At this point, the authors conducted a daily, 6-month longitudinal study to better understand the users' behavior on a logistics startup's website (Zhu, Dong & Luo, 2021) after assessing the subject several times throughout the period. Consequently, two of the main characteristics of the empirical research are the “recreation” of the study and the “generalization” of the finding on a larger scale (Hubbard & Lindsay, 2002; Zhu, Dong & Luo, 2021). In this study, the authors extracted big data from 9 logistics startups of 2022 (Seedtable, 2023; Explodingtopics, 2022); based on size and profit; with the assistance of the web analytics tools Alexa and SEMRush which satisfies this characteristic, since other researchers can follow the same process and extract the same data. Additionally, empirical research has been used since is a reliable method of investigation that eliminates the possibility of data misinterpretation and isolates any possible cognitive bias which is crucial for big data studies (Zhu, Dong & Luo, 2021). At this point in the research, we gathered web analytics data from nine logistics startups’ websites. Web analytics data gathered from all those websites relates to metrics from website users and their behavior in the webpage such as pages per visit and average time on site. To accomplish this, the authors, used decision support systems as well as two platforms that extract these analytics, SEMrush, and Alexa. Those data have been observed and extracted daily for a period of 180 consequent days, which allows a more accurate analysis of the examined metrics.

The logistics startups web analytics that gathered from SEMrush and Alexa are: Wolt, Urbantz, Sennder, Paack, Lalamove, Byrd, FourKites, Centrica, Cargo.one. Since the focus is located on the user experience, the authors extracted the behavioral big data (average time on site, pages per visit) from the two web analytics platforms in order to assess the experience without making a questionnaire and asking questions that might lead to various opinions and consequently to biased results (Power, Cyphert & Roth, 2019). As Table 2 illustrates, we gathered behavioral data for 4.717.930 visitors in total over a six-month period for 9 logistics startup websites. Some other behavioral KPIs that were used for this purpose were “organic traffic”, “bounce rate”, “average time on site” and “pages per visit”. Those data were gathered for every company daily from the web analytics platforms and after six months we initiate the statistical analysis process followed by the creation of a fuzzy cognitive map and agent-based model.

The gathering of web analytics from the startup's webpages, followed by statistical analysis, revealed the presence of correlations between the extracted data. An FCM is a macro-scale analysis model that has been developed to illustrate the power of the correlations, which could be used in the development of an effective marketing strategy as well as on the decision-making process (Salmeron, 2009). The primary goal of developing the FCM is to create a visual illustration of the positive or negative intercorrelations, as well as the cause-and-effect relationships between the web analytics that are under consideration (Salmeron, 2009). By illustrating the correlation between the metrics, FCM offers the ability of a macro perspective on the development of a better digital marketing strategy.

After conducting a macro-scale analysis which depicted an overall description of the study, a micro-scale analysis was also required in order to provide a complete picture of the situation. ABM could indeed produce an effective simulation by displaying the dynamic relationships between key variables (Giabbanelli, Gray & Aminpour, 2017). ABM offers the options to estimate the influence of logistics startups websites’ user engagement parameters such as average time on site, and pages per visit on their global rank and profit, during a 180-day duration.

An ABM works from the interaction of various characters (agents) which are characterized by specific characteristics and run inside the system elicited from predefined “if-then” rules (Barbati, Bruno & Genovese, 2012). Various previous research that implemented relevant methodological approaches has adopted this method, especially in the digital marketing field (Kavak,Padilla, Lynch & Diallo, 2018). The implementation of ABM is advantageous for marketing managers as the simulation that is produced, promotes the improvement of the existing marketing plans (Zhang & Vorobeychik, 2019; Kavak, Padilla, Lynch & Diallo, 2018). In general, an Agent-based model monitors organic traffic and then produces with the assistance of the statistical analysis behavioral data of the user's activity on the website (Sakas & Reklitis, 2021a).

This chapter exhibits the outcomes of the data collection platforms. IBM SPSS Statistics 23 was utilized for the statistical analysis of this article. The findings in question are made up of raw data (gathered from nine logistics startups webpages), as shown in Table 1, which have been examined using statistical analysis software. After the data gathering of the logistics websites, the extracted data from every category was merged in order to conduct a per sector data analysis. Table 2 displays the descriptive statistics for the collected web analytics metrics, more specifically, the standard deviation, min, max, and mean. These values are the outcome of web analytics metrics during a 180-day duration. For instance, the “Logistics Startup's Pages/Visit” means, is the mean of all nine startups’ webpages over 180 days.

