CC BY
0
№ 3 (125)
март, 2025 г.
ECONOMIC THEORY
ARTIFICIAL INTELLIGENCE IN PERSONALIZED MARKETING STRATEGIES: ANALYZING THE EFFECTIVENESS OF ALGORITHMS FOR TARGET AUDIENCE SEGMENTATION
Ekaterina Fomicheva
master's degree, State University of Management,
Russia, Moscow E-mail: _ [email protected]
ИСКУССТВЕННЫЙ ИНТЕЛЛЕКТ В ПЕРСОНАЛИЗИРОВАННЫХ МАРКЕТИНГОВЫХ СТРАТЕГИЯХ: АНАЛИЗ ЭФФЕКТИВНОСТИ АЛГОРИТМОВ ДЛЯ СЕГМЕНТАЦИИ ЦЕЛЕВОЙ АУДИТОРИИ
Фомичева Екатерина Андреевна
магистр,
Государственный Университет Управления,
РФ, г. Москва
ABSTRACT
This article analyzes the use of artificial intelligence (AI) in personalized marketing strategies, focusing on the effectiveness of algorithms for target audience segmentation. The role of machine learning (ML) and neural network approaches in improving segmentation accuracy, predicting customer behavior, and personalizing advertising offers is emphasized. Clustering methods, supervised models, and neural network algorithms are discussed, which allow for real-time adaptation of marketing campaigns, increasing engagement and conversions. The article also explores key metrics for evaluating the effectiveness of AI algorithms and their impact on the outcomes of marketing campaigns.
АННОТАЦИЯ
В статье анализируется использование искусственного интеллекта (ИИ) в персонализированных маркетинговых стратегиях, с акцентом на эффективность алгоритмов для сегментации целевой аудитории. Подчеркивается роль машинного обучения (МО) и нейросетевых подходов в улучшении точности сегментации, прогнозировании поведения клиентов и персонализации рекламных предложений. Рассматриваются методы кластеризации, супервайзед-модели и нейросетевые алгоритмы, которые позволяют адаптировать маркетинговые кампании в реальном времени, повышая вовлеченность и конверсии. В статье также изучаются ключевые метрики для оценки эффективности ИИ-алгоритмов и их влияние на результаты маркетинговых кампаний.
Keywords: artificial intelligence (AI), personalized marketing, segmentation, algorithms, machine learning (ML), neural networks, conversion.
Ключевые слова: искусственный интеллект (ИИ), персонализированный маркетинг, сегментация, алгоритмы, машинное обучение (МО), нейросети, конверсия.
Introduction
Modern marketing faces radical changes in the active implementation of artificial intelligence (AI) technologies. Within the conditions of growing competition and volume of data, companies are forced to make use of new, more qualitative methods of analysis of consumer behavior for the more precise definition of customer needs. Traditional approaches to audience segmentation based on demographic or behavioral characteristics already cannot fully allow for complex mul-tidimensionality of user preferences. In this context,
AI algorithms provide new opportunities for finding hidden patterns in the data automatically, thus contributing to precise targeting and personalized customer interactions.
The aim of this article is to investigate the efficiency of the work of different AI algorithms in the process of target audience segmentation for personalized marketing strategy development. The following research is dedicated to modern machine learning (ML) approaches and their use in marketing; it also includes
Библиографическое описание: Fomicheva E. ARTIFICIAL INTELLIGENCE IN PERSONALIZED MARKETING STRATEGIES: ANALYZING THE EFFECTIVENESS OF ALGORITHMS FOR TARGET AUDIENCE SEGMENTATION // Universum: экономика и юриспруденция : электрон. научн. журн. 2025. 3(125). URL:
https://7universum.com/ru/economy/archive/item/19448
№ 3 (125)
a comparative analysis of different algorithms according to their accuracy, interpretability, and practical applicability.
Main part. The role of AI in personalized marketing
Personalized marketing is a strategy aimed at tailoring marketing offers to individual customer needs based on the analysis of their behavior, preferences, and interactions with the brand [1]. Traditional segmentation methods, such as RFM analysis (recency, frequency, monetary) or demographic segmentation, have limited
март, 2025 г.
accuracy and fail to account for the dynamically changing preferences of consumers.
