THE IMPACT OF SOCIAL MEDIA PROMOTION AND SERVICE QUALITY STRATEGY TOWARD GO-FOOD ONLINE DELIVERY

How to cite this paper: Situmorang, J. R. (2023). media social of The impact strategy quality service promotion and Go toward - delivery Food online [Special issue] Strategy Business .


INTRODUCTION
Since March 2020, the COVID-19 pandemic has plagued Indonesia.Unlike several other countries that imposed lockdowns, the Indonesian government only restricted community activities.These restrictions aimed to reduce the spread of COVID-pe 19 by requesting their limit to ople placed government activities.
The outdoor community activities, on several restrictions 1) including by workplaces/offices limiting 75% and home by from work implementing the protocols; health stricter of application 2) shifting teaching and learning activities online; 3) allowing essential sectors related to basic needs to operate with the application of health protocols and limited operating hours; and 4) regulating the imposition of restrictions on restaurant activities of 25% for dine-in and delivery services according to operating hours (Kominfo, 2021).Therefore, with restrictions like this, people prefer online services to buy food and drinks.
Even before the pandemic, online services for ordering food and drinks were popular in Indonesia.One of them is Go-Food, an online food delivery service established in 2015.According to Redaksi (2022), Go-Food has been named the most useful and user-friendly application in the world during the COVID-19 pandemic.Go-Food, an innovation from the Gojek application (or online driver service), not only provides food delivery service but also helps micro, small, and medium-sized enterprises (MSMEs) become better known by the public.For MSMEs who want to use delivery services but do not have the budget to recruit people, Go-Food can be a solution (Widyayanti & Insiatiningsih, 2021).
One feature of Go-Food that is different from other food delivery service applications is that Go-Food allows consumers to place several orders at different places at one time.Go-Food also often provides discounts in the form of discounts and free shipping.In addition, Go-Food has more fleets, making it easier for customers to place orders (Kusindriani, 2023).Those various facilities and benefits obtained from Go-Food will affect purchasing decisions and the brand image of the Go-Food brand.According to Kotler and Armstrong (2014), purchasing decisions are individual activities that directly involve making decisions to purchase the products.
The method used by Go-Food in influencing purchasing decisions is through the quality of service and promotion on social media.Service quality is seen as an essential part to be provided by companies as it affects the acquisition of new customers and can reduce the probability of old customers switching to other companies.Another way to increase purchases is to run promotions.Promotion is a form of marketing communication designed to disseminate information, influence/ convince and remind a company and its product's target market to accept, buy and be loyal to what the company offers (Tjiptono, 2014).This promotion can be assisted by social media, where social media has developed over the last decade and has become a driving force for obtaining and disseminating information (Stieglitz et al., 2018).Social media's effect on consumer behaviour includes various activities ranging from informing and sharing ideas and attitudes to acquiring awareness, understanding, and visualizing post-purchase behaviour without purchasing (Tatar & Eren-Erdoğmuş, 2016).Besides being a factor in purchasing decisions, service quality and promotions on social media can also enhance a brand's image.Brand image is a people's perception of a brand formed from information and experience they had before (Setiadi, 2013).With the image of a brand, it will encourage potential buyers to buy or use the service.Meanwhile, old buyers continue to shop or use these services.Therefore, a company will get customer loyalty.
Several studies have been done on online food delivery applications.Research by Abderahman and Anggraeni (2020) examines that price, promotion, and brand image positively influence the decision to use a service.In a similar vein, Ulam's (2019) study on the effect of the price, found that price only influences 11.7%, while the remaining 88.3% is influenced by other variables.In line with Aptaguna and Pitaloka's (2016) research, price does not significantly affect the use of services, but the service quality is the main customer's consideration in choosing a service.Thus, this study aims to determine the effect of service quality and promotion on social media on customer loyalty, with the brand image as an intervening variable for Go-Food services.
The paper is structured as follows.Section 2 presents a detailed literature review to identify existing research and hypothesis development.Section 3 analyses the methodology that has been used to conduct empirical research on Go-Food delivery application.The primary findings are presented and discussed in Section 4 and Section 5, respectively.Finally, Section 6 draws conclusions from the study.

Service quality (SQ)
There are two types of service quality (SQ): technical and functional.Technical quality indicates the quality of service goods; such as what consumers buy.Technical quality also describes the technical identifications and standards of service.However, functional quality is a step to distributing the service goods and how is the correlation between service providers and their customers.Technical quality can be in the form of employees' technical skills and knowledge, the quality of specialized solutions provided, and computerized systems and machines.
Meanwhile, the available rate includes behaviour, attitude, accessibility, appearance, customer contact, internal relations, and service thinking (Akhtar et al., 2011).The service quality component depends on active customer service for all services a service or product company offers.Service quality is essential in building a competitive company and increasing its competitiveness.Service quality has drawn significant attention due to its significant influence on corporate performance, customer happiness, loyalty, and profitability.The relationship between service quality and customer loyalty has been studied by Makanyeza and Chikazhe (2017), Özkan et al. (2019), and Uddin (2019).

