Digital Technologies in Logistics Services for the Agri-Food Sector: A Comparison of Poland and Lithuania

The adoption of digital technologies in food supply chains is becoming increasingly important to ensure efficiency, transparency, and sustainability. Academic literature also indicates numerous advantages that digital technologies can provide when effectively adopted in food logistics systems. The food logistics supply chain not only involves the physical movement of food from producers to consumers, but also the movement of information, capital, and services. Effective process management is essential to ensure food quality, reduce costs, improve service levels, and meet consumers’ expectations (Chen, 2024). These benefits encompass both operational improvements and strategic transformations that can increase competitiveness, environmental performance, and customer satisfaction.

Smart logistics systems, especially those that use route optimization and data analytics, help reduce delivery times and operational costs. The application of smart logistics in rural e-commerce has resulted in a 5% cost reduction and improvement in delivery times compared to traditional models (Wei, 2024). Digital transformation also involves the integration of complementary cutting-edge technologies. Smart technologies such as blockchain, the Internet of Things (IoT), big data analytics, cloud computing, smart contracts, 5G, and other advanced technologies support traceability, accuracy, and transparency in logistics. These technologies have become essential for implementing green logistics and reducing carbon footprints, as they allow for real-time cargo tracking, optimized transportation routes, and more efficient use of energy resources (Khatib & Barco, 2021; Ravi et al., 2023).

The literature review indicates that researchers focus on a limited group of solutions that combine two criteria: high utility in the food supply chain and growing maturity of implementation in business practice. Table 1 summarizes these technologies, highlighting their key applications and expected benefits for logistics operators.

As Table 1 shows, empirical studies most often highlight the value of blockchain and IoT in ensuring traceability and transparency, while AI and big data analytics are seen as tools that directly improve operational efficiency. A relatively small number of studies on digital platforms for redistributing food surpluses suggests a gap that is worth addressing in future research projects.

Smart technologies are transforming food logistics by making supply chains more resilient, transparent, and sustainable. Their combined application not only helps reduce waste and costs, but also improves food safety and customer trust. As global food systems face increasing pressure from climate change, population growth, and supply chain disruptions, the adoption of these technologies is important for ensuring long-term efficiency and security in food distribution.

Blockchain technology increases supply chain transparency and helps prevent information tampering, which is a very important advantage in environmentally sensitive industries. These technologies reduce human errors, improve routing, and optimize warehouse operations (Rahman et al., 2024).

Food supply networks are complex and interconnected, and IoT-based systems can monitor them to capture details on food materials and protect the ecosystem. The IoT platform can provide product traceability information in the food supply chain, assisting customers, especially during the COVID-19 pandemic-related disruptions where the information available is so vague. By combining IoT and block-chain technologies, food supply chain can become more transparent and productive by delivering robust and stable information to clients and related stake-holders (Abideen, 2021).

Digitalization significantly improves logistics performance through automation and investment in digital skills (Ntule et al., 2024). New technologies such as big data, artificial intelligence and automation are considered to be high-potential technologies in logistics. They support predictive decision-making and are crucial for the future of smart logistics (Kwon & So, 2023). Digital technologies also enable the development of multi-modal logistics. Smart logistics not only improves infrastructure, but also promotes regional economic equity and service quality. This is evident from the Smart Logistics Development Index (SLDI), which shows the improvement in customer satisfaction and competitiveness due to digital integration (Liu & Zhao, 2024).

Drone technology has transformed the food industry by enhancing logistics and supply chain efficiency through digital advancements. As part of Agriculture 5.0, drones support faster autonomous deliveries, aid humanitarian efforts, and promote sustainability with electric power. They also allow for better data collection and analysis, driving improvements across the food supply chain (Undre & Jokonya, 2024). Drone technology enables stake-holders to access and analyze data that was once unavailable across different stages of the food supply chain. This data supports the development and enhancement of supply chain processes. Logistics within the supply chain involves planning, executing, and managing the efficient and secure flow and storage of goods, services, and information to meet customer needs (Croom et al., 2018).