Regarding the (H1) on the Table 3, a significant positive correlation with ρ=0.945** was found between Visitors and Organic traffic, indicating that when the Organic traffic reach higher levels more users access the websites. Additionally, significant negative correlations have been found between the Global Rank and the Organic traffic with ρ=−.584** and between the Global Rank and Visitors with ρ=−.501**. When Organic traffic and Visitors reach higher levels the Global rank takes lower prices. This is a good sign since lower levels of global ranks provide a better brand name to the company. For instance, when the global rank is 2 (lower level) is better than a company with a global rank of 15 (higher level). The Regression for the (H1) is illustrated in Table 4. Regression analyses are statistically significant, with p values of less than 5%. The model is not significant and with every 1% rise of Global Rank and total Visitors, organic traffic decreases by 14,8% and increases by 87.1% respectively.

As for the (H2) on the Table 5, a significant positive correlation with ρ=0.724** was found between Paid Traffic and Social traffic, showing that, when an advertisement is placed on a search engine benefits the company not only with visitors that access the webpage through the advertisement but also with more visitors from the website's social networks. Moreover, a significant negative correlation has been found among the Global Rank and the Paid traffic with ρ=−.475**. That practical means, the paid advertisements provide a beneficial result to the company's brand name through climbing the site on higher levels, for instance from 12th place to 11th. Finally, a significant positive correlation with ρ=0.624** was found between Global rank and social traffic, showing that, when an advertisement is placed on a social media website profits the company's brand name. The Regression for the (H2) is illustrated in Table 6. Regression analyses are statistically significant, with p values of less than 5%. The model is significant and with every 1% rise of Paid Traffic and Social Traffic, Global Rank decreases by 4,9% and by 58.9% respectively.

In regard to the third hypothesis (H3) on the Table 7, a significant positive correlation with ρ=−0.370** was found between Global rank and Average Time on Site, representing that company's brand name can benefit if the user stays more time on the website. Additionally, non-significant correlations have been found between the Global Rank and the Pages per Visits with ρ=−.046 and between the Average Time on Site and Pages per Visits with ρ=0.054. That practically means, it's unimportant for the company's brand name how many web pages will be viewed per visitor on the website. The Regression for the (H3) is illustrated in Table 8. Regression analyses are statistically significant, with p values of less than 5%. The model is not statistically significant and by every 1% rise of Average Time on Site and Pages per Visits, Global Rank increases by 37.4% and decreases by 6.6% respectively.

Regarding the, the (H4) on the Table 9, a significant positive correlation with ρ=0.694** was found between Organic traffic and User Engagement, meaning that when a visitor has a positive feeling while browsing on a website will access it again. Additionally, non-significant correlations have been found between the Bounce Rate and the Organic traffic with ρ=0.030, and between the User Engagement and Bounce Rate with ρ=−0.107. When a user doesn't find the content attractive or doesn't feel good on the webpage (user engagement) will leave from the webpage (Bounce Rate), which illustrates the user's expected behavior. The Regression for the (H4) is illustrated in Table 10. The model is not significant and with every 1% rise of Organic Traffic and Bounce Rate, User Engagement increases by 69.8% and decreases by 12.7% respectively.

As for the final Hypothesis (H5) on the Table 11, a significant negative correlation with ρ=−.381** was found between User Engagement and Global Rank, meaning that when a user has a positive feeling for the logistics startup website, will access it again and this behavior contributes to better Global rank levels which directly contribute to the brand name. Additionally, a significant negative correlation with ρ=−.433** was found between User Engagement and Profit.That interestingly illustrates, when a user spends more time on a website or browses into different pages on the website has fewer possibilities to financially contribute, buy a product or a service. Practically means, that the websites need to be more technically effective, easy to use, and goal-oriented (to sales or services) instead of amusing the customers. Finally, a non – significant correlation had been found between the Global rank and the Gross Profit. The Regression for the (H5) is illustrated in Table 12. The model is significant and with every 1% rise of Global Rank and Gross Profit, User Engagement decreases by 38.3% and increases by 43.5% respectively. The Durbin-Watson test, according to Savin and White (1977), is the most common way to assess the assumption of independence. The test statistic has a value between 0 and 4, and lower values suggest that consecutive residuals are positively connected. The Durbin-Watson test was applied for all hypotheses to assess for linear regression dependence. The Durbin-Watson test for the first hypothesis is 2189. The Durbin-Watson test for the H2 is 1038, the H3 is 1072, and the H4 is 1994. Finally, the Durbin-Watson test for H5 is 1484. The hypotheses H2 and H3 are close to 1000 which means that although the hypothesis is accepted, generalizations must be made with caution.

