assumptions have been validated for prac-
tical feasibility by the management of SERVICE.
Figure 3. Distributions of orders (left, a) and the average count of requested restaurants per
hour of the day (right, b)
Next, we describe the real-world data showing the applicability of the data as input for the C2RG. With
a noon shift and an evening shift, SERVICE currently operates a two-shift system. However, the noon
Segments
Share of orders (%)
MDRP avg. time (mins)
C2RG avg. time (mins)
Time (mins)
A
10.82%
0:24:12
0:23:18
0:00:54
B
89.18%
0:13:36
0:14:12
+0:00:36
Total
100.00%
0:14:48
0:15:12
+0:00:24
van Dun et al. /Customer-Centric Meal Delivery Processes
Twenty-Eigth European Conference on Information Systems (ECIS2020), Marrakesh, Morocco. 11
shift was just recently established, leading to a significantly higher number of orders during the evening
shift. By absolute numbers, there are on average 42 orders per day. About 93% of all orders within a
day are processed during the evening shift, marking the evening shift as the most relevant part of our
evaluation. Orders must be carried out by six workers on average per evening shift (compared to 1.2
workers per noon shift). During the evening shift, we spot a peak in incoming orders at 7 pm, causing
varying workload within the shift, when compared to the workload from 6 pm to 8 pm (Figure 3a).
Considering this, one might imagine that during peak hours efficient routes must be generated to handle
the number of orders with a limited number of couriers. Data also reveals that the orders are on average
placed to no more than seven restaurants. This allows for building multiple bundles at one restaurant
during calm periods, giving us more options to optimize bundles (see Figure 3b). Per shift, the order-to-
worker ratio is on average 5.2 with a standard deviation of 2.8. A ratio peak in 18.8 challenges the
robustness of the algorithm. Concluding, the described sample data and the case are suitable to test and
validate our proposed prototype with real-time data.
To generate delivery strategies, we use the software prototype and parameterize the application, defining
(1) a window of 15 minutes as the process planning horizon and selecting (2) a five-minute interval for
repeated optimization execution. The newly introduced C2P-Factor which serves as the long-term stra-
tegic orientation of an organization is set to a corridor between 0.6 and 0.9, indicating a strong focus on
customer-centricity with a threshold of 1.3 for selecting poorly-performing customer experiences. In
accordance with managerial practice at SERVICE, we utilize the averaged ready-to-door time as intro-
duced in Section 3.1 as the target measure for customer-centric alignment.
Figure 4. Short-term impact of route variation
Next, we execute a simulation run of the C2RG with the presented parameters and data and discuss the
results. Figure 4 shows an illustrative example of the result of the C2RG compared to the benchmark.
The example is described by three customers () who order a meal at the restaurant . The basic
MDRP solution of this scenario results in the most efficient route of sending a courier from the restaurant
to the customers in the order (1) , (2) , and (3) This route takes 28:36 minutes to complete.
However, considering the historic experience of , C2RG promises a different routing strategy that
significantly changes ’s treatment. Customer retrieves his order first, which reduces his waiting
time. This improvement of ’s experience is made at the expense of customers , who receive
their order later. Hence, customers will receive their delivery depending on their previous experience,
all of which leads to decreasing variance in meal-delivery times.
Table 2 describes C2RG’s impact on different customer segments. As a general trend, we see, that the
C2RG achieves a desired slight decrease in standard deviation. The decrease in variation within the
group of all customers is excelled by the decrease in variation within the filtered set of customers who
experience a direct impact by the algorithm. On average, customers undergo an average increase in their
orders delivery duration of 15 seconds. The set of customers experiencing an impact by the C2RG con-
tains only 59.1% of the orders. This quantity is influenced by either the context of an order (i.e. bundles
of size one do not offer the opportunity for optimization) or the threshold that selects orders to be re-
scheduled.
van Dun et al. /Customer-Centric Meal Delivery Processes
Twenty-Eigth European Conference on Information Systems (ECIS2020), Marrakesh, Morocco. 12
Segment
Share of Orders
Measure
MDRP
C2RG

All customers
100.0%
mean
0:15:09
0:15:24
+0:00:15
std.
0:03:24
0:03:19
−0:00:05
Recurring customers (more than
one order)
86.6%
mean
0:15:00
0:15:03
+0:00:03
std.