The approach to personalization is being radically transformed by AI through the use of ML, natural language processing (NLP), computer vision (CV), and recommendation algorithms. AI algorithms enable the analysis of large datasets, uncover hidden patterns, and offer personalized content to users in real-time. Table 1 presents the main methods and technologies of AI used in personalized marketing, along with their descriptions and examples of application.
Table 1.
Key AI technologies in personalized marketing [2]
Technology Description Examples of use
ML Analyzing user data and predicting customer behavior based on historical information. Content personalization in e-commerce, customer churn prediction.
Deep neural networks (DNN) Identifying complex patterns in data using DNN. Creating intelligent chatbots, analyzing user emotions.
NLP Analyzing textual information: reviews, comments, and user queries. Content optimization, automating customer support, sentiment analysis.
Recommendation systems Generating personalized offers based on user preferences. Product recommendations in online stores, personalized email campaigns.
CV Analyzing visual data (images, videos) for enhancing personalization. Automatically creating personalized ads based on images, emotion analysis in photos
Markov models and probabilistic algorithms Analyzing user behavior sequences, identifying patterns in behavior. Optimizing user flow in mobile apps and websites, behavior prediction.
Unlike traditional approaches, AI algorithms are capable of adapting to changes in user preferences by analyzing their behavior in real-time. This not only enhances the level of personalization but also significantly increases the effectiveness of marketing campaigns.
Artificial intelligence algorithms for audience segmentation
Audience segmentation is one of the key stages in personalized marketing, allowing companies to identify groups of users with common characteristics and preferences. Traditional segmentation methods, such as demographic analysis or behavioral targeting, are
limited in accuracy, whereas AI algorithms can analyze large volumes of data and uncover complex patterns. Among the main AI-based segmentation approaches, clustering methods, supervised models, and neural network algorithms can be distinguished.
Clustering algorithms enable audience segmentation without predefined labels, identifying groups of users with similar characteristics [3]. K-means is one of the most popular methods, which groups objects based on their similarity according to specified features (fig. 1).
Figure l. K-means algorithm for data clustering
№ 3 (125)
март 2025 n
This method is used in marketing to identify consumer segments, such as groups of customers with similar preferences in online stores. Due to its simplicity and effectiveness, k-means is also widely applied in customer base analysis, for example, to segment users based on purchase frequency or spending levels. It is
also used in advertising to identify the most relevant audience segments, allowing for targeted offers and personalized discounts.
DBSCAN (Density-Based Spatial Clustering of Applications with Noise) is a density-based clustering method that does not require predefining the number of clusters and works well with noisy data (fig. 2).
Figure 2. DBSCAN clustering of data
This method is suitable for segmenting users based on behavioral patterns, such as when analyzing webpage visits. Unlike k-means, DBSCAN can identify clusters of arbitrary shapes and effectively handle outliers, making it useful for analyzing irregular user trajectories. It is also applied in anomaly detection, for
example, to identify bots or fraudulent activities in online services.
Agglomerative clustering is a hierarchical method that builds a tree of relationships between objects, allowing for more detailed segmentation (fig. 3).
Figure 3. Agglomerative dustering scheme
Agglomerative clustering finds applications when one needs to have a hierarchical structure of segments, such as starting from very active customers and finishing with rarely purchasing customers. This is applied in loyalty programs for user classification into levels such as bronze, silver, gold customers by enabling finer tuning of marketing strategies via special discounts or reward programs.
The supervised methods are a class of ML algorithms trained on labeled data to predict the membership
of objects to certain categories or classes by identified patterns in the data items [4].
Random forest is a kind of ensemble method that combines several decision trees. It is highly accurate and interpretable, hence widely used in predicting consumer behavior, such as determining the probability of purchase (fig. 4).
№ 3 (125)
MapT, 2025 r.
Final Result
Figure 4. Random forest algorithm: combining multiple decision trees
Random Forest is widely used in marketing for audience segmentation and personalization due to its ability to process large volumes of data and uncover hidden customer behavior patterns. It is used to predict the likelihood of a purchase, allowing marketers to personalize advertising campaigns and offer individual recommendations to users. Additionally, the algorithm helps identify key characteristics of customer segments, such
as loyalty levels or churn propensity, enabling the adaptation of retention strategies and personalized offers.
Logistic regression is a classical classification method used to determine the probability of a user belonging to a specific segment (fig. 5).