Additionally, the connection between the independent and the dependent variables is characterized as an nth-degree polynomial in x in polynomial regression. The fitting of a nonlinear connection between the value of x and the conditional mean of y is described by polynomial regression. It was typically equivalent to the least-squares approach. According to the Gauss Markov Theorem, reduces the variance of the coefficients (Lindgren et al., 2011). This is a form of Linear Regression in which the dependent and independent variables have a curvilinear connection, and the data is fitted with a polynomial equation (Boryga & Graboś, 2009). In this case, the dependence between input and output variables can be described by a higher degree polynomial.

The statistical correlations discovered in Section 3.1 were used to construct the FCM. Figure 2 illustrates the FCM model that has been developed with the assistance of a Mental Modeler. Blue lines represent positive correlations and red lines the negative ones. The thickness of a line represents how strong the correlation is. The plus sign indicates a positive connection between the two measurements, whereas the minus sign indicates a negative correlation. This approach was utilized in this article since it has been widely included in previous research on Search Engine Optimization (SEO) and Search Engine Marketing (SEM) (Giabbanelli, Gray & Aminpour, 2017; Sakas et al., 2022b). For instance, as indicated in Fig. 2, User Engagement can be increased by a rise of organic traffic and can decrease by the percentage of the users that exit the page (bounce rate).

Fig. 2.

Fuzzy Cognitive Map illustrates the correlations between the examined metrics (mentalmodeler.com, accessed on 14 December 2022).

(0.33MB).

4.2.1Adoption of fuzzy cognitive map scenarios to analyze the data

Following the creation of the FCM model, 5 scenarios were conducted to evaluate the projected discrepancies in the KPIs of logistics startups' websites throughout various stages of the human interaction with the website. The hyperbolic tangent function was adopted for all cases since provides the ability to illustrate values that can be negative and fall inside the range (−1, 1) (Kokkinos, Lakioti, Papageorgiou, Moustakas & Karayannis, 2018). Figures 3,4 and 5 illustrate three optimization scenarios. Fig. 3 illustrates, after a lot of trials, the User Engagements optimization scenario for the logistics startups industry according to the extracted web analytics and the statistical analysis. To achieve 10% better user engagement and 10% more visitors, startups need to reduce the Pages per Visits metric by 36% and this can be accomplished by creating a website with fewer pages and easy to use. Additionally, an expected increase by 8% in Average Time on Site can be observed as well as an increase in the Bounce Rate by 20%. Interestingly, a decrease in the “Gross Profit” by 5% is expected and will be explained in the next section.

Fig. 3.

Engagement Optimization Scenario (mentalmodeler.com screenshot).

(0.18MB).

Fig. 4.

Website Development Optimization Scenario (mentalmodeler.com screenshot).

(0.21MB).

Fig. 5.

Profit Optimization Scenario (mentalmodeler.com Screenshot).

(0.15MB).

Figure 4 illustrates a website development optimization scenario. More specifically, examined what will happen if the website developers manage to develop a user-friendly website, with less information, that reduces the Average Time on Site by 25% and increase Organic Traffic by 5%. The bounce rate will increase by 15% which is expected since when more visitors access the website increase the possibility of exiting. Also, a decrease in Pages per Visit by 9% can be observed. This scenario aligns with previous research (Krrabaj, Baxhaku & Sadrijaj, 2017; Pakkala, Presser & Christensen, 2012), it illustrates that the Google algorithm takes into consideration the total Pages per Visit when constructing the rankings. Developers should consider it while developing a logistics startups’ website since there is no need to make it complicated.

Finally, Fig. 5 illustrates a Profit Optimization Scenario. More specifically, examined how the logistics startups will gain 5% more gross profit. According to this scenario, in order to accomplish this goal, startups need to place advertisements in search engines and social media by 7% and 4% accordingly. Additionally, increases in user engagement can be observed by 2% as well as in brand name by 15%.