In addition to technology itself, intelligent tools and workforce support systems have proven vital. Artificial intelligence-based decision support systems and intelligent logistics tools provide real-time monitoring, predictive capabilities, and route optimization. These tools help staff manage perishable goods more effectively, reducing food waste and increasing sustainability (Yousefi et al., 2024). Coordination in logistics networks is another key benefit. Digital solutions such as smart sensors and automation improve coordination, eliminate inefficiencies caused by manual tasks, and improve the overall efficiency of the logistics system (Reddy & Kumar, 2024). From an environmental perspective, digital technologies support the development of green logistics, including through emission reduction, vehicle route optimization, and efficient resource management, supporting the achievement of circular economy goals (Lu et al., 2025).

Despite these advantages, the implementation process faces significant technological, institutional, and organizational challenges, including high initial investment and compatibility with existing systems (Saha et al., 2025), which vary according to the size of the company, the region and the national infrastructure. In addition, lack of skilled staff and resistance to change can hinder the adoption of innovative solutions (Sumets, 2023). As Verdouw et al. (2013) noted, one of the main logistic challenges in food sector is to deal with the high dynamics and uncertainty in supply and demand. There is great uncertainty regarding fresh product quality as well as available volumes in time in a specific place. The sector is characterized by last-minute changes and rush orders. As a consequence, the required prediction and planning concept and accompanying logistics system need to be very flexible, allowing for last minutes changes and reallocations, but also provide a robust planning.

One of the main technological challenges is the high cost of implementation and maintenance, which is especially burdensome for SMEs (Undre & Jokonya, 2024). Many companies still operate on outdated or incompatible systems, which complicates the integration of new technologies (Magdalena et al., 2024). Another common issue is underdeveloped digital infrastructure, particularly in rural or regional areas often involved in food supply chains (Akinbamini et al., 2025). Moreover, advanced technologies like digital twins and IoT remain underutilized in fruit and vegetable supply chains, despite their great potential to optimize operations (Falayi et al., 2024). Mondal et al. (2019) proposed an IoT-enabled distributed ledger technology (DLT) architecture aimed at enhancing transparency in the food supply chain. Their system integrated a proof-of-object authentication mechanism analogous to the proof-of-work used in cryptocurrencies. Such concept integrated with RFID sensors allowed for continuous real-time data collection throughout the supply chain, offering a transparent and traceable food supply system. However, the implementation of such a traceability system is associated with high operational costs, posing a significant barrier to its widespread adoption (Abideen, 2021).

Technological barriers are compounded by organizational and human resources limitations. Lack of qualified human resources, in particular specialists skilled in using IoT, AI, and data analytics, is a significant factor hampering digital technology adoption (Yousefi et al., 2024). Companies often lack internal competencies needed to operate new systems, and staff training is both costly and time-consuming. In addition, there is widespread organizational resistance to change, especially among leadership who may question the return on investment in new technologies (Hasan & Habib et al., 2024). Furthermore, some companies lack a clear strategic vision of how technology implementation aligns with business objectives (Undre & Jokonya, 2024).

At the institutional level, regulatory and financial frameworks can also create barriers. The lack of clear or sufficient regulations, especially with regard to new technologies such as blockchain or drones, impedes their practical application (Blantucas, 2023). In some countries, such as Bangladesh, regulations prohibit the use of certain digital solutions, such as cryptocurrency-based blockchain systems, forcing companies to seek legal alternatives (Hasan & Habib et al., 2024). In addition to regulatory constraints, the lack of financial support at the public or EU level is a significant issue, limiting the implementation of innovations by logistics companies, especially those that do not have sufficient internal resources (Akinbamini et al., 2025).

One of the key factors hampering digitization in food logistics is also the lack of trust and poor quality of cooperation between entities operating within the supply chain. Effective integration of digital solutions is based on open and transparent data exchange between partners, but the lack of mutual trust significantly limits this transparency (Ekhsonov, 2024). Although blockchain technology offers potential solutions for building durable and immutable data registers, its use remains limited, both due to legal uncertainties and low technological awareness (Hasan & Habib et al., 2024).

To summarize, the literature indicates that the implementation of digital technologies in the domain of food logistics is encumbered by many complex multidimensional barriers—not only technological, but also institutional, organizational, and social ones. This necessitates comprehensive modifications, encompassing the modernization of technological infrastructure, the formulation of pertinent regulations, the fortification of companies’ adaptive capacities, and the building of trust in relations between market participants. Despite the evident benefits of digital solutions in the field of food logistics, as demonstrated in extant literature, the transformation is impeded by persistent obstacles. Consequently, a comprehensive understanding of the barriers and benefits is necessary for an accurate assessment of the digital readiness of logistics companies in Poland and Lithuania. This understanding is also crucial for the establishment of practical pathways towards a more effective and sustainable digital transformation in the agri-food sector.

