=Paper=
{{Paper
|id=Vol-2922/paper015
|storemode=property
|title=Optimizing promotional campaigns and evaluating their efficiency using simulation modeling
|pdfUrl=https://ceur-ws.org/Vol-2922/paper015.pdf
|volume=Vol-2922
|authors=Yuriy Leonov,Nadezhda Malyavko,Anna Sazonova,Lyudmila Filippova
}}
==Optimizing promotional campaigns and evaluating their efficiency using simulation modeling==
https://ceur-ws.org/Vol-2922/paper015.pdf
Optimizing promotional campaigns and evaluating their
efficiency using simulation modeling*
Yuriy Leonov, Nadezhda Malyavko, Anna Sazonova and Lyudmila Filippova
Bryansk State Technical University, 7,50-letiya Oktyabrya boulevard, Bryansk, 241035, Russia
libv88@mail.ru
Abstract. The article is devoted to analysing and researching market behaviour
when using promotional campaigns for identical products. The main points in-
fluencing the work of advertising campaigns and users' behaviour are investi-
gated. An algorithm for conducting an optimization experiment is proposed and
recommendations for the promotion are given. The proposed comprehensive
approach allows you to reduce the cost of advertising campaigns and determine
the product optimal competitive price
Keywords: Promotional campaign, simulation modelling, advertising, simula-
tion model, promotional activities.
1 Introduction
Currently, the problem of optimizing promotional campaigns and the duration of us-
ing advertising funds is becoming more urgent. Any company and enterprise strives to
reduce advertising budgets as much as possible, while increasing the market share of
the product. The growth of the advertising direction is due to the fact that great com-
petition among the same products encourages business owners to buy more and more
advertising. However, not every campaign is successful, as it is influenced by many
external factors.
Advertising will be effective only if it is implemented successfully, and it will also
contribute to the quick and uninterrupted sale of products. For a competent launch,
you need to develop a promotional campaign strategy. Most of the Russian owners
and management companies tend to use single promotional campaigns. Often they
resort to them in extreme cases as “first aid” and expect immediate positive results.
[5]
This approach can hardly be called promotion in the modern sense of the word, and
it is unlikely to bring the expected results in the form of the increased sales of prod-
ucts or services.
Developing the strategies helps to avoid mistakes in advertising. It allows minimiz-
ing the risks associated with the consumer's misunderstanding, and also allows in-
*
Copyright c 2021 for this paper by its authors. Use permitted under Creative Commons License Attribu-
tion 4.0 International (CC BY 4.0).
creasing the advertising effectiveness. That means that developing a promotional
campaign strategy allows the company to successfully cope with its sales problems,
even gives an opportunity to compete more prosperously with other companies.
If a firm develops a strategy for a promotional campaign, it avoids a lot of mistakes
in its implementation and makes the advertising that is aimed at the consumer more
accurately than those rash and meaningless promotions, which sometimes just harm
the company.
To build a simulation model, the AnyLogic package is chosen, developed by The
AnyLogic Company, which is of Russia heritage.
The developers of this product did not limit the user to one modelling paradigm, al-
lowing him to choose from three main areas of modelling: discrete-event modelling,
system dynamics, agent-based modelling. Another distinguishing feature of the Any-
Logic package is its object-oriented approach to modelling. It helps to greatly sim-
plify the work with complex systems, making the process of representing the complex
system structure more natural, which ultimately allows facilitating and speeding up
the model creation. Another feature of the AnyLogic package is the ability to combine
various modelling directions within one model (for example, agent-based modelling
and system dynamics). Another AnyLogic package concept is to represent the model
as a set of parallel functioning activities, consisting of both one instance of active
objects and several ones that independently interact with the environment. Another
important part of the package is a user-friendly and intuitive interface supplied with
all kinds of support tools. These tools allow new users to quickly adapt to applying
this product, and simplify the process of creating models. [1].
1.1 Formulating the problem
Developing simulation modelling allows solving the problems of the Internet – entre-
preneur, providing him with a clear vision of the possible consequences when making
decisions for advertising. This tool gives you an opportunity you to quickly analyze
the current state of affairs, optimize the enterprise present activities, reduce advertis-
ing costs, and develop a plan for further actions. [2]
The tasks of simulation modelling include creating an accurate, adequate model of
a real object and the dynamics of its functioning.
Due to the minimal distortion of the object structure, the results obtained are as
close as possible to the real ones.
Thus, a properly constructed model is an almost accurate representation of a real
object with the properties of interest.
Working with models allows experimenting with real social objects and tracking
the possible reaction to a particular promotional campaign, as well as avoiding the
risk of unsuccessful budget investment in advertising.
The promotional campaign model described in AnyLogic software will help soci-
ologists and marketers who know in advance human behavioural principles to analyze
how the population will perceive this innovation; will allow drawing conclusions. [1]
1.2 The problem of modeling promotional campaigns
At first sight, the advertising market allows many companies to get the largest share
of customers, while the cost of advertising is usually reflected in the product price
category. As a result, when using advertising tools in the company, clients are brought
in each attracted client performs such actions as:
─ comes to the trade or commercial premises;
─ visits the website looking for the right product or orders a service;
─ contacts the manager or seller;
─ draws up a deal;
─ pays for the product [3].