Promotion at social media (PSM)
Social media has become a new communication medium through which customers can exchange information, produce content, and generate ideas.Many businesses now use social media as part of their marketing plans (Zhang et al., 2017).Each forum has unique qualities due to marketers' increasing usage of social media platforms to connect with and engage with consumers.However, when used as a promotional mix component, each platform's features have a synergistic effect (Oh et al., 2017).Most businesses have social media integration to communicate and improve customer relationships.Social media is an eminent component of social commerce activities, which includes various tools, such as user reviews, ratings, and recommendations, as well as online communities, forums, and shopping groups (Hajli, 2015).Intensive communication creates a close relationship between the company and the customer.In addition, it can increase customer loyalty (Ajina, 2019; Yadav & Rahman, 2018).

Brand image (BI)
Brand image (BI) is defined as an important part that encourages a brand to differentiate its products from the competitors (Rua & Santos, 2022).Brand image is made up of a consumer's knowledge and beliefs about the brand's various products and is interpreted as the personal symbolism of a brand (Iversen & Hem, 2008).It can help customers differentiate and determine which brand suits them.This brand image is consumers' perception and evaluation.Cho and Fiore (2015) implied that the dimensions of a brand image consist of three different customer associations.First, customers' personal opinions, attitudes, and judgments about a brand are included in the cognitive connections.Second, the emotional associations encapsulate the feelings and emotions customers develop toward a brand.Third, the sensory associations reflect the physical senses.These three customer associations are formed during brand-customer interactions.
For this reason, businesses today put a lot of effort into maintaining their brand image and spend a lot of money on creating names that have a good reputation (Shankar et al., 2008).When customers are drawn to a brand's image, that brand's communications have a greater impact than those of rival brands.Therefore, a buyer's behaviour is significantly influenced by brand image (Burmann et al., 2008).Arslan and Altuna (2010) stated marketers could utilize a variety of marketing techniques, including the product, the brand name, the colours, the logo, the packaging/labelling, point of purchase promotions, commercials, retailers, and all other sorts of promotions, pricing, place of origin, brand owner, target market, or product users.There is plenty of proof that client loyalty is highly influenced by image and reputation (Hardaker & Fill, 2005).Brand image is related to recent consumer experience.The better the idea, the higher its quality, value, satisfaction, and customer loyalty.The brand identity is tailored to the needs and desires of the target market.It facilitates brand loyalty from the customers (Ogba & Tan, 2009).

Customer loyalty (CL)
From a behavioural perspective, customer loyalty (CL) is defined as the proportion of consumers who use particular services from a given category more frequently than the total number of services offered.Meanwhile, the attitudinal perspective sees customer loyalty as the customer's tendency to continue a relationship with a service provider (Zeithaml & Bitner, 2000).Lenka et al. (2009) explained loyalty behaviour as a strong commitment possessed by customers to buying or using a product/service.Loyalty is commonly obtained through behavioural measures rather than attitudinal measures.The predisposition to be loyal and the level of commitment to purchasing a particular brand are two different aspects.Attitudinal measures focus on consumers' feelings about the product, whereas behavioural estimates are based on the responses of the promotional.
Based on the explanation of the relationship between service quality, promotion at social media, brand image, and customer loyalty, the researcher proposes the following hypotheses:

RESEARCH METHODOLOGY
This study used quantitative methodology to analyse the data.The respondents were people who have the Gojek application installed on their phones.Owners of the Gojek application automatically have the Go-Food application integrated into their Gojek platform.The researcher drew a large sample that met the criteria for structural equation modeling (SEM).The study included 400 Gojek application owners selected through the following question: "Have you been actively ordering food and drinks through the Go-Food application since October 2021?"If the answer was affirmative, participants were invited to complete a questionnaire.Questionnaires were distributed through various WhatsApp groups (WAGs).WAGs targeted are groups with a minimum number of 50 people so that the target of 400 respondents can be achieved.The WAGs targeted included five groups of university alums, five groups of community communities, and three groups of family/relative communities.The distribution and collection of the questionnaires lasted for three months, from February 1, 2022, to April 30, 2022.
The questionnaire consists of four sections according to the number of research variables.The SQ variable comprised 16 items adapted from Parasuraman et al. (1988).The PSM variable consisted of five items adapted from Rapp et al. (2013).The BI variable comprised five items adapted from Salinas and Pérez (2009).The CL variable included five items adapted from Amoako et al. (2017).A five-point Likert scale, anchored from 1 (strongly disagree) to 5 (strongly agree), was used to measure each measurement item in order to convey the level of agreement.
The research methodology used in this study can be replicated by other researchers to analyze the impact of service quality and promotion at social media towards customer loyalty in other industries.