0:03:54
0:03:48
−0:00:06
Customers with direct impact (ex-
perienced a change in delivery)
59.1%
mean
0:15:54
0:15:58
+0:00:04
std.
0:04:21
0:04:10
−0:00:11
Table 2. Impact of the C2RG on overall ready-to-door times.
Table 3 looks at groups of customers separated by the cumulated amount of route adjustment that has
occurred to them. The customers within the 5% and 95% quantiles experienced heavy re-bundling. As
a general trend, customers’ mean delivery times across all quantiles shift away from their primal value
in the direction of the overall mean. Data also shows that some cases within the outermost quantiles
exceed their target and almost invert. This is to be read as a hypersensitivity of the algorithm leading to
exaggerated route reconfiguration.
Quantile
Cumulated Route Adjustments
Avg. Impact on Route
MDRP
C2RG
5%
−0:18:00
−0:05:20
0:14:53
0:27:45
10%
−0:11:12
−0:01:25
0:14:08
0:17:09
15%
+0:00:00
0:00:00
0:14:34
0:14:36
85%
+0:00:00
0:00:00
0:14:20
0:13:46
90%
+0:09:15
+0:01:06
0:15:59
0:13:47
95%
+0:18:38
+0:03:49
0:16:01
0:14:34
Table 3. Impact of the C2RG on overall ready-to-door times.
Concluding the practical evaluation, the prototype of the C2RG model tested with real-world data gen-
erated interpretable results. We state that the basic assumptions of our proposed artifact can be validated
with real-world data. Although, while the total variance within delivery times has decreased, heuristic
hypersensitivity has caused irritating effects on segments of customers. This is likely to be explained by
the number of orders that are placed by certain customers. If a customer initially experiences long ready-
to-door times during his first order, the heuristic is likely to consider the customers following order
critical and forces prioritization. This behavior causes bias in data where there are only a few extreme
customer experiences collected. Nevertheless, the expected impact on customer-centricity measured as
a fair treatment among delivery times is achieved within the prototype which underpins our evaluation
criteria of applicability in real-world settings. The experts of our local platform-to-consumer service
agreed that this addition to their routing algorithm can be useful for uncovering undesired effects in
customer prioritizations and mitigating these. However, as we have not tested the algorithm in different
circumstances and with different experts, this can only be considered as the first indication of usefulness.
Furthermore, to prove the reliability of our algorithm, the C2RG should be tested by applying it to more
real-world cases to discover whether it produces satisfactory results in different real-world settings. Fu-
ture work should follow up on evaluating and fine-tuning the C2RG.
5 Concluding Remarks and Limitations
In this study, we examined the meal delivery process in the emerging last-mile delivery sector. Although
academic literature has largely covered delivery processes in platform-to-consumer businesses
(Berbeglia et al., 2010; Reyes et al., 2018), prevalent scheduling, and routing models so far and often
van Dun et al. /Customer-Centric Meal Delivery Processes
Twenty-Eigth European Conference on Information Systems (ECIS2020), Marrakesh, Morocco. 13
narrow their view on short-term efficiency (regarding costs and other efforts). At the same time, im-
proving customer satisfaction, as a measure for the incorporation of customer-centricity (Vakulenko et
al., 2019), can help organizations to build valuable and loyal customer relationships (Fornell et al.,
1996). Adopting the DSR paradigm, we created an artifact that serves as a customizable planning model
for the delivery service process within CCIS. The artifact assists organizations in determining how in-
coming orders should be sequenced and bundled to achieve a positive impact on operational efficiency
as well as customer satisfaction. We refer to our approach as Customer-Centric Route Generation
(C2RG). The C2P-factor we introduce to our model balances the demand for operational efficiency and
the customer-centric perspective. It can allocate strategic importance to customer-centricity within the
delivery process and varies depending on the CCIS process workload. By aggregating the complexity
of user-centric route-bundling into two interpretable components, the selection of measures for short-
term efficiency and long-term impact, the requirements set forth by the derived design objectives are
fulfilled. Striving for validity as well as the applicability of our artifact, we construct a software proto-
type that aims at a fair treatment of customers. This leads to prioritizing those orders whose customers
have experienced long waiting times in the past. We validate the implemented prototype for practical
applicability using real-world data from a Germany-based platform-to-consumer service. Simulation
results show positive and negative effects on delivery times as expected. On average, the ready-to-door
time slightly increases whereas the standard deviation decreases. This algorithmic adjustment is a start-
ing point. The underlying principle could and should be transferred to other real-world cases containing,
e.g., multiple depots or goods other than food.