The more advertising attracts customers, the more profitable the business will be.
However, the amount of funds spent on advertising is not always directly proportional
to increasing customers. Sometimes your advertising budget is wasted, which is the
main problem of promotional campaigns.
2 Materials and methods
2.1 Promotional campaign model algorithm
An abstract consumer market and the competition with the influence of promo-
tional campaigns on consumers are selected for modelling and studying. Consum-
ers live in five different towns and buy a certain product. The product is manufac-
tured by three competing companies. Each consumer has one favourite type of
product (and shows a certain group loyalty), but may switch to a different type due
to the influence of other consumers, prices, promotions, or by accident. Consumers
can move from one town to another and interact with other people in the same city.
An integrated approach to making a decision on the most favourable product price
and the number of promotional campaigns is proposed:
─ Visual observation and the indicator analysis such as market share, sales dynamics,
weekly revenue and total revenue.
─ Creating a consumer behaviour model in AnyLogic.
─ Graphic representation of towns and consumer group interaction.
─ Testing the proposed solutions on a simulation model to identify their advantages
and disadvantages.
─ Modernizing a particular company work by adjusting the parameters affecting the
user's behaviour and the promotional campaigns themselves.
On the basis of the proposed integrated approach, an algorithm for optimizing a
promotional campaign is developed and implemented to increase the market share of
a particular company (Figure 1).
Start
Generating towns
with consumers
Controlling consumers' preferences
Initial data: random selection of products, rate of ran-
dom selection, contact, duration of the contact with adver-
tising, contact rate, contact volume, daily necessity, weight
influence, memory influence, movement, rate of movement,
number of product types, price, remaining influence.
Collecting statistical data on the launch time of promo-
tional campaigns and the number of users who interacted
with advertising in AnyLogic
Assessing the influence of the Republic of Ka-
zakhstan on consumers
Yes Is the market share
of the product maxi-
mum?
No
Editing the advertising weight and the duration of the interaction
Editing the product price
No Is advertising
effective?
Yes
The goal has been achieved.
The end
Fig.1 Algorithm for optimizing a promotional campaign.
2.2 Developing a simulation model in the AnyLogic environment
A simulation model of influencing a promotional campaign on the consumer market
is built.
A number of agents are created to implement the model. In the agent-based model-
ling, an agent is a model element that can have behaviour, memory, contacts, etc. [4]
Agents simulate the target audience, publishing campaign impact, the town, as well as
the general environment of Main.
Parameters are used to create an agent's static characteristics. With their help, you
can set different parameter values for various agents of the same type, which is re-
quired in cases where agents have similar behaviour, but they have different charac-
teristics. For the “Action Impact” agent, the following parameters are set:
─ type of product;
─ duration of the campaign;
─ start of the campaign;
─ volume.
The agent displaying all consumers of the target audience is named “CA”, its
graphic image of one representative of the target audience is presented in the form of
an oval.
The following parameters are added for the “CA” agent:
─ accidental event “accidentalSwitch”;
─ random event rate “accidentalSwitchRate”;
─ contact “contact”;
─ duration of the contact with advertising “contactPromotionDuration”;
─ contact rate “contactRate”;
─ contact volume “contactVolume”;
─ daily necessity “dailyNeed”;
─ weight influence “influenceWeight”;
─ memory influence “memoryWeight”;
─ movement “move”;
─ rate of movement “moveRate”;
─ number of the product types “nProductTypes”;
─ price “price”;
─ remaining influence “remainingInfluence”.
Also, variables are added to the CA agent to model the agent’s changing character-
istics and to store the model work results.
─ changing “change” with the initial parameter value of the number of the product
types nProductTypes;
─ favourite among the products “favoriteproductType” with the initial value of 0;
─ advertising influence “influence”;
─ memory “memory”;
─ preference “preference”;
─ town “town”.
Next, functions are created, with their help certain sequences of actions are set that
are performed from different places or at various moments in the model life.
The following functions are created:
─ Searching for a specific consumer's location “findLocation”.
─ Calculating consumers' maximum preference.
─ Since consumers can change their product preference, a function is created to
switch in the user's preference for the product.
─ After a consumer changes preferences, it is necessary to update the consumers'
state.
─ Events are created to plan any actions in the model. Events are often used to simu-
late delays and timeouts. In some cases, behaviour can only be modelled using
events.
Therefore, the following events are generated:
─ accidentally changing the product type “accidentallySwitchProductType”;
─ influence on other consumers “contactBuddy”;
─ moving to a different town “moveToAnotherTown”;
Dynamic events are used to schedule an unlimited number of events that perform
similar actions, which can be scheduled independently and in parallel with each other.