This study forms the basis of an international research project carried out in Polish–Lithuanian collaboration, which concerns the impact of smart technologies on sustainable development in logistics in Poland and Lithuania. The aim of the project is to assess and compare the use of smart technologies in logistics companies in both countries, taking into account their impact on green logistics practices and sustainable development. The paper focuses on the research results, concentrating on food supply chains.

The study covered all logistics companies operating in Poland and Lithuania and providing services to the agri-food sector. In order to ensure representativeness and enable analysis of the effect of scale, stratified sampling was used with the criterion of company size measured by the number of employees: small (up to 10 persons), medium (11–50), large (51–250), and very large (over 250). The proportions in each stratum reflected the base structure, determined on the basis of national registers of carriers and logistics operators.

The primary data was collected using a standardized questionnaire, developed on the basis of previous literature studies on smart logistics, innovation diffusion, and sustainable development practices. The research instrument consisted of four modules: 1) company characteristics, 2) scope of smart technology implementation, 3) motivations and barriers to adoption, 4) pro-environmental activities and CSR. For modules 3 and 4, a five-point Likert scale (1 = ‘definitely no’, 5 = ‘definitely yes’) was used, which allowed for the subsequent construction of synthetic indices.

The questionnaire was made available to respondents in electronic form in October 2024. The online format made it possible to reach a geographically dispersed population and ensured flexibility in responding. Conducting the study in both countries simultaneously, using identical tools and procedures, created a comparable database for further analysis of differences and determinants of smart logistics technology adoption.

The files in MS Excel format were imported into the Python environment, more specifically the Pandas library. Multiple-choice questions were converted into binary variables, while Likert scales were retained as ordinal variables. Data gaps not exceeding 5% were imputed modally (for binary variables) or by the median (for ordinal variables). Larger gaps resulted in the exclusion of observations from analyses requiring complete vectors. This scheme minimizes information loss while ensuring transparency of the data processing procedure.

The key measure in the study is the Logistics Technology Adoption Index (LTAI). It was defined as the sum of seven technologies normalized to the 0–1 range, with the weights of individual components determined on the basis of factor loadings obtained in the principal component analysis (PCA). The criterion for the one-dimensionality of the construct was a single component with an eigenvalue greater than one, explaining at least half of the variance. The technologies examined included: sensor-based load monitoring, extended telematics systems, big data / AI analytics, sensor networks, electronic waybills, digital freight exchanges, and warehouse automation with WMS systems. The LTAI (Equation 1) for the i-th company is therefore the arithmetic mean of the binary values corresponding to the seven technologies above.1IATLi−1k∑j=1kTij,k−7IAT{L_i} - {1 \over k}\sum\nolimits_{j = 1}^k {{T_{ij}}} ,k - 7

IATL_i – value of the Logistics Technology Adoption Index for the i-th company,

k – number of technologies included in the study,

T_ij – a binary variable (0 or 1) indicating the implementation of the j-th technology in the i-th company.

The internal consistency of the LTAI was assessed using Cronbach’s a coefficient, calculated according to the following formula: 2α=kk−1(1−∑j=1ksTj2s∑ jTj2)\alpha = {k \over {k - 1}}\left( {1 - {{\sum\nolimits_{j = 1}^k {s_{{T_j}}^2} } \over {s_{\sum j Tj}^2}}} \right)

α – Cronbach’s reliability coefficient,

k – number of technology variables in the analysis,

sTj2s_{{T_j}}^2 – variance of the binary variable describing the j-th technology,

s∑ jTj2s_{\sum j Tj}^2 – variance of the sum of variables in the sample.

Descriptive statistics were performed for all variables. In order to locate the source of differences, the share of companies using each technology was examined. For Tj technology, a 2 × 2 contingency table (country × adoption) was compiled and the χ² independence test was applied: 3χ2=∑ (ORC−ERC)2ERC{\chi ^2} = \sum {{{\left( {{O_{RC}} - {E_{RC}}} \right)2} \over {{E_{RC}}}}}

χ² – value of the χ² independence test statistic,

O_RC – real number observed in the table cell,

E_RC – expected number in this cell assuming independence.