Our planning model contributes to the prescriptive body of knowledge related to customer-centricity
and routing problems. We provide an empirical contribution in the field of CCIS and process decision-
making. The artifact is the first instantiated application of an integrated customer-centric perspective
with classical order bundling and route optimization techniques in a configurable service process fol-
lowing CCIS theory. To the best of our knowledge, the theory about customer-centricity has not yet
been applied in the field of last-mile meal delivery processes. In line with the paradigm of customer-
centric organizations, the integration of secondary measures into route bundling is only sensible when
considering how strongly customer experiences affect customer satisfaction and loyalty. These previ-
ously undocumented interrelationships between CRM and OR in CCIS set theoretical implications to be
uncovered in future work. Furthermore, we provide a practical contribution by providing a real-world-
validated model to solve the MDRP and its C2RG enhancement. The concept of including the order
history into route-generation allows organizations to strengthen customer-centric structures and set
themselves apart in the very competitive meal-delivery market. Moreover, by enabling a multitude of
different metrics for customer-centricity, organizations can customize the C2RG according to their own
business goals.
Concluding the paper, we also identified limitations and directions in which the C2RG can be further
developed. As for the model’s applicability, we see potential in enhancing the implemented bundling
heuristic to mitigate the effect of excessive re-bundling and to allow for pickups at multiple stores. The
overall approach needs to be evaluated in different practical settings to underpin our indication of ap-
plicability. In further examining indicators for customer satisfaction as appropriate proxy parameters for
configuring service processes in CCIS in last-mile meal delivery, we also see implications for academic
literature as well as to practice. Complementary, an opportunity for the improvement of routing pro-
cesses lies in the statistical forecasting of meal preparation times. However, we are confident to find
similar positive results in other settings of process planning and the incorporation of customer-centricity.
Hence, we encourage researchers to further explore the aspect of customer-centricity in last-mile deliv-
ery as well as general logistics processes.
van Dun et al. /Customer-Centric Meal Delivery Processes
Twenty-Eigth European Conference on Information Systems (ECIS2020), Marrakesh, Morocco. 14
References
Ahmadi-Javid, A., E. Amiri and M. Meskar (2018). “A Profit-Maximization Location-Routing-Pricing
Problem: A Branch-and-Price Algorithm.” European Journal of Operational Research 271 (3),
866881.
Alan Eera (2017). Optimization Algorithms for Meal Delivery Operations. URL:
https://www.slideshare.net/alerera/optimization-algorithms-for-meal-delivery-operations (visited
on 11/28/2019).
Berbeglia, G., J.-F. Cordeau and G. Laporte (2010). “Dynamic pickup and delivery problems.” Euro-
pean Journal of Operational Research 202 (1), 815.
Braekers, K., K. Ramaekers and I. van Nieuwenhuyse (2016). “The vehicle routing problem: State of
the art classification and review.” Computers & Industrial Engineering 99, 300313.
Buhl, H. U., R. Klein, J. Kolb and A. Landherr (2011). “CR2M—an approach for capacity control
considering long-term effects on the value of a customer for the company.” Journal of Manage-
ment Control 22 (2), 187204.
Deliveryhero (2018). Delivery Hero sells food delivery operations in Germany to Takeaway.com for
cash and shares and reinvests for further growth. URL: https://www.deliveryhero.com/delivery-
hero-sells-food-delivery-operations-germany-takeaway-com-cash-shares-reinvests-growth/ (visited
on 29-Nov-19).
Fornell, C., M. D. Johnson, E. W. Anderson, J. Cha and B. E. Bryant (1996). “The American Cus-
tomer Satisfaction Index: Nature, Purpose, and Findings.” Journal of Marketing 60 (4), 7.
Galbraith, J. R. (2005). Designing the Customer-Centric Organization: A Guide to Strategy, Structure,
and Process. San Francisco: Jossey-Bass.
Gartner (2019). Is Your Organization Customer Centric? URL: https://www.gartner.com/smarterwith-
gartner/is-your-organization-customer-centric/ (visited on 11/28/19).