Convergent validity test
Convergent validity has criteria for a correlation coefficient value above 0.50.Based on the SmartPLS 3.0, the data of the study showed that all loading factors are worth greater than 0.5 that shows that each construct has good validity.After that, the average variance extracted (AVE) test was carried out to strengthen the convergent validity results with the criterion of AVE > 0.5.The results of the AVE test are presented in Table 1.The results in Table 1 show that there are latent variables with an AVE value of less than 0.5, including responsiveness and service quality.It suggests that the latent construct indicators have inadequate construct validity.Considering the need for the formation of latent constructs and all indicators having a good category loading factor, the latent having an AVE of less than 0.5 can still be maintained.

Discriminant validity test
It is possible to see discriminant validity tests from the cross-loading value.The connection between indicators and other constructs must be smaller than the correlation between indicators themselves.It is clear from comparing the correlation between latent constructs and the AVE square root.The latent construct has excellent discriminant validity if the square root value exceeds the correlation between latent constructs.The outcomes, conducted using SmartPLS 3.0, are shown in Table 2.

Reliability test
The reliability coefficient, known as Cronbach's alpha, measures how positively correlated the items in a set are.It can be valid if Cronbach's alpha is 0.7 or higher.Due to Cronbach alpha's population limitations, adopting a different internal consistency reliability measure is recommended.The composite reliability is generally interpreted similarly to Cronbach's alpha.For exploratory research, composite reliability scores between 0.60 and 0.70 are acceptable.Values between 0.70 and 0.90, however, can be regarded as good.Values greater than 0.90 are undesirable as they suggest that all indicator variables measure the same occurrence.A lack of internal consistency dependability is shown by composite reliability ratings of less than 0.60.The following presents the reliability test results using the SmartPLS 3.0 software.

F-square
F-square is the effect size used to determine the effect of latent variable predictors at the structural level.An effect size ≥ 0.02 is considered small; ≥ 0.15 is considered medium; ≥ 0.35 is considered large.Based on the test results with SmartPLS 3.0 software, the results can be seen in Table 5.Table 5 demonstrated that the PSM variable has a big effect size on BI, whereas the SQ variable has a moderate effect size.The SQ and PSM variables have a small effect size on CL, while BI has a large effect size on CL.

Q-square
If the Q-square value is greater than zero, it determines the predictive importance of the endogenous constructs.It shows that values are generated properly, and the model is predictively relevant.The Q-square value obtained using the R-square value can be seen in Table 6.Table 6 indicates the Q-square is greater than zero, which means the observed values have been reconstructed properly so that the model has predictive relevance.The relative impact of the structural model on observational measurements for endogenous latent variables can also be seen using the Q-square value.It means that 0.761 or 76.1% of the relative impact of the structural model on observational measures for endogenous latent variables and as much as 23.9% is an error.

Hypotheses test
In the assessment of significance and predictions, hypothesis testing is carried out through the path coefficient and t-value.The t-table value with a 95% confidence level, two-tailed, and a total of 400 respondents is 1.96.If the t-statistic value is bigger than 1.96, then the null hypothesis (H 0 ) is rejected, and H1 is accepted, but if it is smaller than 1.96, then H 0 is accepted and the first hypothesis (H1) is rejected.The significance value among the variables is presented in Figure 1.From Table 9 above it is seen that the path coefficient of the original sample estimate (beta) of the SQ variable is positive (0.454), indicating that the direction of SQ's influence on CL through BI is positive or unidirectional.It means that if SQ increases, CL through B) will increase, and vice versa.The effect of SQ on CL through BI was significant in the 2-tailed test (t-table = 1.96) with a t-statistic value of 13.590 greater than the t-table and a p-value of 0.000 less than alpha 5% (0.000 < 0.05).Thus, H3 that SQ has a partially significant effect on CL through BI is accepted.

The influence of service quality, promotion at social media, and brand image on customer loyalty
The service quality variable's original sample estimate (beta) path coefficient is 0.744.It indicates that the direction of service quality's influence on CL is positive, meaning that if service quality increases, CL will increase, and vice versa.The effect of service quality has a significant effect on CL.These findings align with the research results by Makanyeza and Chikazhe (2017), which found that service quality, customer satisfaction, and brand image positively affect CL.It implies that the better the service quality, the more loyal customers are.
The path coefficient of the PSM variable's original sample estimate (beta) is 0.077, indicating the positive influence between promotion on social media and CL.If a promotion on social media increases, CL will increase, and vice versa.The effect of promotion on social media has no significant effect on CL.This result contradicts the research by Yadav and Rahman (2018), who found that social media marketing activities (SMMA) positively influence CL.Similar to Ajina (2019), the results imply that social media engagement plays a significant role in enhancing the loyalty of consumers of banking organizations.
The BI variable's original sample estimate (beta) path coefficient is -0.008.It indicates that the direction of influence between the BI on CL is negative or reversed, meaning that if BI increases, CL will decrease, and vice versa.The effect of BI has no significant effect on CL.This finding differs from Mahothan et al. (2022) study, which indicates that the better the BI, the higher the CL.