Figure 5. Regression example
Logistic regression has a great number of applications in marketing; it can be used to predict the odds related to customers' responses after promotional offers or subscription to a newsletter. The approach is also utilized in segmenting the audience based on their level of activity. Therefore, the groups of active, passive, and potential customers can be determined.
Gradient boosting (XGBoost, LightGBM, Cat-Boost) is a method that outperforms many other algorithms in terms of accuracy due to its ability to effectively handle unstructured data and uncover complex nonlinear relationships (fig. 6).
№ 3 (125)
март 2025 n
Figure 6. Gradient boosting: combining multiple classifiers
It finds wide applications in marketing: segmenting customers, predicting behavior, and optimizing advertising campaigns. It enables the determination of the possibility of response to a promotional message in email marketing, analyzing historic data on user interactions with emails: opening, clicking, or unsubscribing. Gradient boosting is also applied for personal recommendations, determining what products or services will be of interest for a certain customer.
Neural network approaches are AI methods based on artificial neural networks, which allow modeling complex relationships in data, automatically uncovering hidden patterns, and adapting to changes. This makes them effective for segmentation and personalization tasks in marketing (table 2).
Table 2.
Neural network approaches in marketing [5, 6]
Method Description Application
DNN Used for processing large datasets and creating dynamic segments. Analysis of social networks to segment audiences based on interests.
Autoencoders Neural models capable of identifying latent features. Recommendation systems for identifying «hidden» customer needs.
Algorithms in AI allow marketers to segment audiences more accurately by analyzing multidimensional data and predicting customer behavior patterns. The method will be selected according to the volume of data, the necessity to interpret the results, and the aims of the marketing campaign. Approaches like k-means and DBSCAN are convenient in discovering hidden user groups; however, they are based on tuning of their parameters, which complicates the whole analysis process. Supervised ML methods, like Random Forest and gradient boosting, have high accuracy, especially for labeled data, hence being quite effective in predicting customer behavior. Neural network algorithms, including DNN and autoencoders, are the most computationally powerful and capable of identifying complex patterns; however, they require the most resources and are difficult to interpret. Therefore, AI applied in the segmentation of audiences helps increase the power of personalized marketing, reduce the cost of advertising, and enhance customer engagement levels.
Evaluation of AI methods' effectiveness in marketing strategies
Implementation of AI algorithms into marketing strategies is impossible without considering the testing of their results, as a choice of appropriate method influences the accuracy of audience segmentation, personalization of offers, and further conversion and return on investment in marketing campaigns. Critical success metrics will involve the accuracy of the forecast from the models, algorithms that will work to adapt changing user preferences, and scalability to big volumes of data.
Those of segmentation and personalization algorithms rely on a number of metrics as the driving undercurrent, further divided into the technical accuracy indications of the model and business metrics that would deal with the accountability of algorithms due to the consequence of marketing outcomes (table 3).
№ 3 (125)
март 2025 n
Table 3.
Metrics for evaluating algorithm effectiveness [7]
Type Metric Description
Technical Silhouette Score Evaluates clustering quality; shows the degree of separation between segments.
Rand Index Measures the agreement between segmentation and real data, used for comparing various clustering method
F1-score A combined metric that considers both the accuracy and completeness of model predictions; important when using supervised methods.
Business CTR (Click-Through Rate) Shows how well personalized offers attract user attention.
Conversion Rate (CR) Reflects the proportion of users who complete a targeted action (purchase, subscription, etc.).
ROI (Return on Investment) Calculates the return on investment in marketing, measuring the financial return on personalized marketing.
Evaluation of the effectiveness of segmentation and personalization algorithms in marketing requires a comprehensive approach that combines technical and business metrics. Technical indicators help determine the accuracy and relevance of model performance, ensuring quality segmentation of the audience. However, to assess the actual impact of AI on marketing strategies, business metrics are crucial, as they reflect user engagement, conversion rates, and the economic benefits of personalized solutions. The combined use of these metrics helps companies tailor algorithms to specific business objectives, improving their effectiveness and reducing costs.
One of the most successful examples of AI application is recommendation systems in e-commerce. Large online retailers such as Amazon actively use gradient boosting and neural network models to analyze customer preferences. These algorithms track browsing history, purchases, and interactions with the platform, forming personalized recommendations. This approach substantially raises the chances of purchase, enhances average order value, and encourages repeat purchase rates. All in all, customers get more relevant offers, while companies benefit from increased revenue and stronger customer loyalty.