Gregor, S. and A. R. Hevner (2013). “Positioning and Presenting Design Science Research for Maxi-
mum Impact.” MIS Quarterly 37 (2), 337355.
Henger, R. and C. Oberst (2019). Immer mehr Menschen verlassen die Großstädte wegen Wohnungs-
knappheit. IW-Kurzbericht Nr. 20.
Homburg, C., M. Droll and D. Totzek (2008). “Customer Prioritization: Does It Pay off, and How
Should It Be Implemented?” Journal of Marketing 72 (5), 110130.
Ioannou, G., M. Kritikos and G. Prastacos (2001). “A greedy look-ahead heuristic for the vehicle rout-
ing problem with time windows.” Journal of the Operational Research Society 52 (5), 523537.
Jayachandran, S., S. Sharma, P. Kaufman and P. Raman (2005). “The Role of Relational Information
Processes and Technology Use in Customer Relationship Management.” Journal of Marketing 69
(4), 177192.
Khan, M. M. and M. Fasih (2014). “Impact of Service Quality on Customer Satisfaction and Customer
Loyalty: Evidence from Banking Sector.” In: Pakistan Journal of Commerce and Social Sciences,
p. 331354.
Liang, T.-P. and M. Tanniru (2006). “Special Section: Customer-Centric Information Systems.” Jour-
nal of Management Information Systems 23 (3), 915.
Liu, W. and W. J. Florkowski (2018). “Online Meal delivery services: Perception of service quality
and delivery speed among Chinese consumers.” In: Annual Meeting of Southern Agricultural Eco-
nomics Association (SAEA). Ed. by M. Reed and S. Saghaian. Jacksonville, Florida.
van Dun et al. /Customer-Centric Meal Delivery Processes
Twenty-Eigth European Conference on Information Systems (ECIS2020), Marrakesh, Morocco. 15
Luxen, D. and C. Vetter (2011). “Real-time routing with OpenStreetMap data.” In: Proceedings of the
19th ACM SIGSPATIAL International Conference on Advances in Geographic Information Sys-
tems. Ed. by D. Agrawal. New York: ACM, p. 513.
March, S. T. and G. F. Smith (1995). “Design and natural science research on information technol-
ogy.” Decision Support Systems 15 (4), 251266.
Oliver, R. L. (1999). “Whence Consumer Loyalty?” Journal of Marketing 63 (4_suppl1), 3344.
Peffers, K., T. Tuunanen, M. A. Rothenberger and S. Chatterjee (2007). “A Design Science Research
Methodology for Information Systems Research.” Journal of Management Information Systems 24
(3), 4577.
Reyes, D., A. Erera, M. Savelsbergh, S. Sahasrabudhe and R. O'Neil (2018). “The Meal Delivery
Routing Problem.” Optimization Online.
Rust, R. T., C. Moorman and G. Bhalla (2010). “Rethinking marketing.” Harvard business review 88
(1/2), 94101.
Sander, N. (2017). “Germany. Internal Migration Within a Changing Nation.” In: Internal Migration
in the Developed World. Ed. by T. Champion, T. Cooke and I. Shuttleworth. Routledge, p. 226
241.
Schrage, M. (1994). On-Line Pizza Idea Is Clever but Only Half-Baked. URL:
https://www.latimes.com/archives/la-xpm-1994-08-25-fi-31168-story.html (visited on
11/18/2019).
Shah, D., R. T. Rust, A. Parasuraman, R. Staelin and G. S. Day (2006). “The Path to Customer Cen-
tricity.” Journal of Service Research 9 (2), 113124.
Sheth, J. N., R. S. Sisodia and A. Sharma (2000). “The Antecedents and Consequences of Customer-
Centric Marketing.” Journal of the Academy of Marketing Science 28 (1), 5566.
Sonnenberg, C. and J. vom Brocke (2012). “Evaluations in the Science of the Artificial Reconsider-
ing the Build-Evaluate Pattern in Design Science Research.” Design Science Research in Infor-
mation. Advances in Theory and Practice 7286, 381397.
Statista (2019). eServices Report 2019 - Online Food Delivery. URL: https://de.statista.com/statis-
tik/studie/id/40371/dokument/food-delivery/ (visited on 11/16/2019).
Steiner, E. (1994). PizzaNet -- the killer app. URL: https://web.ar-
chive.org/web/20120609100313/http://www.interesting-people.org/archives/interesting-peo-
ple/199408/msg00057.html (visited on 11/18/2019).