In order to compare the level of logistics technologies adoption between companies in Poland and Lithuania, the Mann–Whitney test was performed. This is a non-parametric equivalent of the Student’s t-test for independent samples. Prior to the application of the LTAI distribution, the normality of this distribution in both samples was assessed using the Shapiro–Wilk test. This analysis revealed significant deviations from normality (p < .05). Consequently, the Mann–Whitney test was selected, which is robust to non-normal data and is suited for the comparison of two independent groups. The calculation was conducted in accordance with the prevailing standard formula.4U=n1n2+n1(n1+1)2−R1U = {n_1}{n_2} + {{{n_1}\left( {{n_1} + 1} \right)} \over 2} - {R_1}

n₁, n₂ – group sizes,

R₁ – sum of ranks in the first group.

In order to compare the assessments of cost and environmental motivations related to the implementation of logistics technologies, the Wilcoxon signed-rank test for paired samples was performed. This is a non-parametric test that allows us to assess whether the differences in the assessments of two related categories are statistically significant, regardless of the normality of the distributions. The analyses were performed at a significance level of p < .05 and the following formula was used: 5W=∑i=1nRiW = \sum\nolimits_{i = 1}^n {{R_i}}

W – sum of positive ranks,

R_i – ranks of absolute differences between pairs of observations.

To assess the relationship between the perception of barriers and the level of adoption of logistics technologies, Spearman’s rank correlation coefficients were calculated for both countries.6p=1−6∑ di2n(n2−1)p = 1 - {{6\sum {d_i^2} } \over {n\left( {{n^2} - 1} \right)}}

p – Spearman’s rank correlation coefficient,

d_i – rank difference for a pair of observations,

n – number of pairs.

Furthermore, in order to identify the determinants of adoption, logistic regression modelling was performed for the dependent variable: high adoption level (LTAI ≥ 0.5).7log(p1−p)=β0+β1x1+β2x2+…+βkxk\log \left( {{p \over {1 - p}}} \right) = {\beta _0} + {\beta _1}{x_1} + {\beta _2}{x_2} + \ldots + {\beta _k}{x_k}

The average rating of barriers and the status of a large enterprise were adopted as predictors. The analyses were performed at a significance level of p < .05.

In order to examine the effect of scale (differences between small/medium-sized and large companies), the Mann–Whitney test was used for two independent groups: SMEs (companies below the median number of employees) and large companies (companies above the median). The analyses were performed separately for Poland and Lithuania, at a significance level of p < .05.

Although the project is cross-sectional in nature and based on respondents’ declarations—which carries the risk of self-perception errors—its design is fully replicable. The implemented analytical code can be used for research in other countries in the region. In the future, this tool will allow for the expansion of analyses to include a panel approach, allowing for tracking the trajectory of digitalization and the assessment of the effectiveness of public support instruments.

The analysis of the calculated LTAI showed that the index values in the total sample and in individual countries differ significantly. The range of LTAI values for the entire sample was 0.00–0.86, with the lower quartile (Q1) at 0.00 and the upper quartile (Q3) was 0.67, indicating a clear right-sided asymmetry in the distribution. In further analyses, the LTAI level of ≥ 0.50 was considered the threshold for high adoption (implementation of at least four of the seven technologies studied), allowing digitally advanced enterprises to be identified.

The average LTAI value for enterprises in Poland was 0.431, while for Lithuania the average was significantly lower at 0.044, as shown in Table 2.

The internal consistency of the LTAI was confirmed using Cronbach’s a coefficient, which was 0.825 for the entire data set. This result exceeds the threshold of 0.7, indicating a high level of reliability in exploratory research. In the analysis divided by country, the value of the α coefficient for the Polish sample was 0.83, while for the Lithuanian sample it was 0.78. All values indicate high construct consistency.

In addition, a principal component analysis (PCA) was performed to assess the unidimensionality of the construct. The first component explained 57.3% of the total variance of the variables, which meets the eigenvalue criterion > 1 and confirms that the LTAI is unidimensional.

The analysis of differences in the use of specific technologies between Poland and Lithuania was conducted using χ² independence tests. The results are presented in Table 3.