ABMs provide double attention to the user behavior inside the platform as well as the system operations overall (Barbati, Bruno & Genovese, 2012; Bonabeau, 2002). The Agent-Based Model depicted in Fig. 6 is built up of unique features that interact inside a “cause-and-effect” system to provide an outflow of correlations (Aguilar, 2005). The application of ABM provides the ability for the monitoring of both user behavioral data (User engagement) and their effects on the examined website (Global rank, profit) (Barbati, Bruno & Genovese, 2012). As a consequence, various findings of the correlations from these interactions are produced. Because ABM emphasizes the presence and discovery of connections between distinct system elements, this developmental method prioritizes the identification of conditions that impact the agents’ behavior inside the model (Sakas & Reklitis, 2021a). As a result, the adoption of Agent-Based Models enables businesses to completely understand the insights supplied by data in terms of user interaction with their websites as well as opportunities for expansion (Giabbanelli, Gray & Aminpour, 2017). The system Anylogic 8.7.9 with the computer language “Java” has been used to create the entire model.

Fig. 6.

ABM for the optimization of logistics startups’ User Engagement, Global Rank, and gross profit through Digital Marketing and SEO.

(0.73MB).

The ABM visualizes the impact of display advertisements on visitors and their interactions with the logistics startups ‘websites. The circular blocks depict the agent's initial status before any move. Movements among the blocks are depicted with the arrows on the ABM, and such movements come from changes to the required circumstances dictated on the bottom of the picture in Fig. 3. The Poisson distribution has been adopted for this study for two main reasons (Consul & Jain, 1973). The specified variables for the ABM were implemented with the poison distribution and are the outputs of the statistical analysis illustrated in Section 4.1.  Additionally, the findings from the statistical correlations illustrated above are applied in the ABM for the agents to travel from block to block.

The model on the Fig. 6, initially depicts potential users in the upper center block, who are separated into two groups. Users that accessed the logistics startups’ websites through a paid advertisement are divided into “Paid Traffic” (e.g., google ads.) and “Social Traffic” (social media advertisements, e.g. Facebook ads.). On the other hand, a website can generate users directly from a search engines search depicted as “Search Traffic” and from “Referral Traffic” which are users that entered a website through a link from another website. Next, all those types of traffics constitute the general “Organic traffic” of the web page. On the top right-hand side “Bounce rate” can be identified. “Bounce rate” is the percentage of the visitors that exit from the website without searching anything else. After a user enters the website (“Organic Traffic”) produce some behavioral analytics are divided into “Average Time on Site”, “Pages/Visits” and “Total Visitors”. All those parameters are constituting the “User Engagement” parameter above. Continuously, the model agents move to the next block and produce the “Global Rank” before they exit the website. Finally, inside the “Global Rank” block runs the procedure of “Gross Profit” with a triangular calculation method (Anylogic, 2022). Those variables are the outcome of the statistical analysis of the web analytics and the behavior of the users in the websites. The Agent-based model produces an agent population distribution (Fig. 7a and b) and two time charts (Fig. 8a and b) that present the fluctuations in logistics websites’ User engagement parameters in correlation to web ranking and profit as a result of changes in paid and nonpaid traffic.

Fig. 7.

(a) Experiment's population allocation with 1000 agents for a period of 180 days. Day 12. (b) Population allocation with 1000 agents for 180 days. Day 54.

(0.5MB).

Fig. 8.

(a). The time chart presents the history of the contribution of the parameters: Organic traffic, Social traffic, Paid traffic from the time that placed an advertisement for 180 days. (b). The time chart presents the history of the contribution of the parameters: Global rank, User Engagement, and Gross Profit from the time that placed an advertisement for 180 days.

(0.52MB).

Figure 7(a) and (b) depict the execution of the ABM model shown in Fig. 6. The study explores the movement of the users in the Agent-based model for 180 days period. The agents embody the potential users moving through the model until they access the ending block named “User engagement” and provide an outcome for the parameters “User Engagement” and the variables “Global rank” and “Gross Profit” in Fig. 7. The gray agents represent the users that access the website through a paid traffic source, blue agents represent the user that generated from a search traffic source. The red dots are the users that left from the website (bounce rate). The yellow agents illustrate users that contribute to user engagement, the light blue agents illustrate the users that contribute to the Global Rank and the green agents embody the users of the webpage that contribute to the increase or decrease of profit. This logical, since, once the paid advertisement campaign was launched, the website's Global Rank increased, and consumers reached the startup's website directly, as the corporate brand name grew.