United Nations, Department of Economic and Social Affairs, Population Division (2019). World Ur-
banization Prospects: The 2018 Revision. Custom data acquired via website.
Vakulenko, Y., P. Shams, D. Hellström and K. Hjort (2019). “Service innovation in e-commerce last
mile delivery: Mapping the e-customer journey.” Journal of Business Research 101, 461468.
van den Hemel, C. and M. F. Rademakers (2016). “Building Customer-centric Organizations: Shaping
Factors and Barriers.” Journal of Creating Value 2 (2), 211230.
Xiang, Z., C. Chu and H. Chen (2008). “The study of a dynamic dial-a-ride problem under time-de-
pendent and stochastic environments.” European Journal of Operational Research 185 (2), 534
551.
Yildiz, B. and M. Savelsbergh (2019). “Provably High-Quality Solutions for the Meal Delivery Rout-
ing Problem.” Transportation Science 53 (5), 12131499.
... van Dun et al. [15] explored meal delivery processes by incorporating customer preferences, focusing on the impact of temperature and speed on customer SAT. Findings revealed a strong preference for hot and fast deliveries, with customers expressing higher SAT levels. ...
... The investigation's results indicate a notable correlation between PU and SAT with FDMA. According to Mohammed and Rozsa [36], and van Dun et al. [15], PU is a critical factor influencing SAT with mobile apps. Their findings emphasized that user who perceived the app as useful in facilitating their needs reported higher SAT levels. ...
Article
Full-text available
The rise of mobile technology has significantly transformed numerous aspects of our everyday lives, especially within food delivery services. The investigation aims to explore the food delivery mobile apps (FDMA) satisfaction (SAT) and the influence of familiarity (FAM). Data was gathered from 381 individuals who have experience in using any FDMA services specifically in Shah Alam, Selangor with the aid of online questionnaires. The study findings indicate user satisfaction (US) with FDMA is strongly influenced by the level of familiarity users have with the platform. The research result shows the satisfaction of users with FDMA is strongly linked to how easy they find the platform to use. The research provides a unique contribution by exploring the influence of familiarity on the US with FDMA. Investigating how users' prior experiences and comfort levels impact their satisfaction provides valuable insights for enhancing app design and user experience in the rapidly evolving food delivery industry. The study contributes by elucidating the significant impact of FAM on FDMA satisfaction. This insight aids in refining app design and strategies to enhance user experience. The study suggests optimizing FDMA by prioritizing features that enhance user FAM, ultimately developing higher SAT levels and improving overall user experience. The research findings indicate a notable correlation between the US and the inclination to maintain the usage of FDMA systems.
Technical Report
Full-text available
Das knappe Wohnungsangebot, steigende Mieten und Immobilienpreise dämpfen den zuletzt hohen Zuzug in die Großstädte. Die Ballungszentren bleiben zwar insbesondere für Studierende und Berufseinsteiger attraktiv. Familien bevorzugen dagegen zunehmend das Umland der Großstädte. Bereits seit 2014 ziehen im Saldo mehr Inländer aus den Großstädten heraus – Tendenz steigend.
Article
Full-text available
Although the number of companies declaring their commitment to the concept of customer centricity is increasing, many of these companies are struggling to actually shape their organization accordingly. In this article we share the results of our research, which aims to identify key shaping factors that help organizations move towards customer centricity, and key barriers that typically deter organizations from becoming customer-centric. Based on a literature review and field research among both practitioners and academics, we have identified a set of nine shaping factors and three barriers to building customer-centric organizations and explored these in more detail. Our study suggests that there are no simple recipes or shortcuts for organizations to become customer-centric. Instead, companies can reap sustainable competitive advantage from evolving the ability to master nine shaping factors and evade three barriers to customer centricity.
Article
Full-text available
Over the past decades, the Vehicle Routing Problem (VRP) and its variants have grown ever more popular in the academic literature. Yet, the problem characteristics and assumptions vary widely and few literature reviews have made an effort to classify the existing articles accordingly. In this article, we present a taxonomic review of the VRP literature published between 2009 and June 2015. Based on an adapted version of an existing comprehensive taxonomy, we classify 277 articles and analyze the trends in the VRP literature. This classification is the first to categorize the articles to this level of detail.