The influence of service quality and promotion on social media on brand image
The path coefficient of the SQ variable's original sample estimate (beta) is -0.044, indicating that if service quality increases, BI will decrease, and vice versa.The effect of service quality does not have a partially significant effect on BI.This finding differs from Hassan and Salem (2021), which service quality directly affects private label brand (PLB) image but does not affect the purchase intention.
The path coefficient of the PSM variable's original sample estimate (beta) is 0.611, with a positive sign, meaning that if promotion on social media increases, BI will increase, and vice versa.The effect of PSM has a partially significant effect on BI.This finding aligns with the results of research by Bilgin (2018), which states that SMMA have been seen as having a significant effect on BI.

The influence of service quality, promotion at social media toward customer loyalty through brand image
The path coefficient of the PSM variable's original sample estimate (beta) is -0.033 with a negative sign indicating that the direction of PSM's influence on CL through BI is negative or reversed, meaning that if PSM increases, CL through BI will decrease, and vice versa.The effect of PSM on CL through BI was not significant.Thus, PSM does not partially affect CL through BI.

CONCLUSION
This study aimed to examine the effects of service and promotion on social media on customer loyalty when using Go-Food delivery services, with brand image as the intervening variable.The results showed that promotion on social media and service quality significantly influence customer loyalty through their impact on brand image.Unlike service quality that positively influences customer loyalty, promotion on social media did not affect customer loyalty because the other online food delivery service also gives promotions.Thus, customers can choose which services they want to use based on the promotion given.Whereas brand image also does not influence customer loyalty, even if Go-Food is chosen as the most used online food delivery application, it cannot be said that brand image is the factor that affects customer loyalty.Therefore, in conclusion, the factor influencing the customer loyalty to using Go-Food as an online food delivery service is the service quality.
However, this study also has some limitations, such as the study only focusing on the Go-Food online delivery service, which may limit the generalizability of the findings to other online food delivery services or industries.The study only considers the brand image as an intervening variable, while other variables such as price, convenience, and product quality may also affect customer loyalty.The study relies on self-reported data from a small sample size of 400 active respondents, which may not be representative of the entire population of Go-Food users.The study only uses quantitative data analysis, which may not capture the full complexity of customer loyalty and the impact of service quality and promotion on social media.The study does not consider the impact of external factors such as competition, economic conditions, and cultural differences on customer loyalty.The practical implications of this paper can help to allocate their resources effectively by focusing on the factors that have a significant impact on customer loyalty.It also helps to improve their services and retain customers, which can lead to increased profitability and growth.
Further investigation is needed to understand the specific dimensions of service quality that have the greatest impact on customer loyalty in the context of online food delivery services.

Table 1 .
Average variance extracted values

Table 2 .
Cross loading discriminant validity test values

Table 3 .
Cronbach's alpha and composite reliability values

Table 4
shows the BI variable has an R-square value of 0.361, meaning that SQ and PSM contribute 36.1% to BI in the moderate category.At the same time, the remaining 63.9% is the influence of other factors that are not observed.The R-square for the CL variable is 0.626, which means that SQ, PSM, and BI contribute 62.6% to CL in the moderate category.While the remaining 37.4% influences other factors not observed in this research.

Table 7
above shows the effect of SQ, PSM, and BI to CL both simultaneously and partially.Simultaneously, an R-square of 0.626 indicates that SQ, PSM, and BI contribute 62.6% to CL, while the remaining 37.4% influenced by other factors that are not observed.The effect of SQ, PSM, and BI toward CL is greater than the t-table (20.849 > 1.96), and a p-value of 0.000 which is less than alpha 5% (0.05).Thus, H1 that SQ, PSM, and BI simultaneously significantly affect CL is accepted.

Table 8 .
Path coefficient, R-square, t-count, and p-value: SQ and PSM  BI

Table 8
shows the effect of SQ and PSM on BI both simultaneously and partially.Simultaneously, an R-square of 0.361 indicates that SQ and PSM contribute 36.1% to BI, while the 63.9% is the influence by other factors that are not observed.The effect of SQ and PSM on BI showed greater than the t-table (8.815 > 1.96), and a p-value of 0.000 which is less than alpha 5% (0.05).Thus, H2 that SQ and PSM simultaneously significantly affect BI is accepted.