Figure 8a and b illustrate the results of the predictive model. According to Anylogic the chart “displays the history of contribution of several data items into a total during the latest time horizon as stacked areas” (Anylogic, 2022). The values of the simulation results are indicated on the vertical axis. The horizontal axis depicts the period over which the simulation takes place (180 days) for the logistics startups’ websites. Figure 8a illustrates the placement of an advertisement in social media at the time “0”. We can observe that the social traffic from the advertisement increased as well as the general organic traffic of the website. When the advertisement stopped on the day “5”, the social traffic and organic traffic dropped. Additionally, on day “8” the second placement of a search engine advertisement took place and remained until the end of the simulation. As can be observed, to maintain good visibility and high organic traffic, logistics startups’ websites need to keep a steady stream of advertisements and keep them going in order to be effective in contrast to the FinTech industry websites (Bapat, 2018; Tien, Cheng & Pei-Ling, 2018).

Figure 8b illustrates the results of the main examined metrics after the placement on an advertisement in the day “0”. There is a sudden spike in global rank that remained until the day “120” which decreases. A decrease in global rank is beneficial since a website ranking in 10th place is better than a website ranking in 100th place. Additionally, an increase in user engagement can be observed after day “8” on the model and correlates with a placed advertisement in social media as can be observed in Fig. 8a. As can be observed on the day “120” with the sudden drop of Global Rank, a brand name has been created. Finally, there is no correlation between Engagement and the Brand name with the profitability.

Marketers and developers need to create easy-to-use and dedicated to the result websites in order to enhance engagement. For instance, the customer must have easy access to tracking numbers or booking parcels for delivery, which will lead to the creation of brand loyalty (Zheng, Cheung, Lee & Liang, 2015). This is crucial because the needs of logistics websites customers are diametrically opposite from other websites such as online clothing stores. For instance, in online clothing stores, such as Zara and Mango, the importance is located on creating different and strong emotions to maximize the user experience, which will lead to a sale (Goldsmith & Flynn, 2004). Consequently, it is crucial for a logistics startup website to place paid advertisements both on google and social media regularly to maintain high levels of visibility.

Decision support systems and simulation models enable continuous monitoring of customers' digital behavior across real-time market data, with the goal of eliminating digital marketing inefficiencies and improving user engagement. For instance, digital marketing metrics (global ranking and gross profit) have a beneficial impact on traffic-related Key Performance Indicators (organic and paid traffic) and behavioral-related KPI metrics (pages per visit). In this logic, marketing efforts should be directed toward paid ads and targeting specific keywords in order to improve profit and brand recognition; which will lead eventually to the sustainability of a startup company since the vast majority of them fail in the first year of operation (Shlomo & Maital, 2021). Furthermore, resources need to be allocated to producing unique content in order to increase user engagement (Byun et al., 2020). Big data-powered trigger-based customer journeys help businesses to foresee potential developments and concentrate on delivering exceptional digital experiences (Terragni & Hassani, 2019). Additionally, in contrast to other industries, which can create advertisements seasonally (Zheng et al., 2015; Fitz-Gibbon, 1990), logistic startups require a consistent stream of advertisements in both social media and search engines in order to maintain and develop a digital brand name and customer loyalty.

Digital advertising campaigns; as a part of a general digital marketing strategy, are centered on boosting logistics startups’ websites visibility, resulting in higher corporate profit. Paid marketing messages, such as banner ads, are an essential aspect of digital advertising; and this research focused specifically on paid ads from search engines as well as from social media and their effects on startups’ websites user engagement. Additionally, the findings of this study highlight the necessity of monitoring and analyzing the web analytic metrics for the optimization of corporate brand name through the global rank metric (Sakas & Reklitis, 2021a; Drivas, Kouis, Kyriaki-Manessi & Giannakopoulou, 2022).