Article
Online restaurant aggregators with integrated meal delivery networks have become more common and more popular in the past few years. Meal delivery is arguably the ultimate challenge in last-mile logistics: a typical order is expected to be delivered within an hour (much less if possible) and within minutes of the food becoming ready. We introduce a novel formulation for a meal delivery routing problem (in which we assume perfect information about order arrivals) and develop a simultaneous column- and row-generation method for its solution. The analysis of the results of an extensive computational study, using instances derived from real-life data, demonstrates the efficacy of the solution approach, and provides valuable insights into, among others, the (potential) benefits of order bundling, courier-shift scheduling, and demand management.
Article
The remarkable growth of e-commerce has defined the recent years of various industries worldwide. Driven by consumers, the e-commerce surge (e-retail in particular) stems from the final leg of the supply chain: the last mile. As the growing flow of e-commerce orders continues to generate new records for annual revenues, key actors in the last mile face the challenges of increasing customer demands and transportation volumes. In response, e-retailers and logistics service providers seek innovative service solutions, often powered by technological advancements. This study consisted of focus group interviews and a usability test that incorporated an innovative technology in the delivery service. The study provides insights into how service innovation affects e-customer behavior and presents a basic map of the e-customer journey. The findings also provide a foundation for improving management of the customer experience and aiding managerial decision-making when designing new e-commerce last mile services.
Article
Both practitioners and academics understand that consumer loyalty and satisfaction are linked inextricably. They also understand that this relation is asymmetric. Although loyal consumers are most typically satisfied, satisfaction does not universally translate into loyalty. To explain the satisfaction–loyalty conundrum, the author investigates what aspect of the consumer satisfaction response has implications for loyalty and what portion of the loyalty response is due to this satisfaction component. The analysis concludes that satisfaction is a necessary step in loyalty formation but becomes less significant as loyalty begins to set through other mechanisms. These mechanisms, omitted from consideration in current models, include the roles of personal determinism (“fortitude”) and social bonding at the institutional and personal level. When these additional factors are brought into account, ultimate loyalty emerges as a combination of perceived product superiority, personal fortitude, social bonding, and their synergistic effects. As each fails to be attained or is unattainable by individual firms that serve consumer markets, the potential for loyalty erodes. A disquieting conclusion from this analysis is that loyalty cannot be achieved or pursued as a reasonable goal by many providers because of the nature of the product category or consumer disinterest. For some firms, satisfaction is the only feasible goal for which they should strive; thus, satisfaction remains a worthy pursuit among the consumer marketing community. The disparity between the pursuit of satisfaction versus loyalty, as well as the fundamental content of the loyalty response, poses several investigative directions for the next wave of postconsumption research.
Article
The American Customer Satisfaction Index (ACSI) is a new type of market-based performance measure for firms, industries, economic sectors, and national economies. The authors discuss the nature and purpose of ACSI and explain the theory underlying the ACSI model, the nation-wide survey methodology used to collect the data, and the econometric approach employed to estimate the indices. They also illustrate the use of ACSI in conducting benchmarking studies, both cross-sectionally and over time. The authors find customer satisfaction to be greater for goods than for services and, in turn, greater for services than for government agencies, as well as find cause for concern in the observation that customer satisfaction in the United States is declining, primarily because of decreasing satisfaction with services. The authors estimate the model for the seven major economic sectors for which data are collected. Highlights of the findings include that (1) customization is more important than reliability in determining customer satisfaction, (2) customer expectations play a greater role in sectors in which variance in production and consumption is relatively low, and (3) customer satisfaction is more quality-driven than value- or price-driven. The authors conclude with a discussion of the implications of ACSI for public policymakers, managers, consumers, and marketing in general.
Article
This paper for the first time considers a profit-maximization location-routing problem with price-sensitive demands. The problem determines the location of facilities, the allocation of vehicles and customers to established facilities, and the pricing and routing decisions in order to maximize the total profit of serving customers. A mixed-integer linear programming model is presented, which can only be used to solve small-size instances with commercial optimization solvers. Then, the model is reformulated as a set-packing model and solved by an efficient branch-and-price algorithm for large-size instances. The proposed algorithm can also be used to solve the more basic location-routing problem with profit where demands are not price-sensitive, which has not been considered by any research earlier. Our numerical study indicates the substantial advantage of the integrated model.