The table presents the outcomes of χ² tests for five technologies that exhibited substantial differences in adoption rates across different countries. The remaining technologies were omitted, since they involved a variety of less significant solutions, and their results were aggregated and statistically insignificant. This approach serves to enhance the clarity of the presentation of results, thereby focusing on the key innovations that exert the most significant influence on the digitization of logistics services.

The findings of the study indicate substantial disparities in the implementation of the analyzed technologies. The greatest ones were observed in the domain of capital-intensive technologies, including intelligent vehicle systems and predictive analytics. It is evident that these technologies are being implemented by a significant proportion of Polish companies, while in Lithuania their use is practically non-existent. It is evident that other technologies have also demonstrated significant disparities, thereby confirming Poland’s distinct advantage in the implementation of smart logistics solutions.

The results of the Mann–Whitney test confirmed significant differences in the level of adoption of logistics technologies between Poland and Lithuania. The U statistic value was 2295.5, and the accompanying p value was less than .001, which indicates the significance of the differences. This result points to a significantly higher level of adoption of logistics technologies in Polish enterprises and falsifies the null hypothesis of no differences between countries.

The results of the Wilcoxon rank test showed that in both countries, the assessments of cost motivations were significantly higher than the assessments of environmental motivations. In Poland, the Wilcoxon statistic was W = 614.0 (p = .002), while in Lithuania it accounted for W = 441.5 (p = .011). The values obtained indicate that these differences are statistically significant (p < 0.05), which confirms the dominant role of cost motivations in the digitization of food logistics.

The results of the analysis showed a strong negative relationship between the perception of barriers and the level of adoption in both countries. Spearman’s correlation coefficients were –0.46 in Poland (p = .004) and -0.38 in Lithuania (p = .013), confirming the significance of the relationship (p < .05).

Table 4 presents the odds ratios (OR), 95% confidence intervals (CI), and p values for the predictors ‘average barrier rating’ and ‘large enterprise status’ in logistic regression models estimated separately on Polish and Lithuanian samples.

In the Polish sample, OR = 0.60 (95% CI [0.41–0.89]; p < .01) means that each 1-point increase in the mean inverse score of barriers reduces the chance of ‘high adoption’ by approximately 40%. The confidence interval is entirely below 1 (upper limit 0.89), confirming a significant (α = 0.05) negative relationship between the perception of barriers and the likelihood of implementing at least half of the possible smart technologies.

In the Lithuanian sample, OR = 0.576 (95% CI [0.0767–4.322]; p = .591) also indicates a decrease in the likelihood of adoption with an increase in perceived barriers, but the effect is not significant. The very wide confidence interval (more than a fifty-fold difference between the lower and upper limits) indicates considerable uncertainty in the estimation, mainly due to the small sample size (n = 44) and high variability in the responses of Lithuanian companies.

In Poland, OR = 2.15 (95% CI [1.00–4.62]; p <.05) means that large enterprises are more than twice as likely to have a ‘high adoption’ of smart technologies than smaller companies. The lower CI limit of exactly 1.00 suggests that the effect is marginally significant (p < .05) but still statistically confirmed. The upper limit of 4.62 indicates a possible fourfold increase in chances under favorable conditions.

In Lithuania, OR = 1.904 (95% CI [0.4698–7.716]; p = .367) once again points in a similar direction – large companies seem to have a higher chance of adoption – but the lack of significance and the wide 95% CI (covering 1 and reaching 7.7) mean that the effect is not statistically confirmed in this sample.

The following section presents a comparison of the mean LTAI values between SMEs and large companies, along with the results of the Mann–Whitney (U) test for the Polish and Lithuanian samples. The effect of scale is illustrated in Table 5, which compares the average LTAI values for SMEs and large companies in both countries, together with the results of the Mann–Whitney test. This comparison enables us to ascertain whether the size of the company significantly differentiates the level of digitization and whether this relationship occurs in parallel in Poland and Lithuania.

The analysis revealed a significant scale effect in the Polish sample: large companies achieved an average LTAI score of 0.591, while SMEs scored only 0.271. The Mann–Whitney test result (U = 306.0; p = 0.007) confirms the significance of these differences (p < .05). In the Lithuanian sample, the differences between large companies and SMEs were not statistically significant (p = .118). These results suggest that the scale effect only becomes clearly apparent in a more mature digitalization market, which was only observed in Poland in the study.