Personalization within email marketing is also greatly aided by the effective use of AI. Brands such as Netflix and Spotify have turned to using AI algorithms that generate dynamic email content and automatically choose the optimal sending time. The technology analyzes data about the users' behavior for predictions of topics, subject lines, and offers that would receive the most interest from a specific recipient. This approach amazingly increases open rates and CTRs, increasing audience engagement hence conversion. Thereby,
an accurate personalized mail campaign and effectively communicating with one's customers happens.
In today's environment, no business can afford to overlook the need for AI in marketing, as it is a crucial tool for ensuring heightened accuracy in segmenting and personalizing content. The usage of ML algorithms here helps to optimize the marketing budgets for enterprises by putting forward relevant products, services, and promotional offers for the customer experience. However, when the volume of processed data rises along with personalization, personal data security and cy-bersecurity standards become a matter of high importance [8]. Protecting customer privacy and preventing data leaks should be an integral part of implementing marketing strategies using AI.
Conclusion
The use of AI in personalized marketing improves audience segmentation, predicts customer behavior, and enables real-time campaign adaptation. Machine learning and neural networks allow for deeper data analysis, revealing patterns that enhance personalization effectiveness. Clustering methods identify customer groups, while supervised models predict purchases, boosting conversion rates and engagement. AI technologies optimize marketing budgets by reducing ineffective campaigns. Successful implementation requires evaluating both technical metrics (like Silhouette Score, Rand Index, and F1-score) and business metrics (such as CTR, conversion rate, and ROI). Leading companies like Amazon, Netflix and Spotify show that AI automation not only enhances personalization but also drives significant sales growth and customer loyalty.
References:
1. Babatunde S. O., Odejide O. A., Edunjobi T. E., Ogundipe D. O. The role of AI in marketing personalization: A theoretical exploration of consumer engagement strategies // International Journal of Management & Entrepreneurship Research. 2024. Vol. 6. №. 3. P. 936-949. DOI: 10.51594/ijmer.v6i3.964 EDN: LUETQI
2. Lee G. H., Lee K. J., Jeong B. Kim T. K. Developing personalized marketing service using generative AI //IEEE Access. 2024. Vol. 12. P. 22394-22402. DOI: 10.1109/access.2024.3361946 EDN: QXKTZN
№ 3 (125)
UNIVERSUM:
ЭКОНОМИКА И ЮРИСПРУДЕНЦИЯ
• 7universum.com
март, 2025 г.
3. Ikotun A. M., Ezugwu A. E., Abualigah L., Abuhaija B., Heming J. K-means clustering algorithms: A comprehensive review, variants analysis, and advances in the era of big data //Information Sciences. 2023. Vol. 622. P. 178-210. DOI: 10.1016/j.ins.2022.11.139 EDN: DQYTMO
4. Manoharan G., Mishra A. B., Adusumilli S. B. K., Chavva M., Damancharla H., Lenin D. S. Supervised Learning for Personalized Marketing Strategies //2024 International Conference on Advances in Computing, Communication and Applied Informatics (ACCAI). IEEE, 2024. P. 1-6. DOI: 10.1109/ACCAI61061.2024.10602307
5. Kidassova M. Innovative business strategies in the digital transformation era // Economic Development Research Journal. 2025. № 1. P. 118-124.
6. Yang Q., Hayat N., Mamun A. A., Makhbul Z. K. M., Zainol N. R. Sustainable customer retention through social media marketing activities using hybrid SEM-neural network approach //Plos one. 2022. Vol. 17. №. 3. P. e0264899. DOI: 10.1371/journal.pone.0264899 EDN: PWNEJW
7. Saura J. R. Using data sciences in digital marketing: Framework, methods, and performance metrics //Journal of Innovation & Knowledge. 2021. Vol. 6. №. 2. P. 92-102. DOI: 10.1016/j.jik.2020.08.001 EDN: FQKXLB
8. Israfilov A., Yakovishin A. Protection of privacy and data on the internet. ISJ Theoretical & Applied Science. 2024. Vol 139. № 11. P. 42-46. DOI: 10.15863/tas.2024.11.139.10 EDN: YTTQHG