Also dynamic events, namely the timer for the end of the influence “InfluenceEnd-
Timer” and the timer for the beginning of the influence “InfluenceStartTimer” are
added.
The agent Town is created similarly with the following parameters:
─ border – boundary;
─ relativeArea – area of the town.
Since a certain number of consumers live in town, it is necessary to create a group
of objects called a collection which is used to define a data object that combines sev-
eral elements of the same type (consumers) at once.
Next, a function is created that is responsible for the goods promotion in town (lo-
cal promotional campaign).
After all the agents are created, they are linked and located in the Main environ-
ment.
The CA and Town agents are added to the diagram, the variables are created, as
well as the parameters that will be used to graphically represent the model:
─ currentSales – present sales;
─ pickedCA – brought consumer;
─ price – value;
─ promotions – promotional campaign;
─ totalRevenue – total receipts;
─ totalSales – total sales amount;
─ weeklyRevenue – weekly returns.
We register the functions:
─ globalPromotion – global promotion;
─ randomTown – selecting a random town;
─ townArea – forming a town area;
─ townBounds – town boundaries;
─ updateCas – updating consumers;
─ updateStats – updating statistics.
Next, a daily update event is created.
As a result of all the transformations, the Main agent looks like this.
The last stage of modelling is the graphical presentation of the model. When draw-
ing by adding elements for 3 different companies, using the sliders you can adjust the
price and enable the Global promotion.
Also, using the sliders, you can control the variables responsible for the product
promotion.
Then we create a graphical representation of the towns.
The main information of modelling is presented in the lower part, graphs display
the preferences of a random consumer, market share of the companies, sales dynamics
and the company revenues (Figure 2).
Fig. 2. Graphical representation of the model.
3 Discussion
When you launch the model, you can visually see the distribution of a particular
product depending on the promotion parameters.
Let us conduct an experiment, i.e. we will set the minimum price for one company
without considering the promotion, for the second we will set the average price and
consider one global promotion, and for the third we will set the maximum price and
consider the largest number of global promotions.
Then, after 130 days, you can see that the most popular products are those of com-
pany 1 (Figure 3).
Thus, we can conclude that with a strong difference in price, consumers choose the
cheapest product. However, at the same price, promotion becomes a decisive factor in
the product selection.
4 Conclusion
In conclusion, it can be noted that the most important condition for effective market
activity is creating and producing competitive goods and services.
Fig. 3. Experiment results.
A well-thought-out promotional campaign influences various aspects of entrepre-
neurial activity, encouraging the initiative in producing new products, using the
achievements of scientific and technological progress, the fashion factor, and so on.
Advertising promotes the sale of products, the process of converting goods into
money, contributes to the accelerated and successful completion of the turnover proc-
ess, that is, to the reproduction at the firm level. It can design and manage the demand
and the market. Advertising is a channel for disseminating information in the market,
as well as a prerequisite for the feedback from it.
Then we will make the price the average for all the goods and track the subsequent
results (Figure 4).
Fig. 4 Results of the modified experiment.
The effectiveness of promotion is achieved not only by means of the simultaneous
use of many advertising tools and techniques, some of which complement and en-
hance the action of others, but also by means of a competitive price. The model shows
that at very different prices for similar products, the user will choose a cheap analogue
rather than an advertised brand.
References
1. Boev, V.: Conceptual design of systems in AnyLogic and GPSS World (Moscow: The
Internet University of Information Technologies (INTUIT)) (2016).
2. Vasiliev, A., Isoyanko E. V.: Organizing and planning advertising activities: manual for
universities ( Moscow: UNITY-DANA) (2017).
3. Vasiliev, G., Polyakov V. A.: Basics of advertising; manual for universities(Moscow:
UNITY-DANA) (2017).
4. Efromeeva, E., Efromeev, N. M.: Simulation: the basics of practical application in the
AnyLogic environment: manual (Saratov: University education) (2020).
5. Shchepilova, G., Shchepilov, K., Krasnyuk V.: Introduction to advertising research.
(“Elite-2000”).
6. Filippov R.: Internet of Things and Security (International Scientific and Practical Confer-
ence "Innovations in Industry, Management and Education", Bryansk: BSTU, 76-78
(2017).
7. Kuzmenko A., Kondrashin D.: Methods and approaches to the development of a system
for automated analysis of the dynamics of changes in the surfaces of forest stands on the
basis of automatic recognition methods, 6, 230-240 ( 2019).
8. Filippov, R. A., Filippova, L. B., Sazonova, A. S.: Internet veshchey: osnovnyye ponyatiya
(Bryansk: BGTU) (2016)
9. Leonov, E., Leonov, Y., Kazakov, Y., Filippova, L.: Intellectual Subsystems for Collecting
Information from the Internet to Create Knowledge Bases for Self-Learning Systems (Pro-
ceedings of the Second International Scientific Conference “Intelligent Information Tech-
nologies for Industry” (IITI’17). IITI 2017. Advances in Intelligent Systems and Comput-
ing) 679. Springer, Cham. 94-103(2018).