Finally, we suggest that in order to establish their company in the market, startup marketing managers should monitor changes in their web pages' organic traffic and global rankings in relation to user engagement metrics and profit. As a consequence, decision-makers will have a better view of advertising campaigns for their products, and services. The current study's findings lend credibility to various big data research on the digital marketing process of logistic startups. Behl et al. (2019) have published a study expressing the valuable potential of big data on marketing capacities in favor of startups. The authors emphasize the startups’ ignorance of big data capabilities, which reduces the effectiveness of a big data-driven digital marketing approach in startups. Furthermore, the current study aligns with the Cavicchioli and Kocollari (2021) and Roumeliotis et al. (2022) studies, in which important key marketing metrics are investigated, resulting in efficient digital marketing strategies, such as increased lead generation and conversion rates. Finally, this study highlights that customers' behavioral web and social analytics, on the other hand, can provide added value to startups' digital marketing strategies.

The findings of the present research highlight the necessity of logistics startups to devote more resources in understanding the behavioral patterns of digital customers using big data analytics. By doing so, logistics startups can be able to follow existing trends and identify the aspects that significantly influence customers' online purchase behavior based on their web and social media engagement (Negedu & Isik, 2020). The integration of behavioral analytics into concrete digital advertising strategies is a tough assignment that involves numerous components, including the behavioral KPIs (Saura, Palos-Sánchez & Cerdá Suárez, 2017).

More specifically, marketers and decision-makers must extract, evaluate, and utilize web analytics and big data to restructure the startup's marketing strategy. Startups may advantage from the proper execution of digital marketing strategy since advertising and campaigns allow them to contact potential clients more quickly (Tardan, Shihab & Yudhoatmojo, 2017; Shlomo & Maital, 2021; Mariani & Fosso Wamba, 2020).

Additionally, in order to construct a user-friendly website that suits their business purpose, website developers need to obtain a thorough understanding of the implications of their website's user engagement on gross profit and digital brand name. Furthermore, with the knowledge gained from big data, they could contribute to a greater degree to the startups' operations. Finally, we suggest that in order to establish their company in the market, startup marketing managers should monitor changes in their web pages' organic traffic and global rankings in relation to user engagement metrics and profit. As a consequence, decision-makers will have a better view of advertising campaigns for their products, and services.

The vitality of a startup is based on reaching new consumers and maintaining the existing ones. The first parameter can be accomplished with digital advertising; the second one is the adoption of a well-designed and user-friendly website. In order to develop an effective website, the user engagement analytics must be examined in correlation to the technical parameters of a website (Sakas et al., 2022b). Since technical parameters examination are crucial for identifying a company's website usability; their investigation in correlation to behavioral analytics can provide valuable insights to marketers. It could be interesting for instance, to identify the effects of a slow-loading startup's webpage on the purchase rate or the correlation between the total webpage size with the total parcel search time as studied before in the agri-logistics sector (Sakas & Reklitis, 2021b). Additionally, it would be beneficial to examine the effects of the logistics startups' social media on the overall user experience and interactivity of their websites.

Moreover, since a correctly placed advertisement is crucial for maximizing market share (Lin, Paragas & Bautista, 2016; Grewal, Bart, Spann & Zubcsek, 2016) and the usage of emotions in ads promotes the brand name (Hutchins & Xiomara, 2018), it could be beneficial to examine the effectiveness of the advertisement with the assistance of neuromarketing tools. For instance, it could be beneficial to examine where the customers focus on a website in order to make it easy to use and user-friendly as studied in other industries' websites and social media (González-Mena et al., 2022). Additionally, it could be examined if a well-designed logistics startup website can reduce customers' frustration and anger while searching for a parcel with the assistance of an EEG. Eye-tracking could be used to identify where the customer looks frequently to place the advertisements and EEG could be used to illustrate a consumer's emotional reaction to a specific advertisement (Vences, Díaz-Campo & Rosales, 2020).

Finally, the first limitation relies on the quantity of the examined websites. More websites could be examined to acquire better insights into the sector. Secondly, the data gathered from the ``SEMrush'' data extraction webpage. More, behavioral data could be beneficial in order to examine the logistics startups' user engagement in depth. Additionally, social analytics can be extracted and analyzed in correlation to behavioral and technical factors in order to get a holistic approach to the logistics startups' digital entities. Further research can be held with the assistance of interviews and questionnaires for a more holistic qualitative and quantitative approach (Reklitis et al., 2017; Loor-Zambrano, Santos-Roldán & Palacios-Florencio, 2022). Finally, the research is limited to the logistics startups sector, further research to other startup sectors could be beneficial. The interviewees' opinions extracted from questionnaires in relation to their actual behavior on a website as extracted from web and social analytics could provide valuable insights to marketers, managers, and startup owners.