Evaluation of WeMedia Ads Marketing Based on the Demand of Internet Economy

,


Introduction
In the context of Internet economy, online trading platforms provide numerous small businesses with entrepreneurial and employment opportunities and offer convenient consumption approaches and choices to all citizens [1][2][3][4][5]. As an indispensable part of the modern market economy, the WeMedia marketing channels (e.g., live broadcast, WeChat, forums, short videos, and Weibo) on Chinese online e-commerce platforms achieved a turnover of more than 100 billion yuan [6][7][8][9][10]. To find the root causes for consumers' trust and preference for WeMedia marketing, it is important to deeply explore the features, merits, and profit factors of the marketing model of WeMedia marketing platforms. e exploration results help to propose countermeasures that update and optimize the marketing model for better profit and performance.
To adapt to the latest trend of the market pattern, each brand needs to rely on the new model of WeMedia marketing to expand the audience and look for new ideas and growth points for product sales [11][12][13][14][15][16][17]. Based on the theories of online marketing and online opinion leaders, Kollintza-Kyriakoulia et al. [18] determined the evaluation indices for the performance of the trendy live broadcast marketing and constructed a system of 24 indices in five aspects, namely, marketing amount, after-sale service, marketing value, platform performance, and social benefit. To help policymakers and media platform operators promote the development of online media industry, Kim [19] demonstrated the influence of three factors, namely, matching technology, consumer strategy, and advertising technology, on network scale and social welfare and provided helpful suggestions on how to prevent media platform operators from abusing their power. Wang and Gao [20] deeply analyzed the inadequacies of WeChat WeMedia marketing models in precision marketing, brand perception, and interestingness through questionnaire survey and indepth interview and discussed the relationship between WeChat WeMedia platform, consumers, suppliers, and advertisers.
WeMedia differs greatly from the traditional media in the marketing model. To make more profits, it is necessary to sort out and summarize the features, functions, and development levels of WeMedia ads marketing [21][22][23][24][25][26]. Chen et al. [27] detailed the connotations of integrated marketing communication theory, combined the theory with WeMedia ads marketing, and modeled the factors affecting the consumption willingness under WeMedia ads marketing. Brekke and Nilssen [28] carried out a mathematical analysis on the consumption data under WeMedia ads marketing, compared the results of the analysis with those obtained by the theoretical model, and concluded that WeMedia ads marketing strategies affect the marketing channels and contents in four dimensions: interestingness, interactivity, practicality, and economic benefits. Taking short video WeMedia as the objects, Mujahid [29] reviewed the impact of green WeMedia marketing on the organizational performance of small enterprises and discussed how the integration of learning orientation in organization's internal culture determines the influence of green strategies on corporate value improvement.
e existing studies at home and abroad rarely quantify WeMedia ads marketing. Rather, most of them are descriptive analyses on the relevant concepts, the importance of WeMedia ads marketing, product quality monitoring, and the reasonability of marketing models. ere is a lack of quantitative results on deeper layers. To make up for the gap, this paper attempts to evaluate WeMedia ads marketing based on the demand of Internet economy. e main contents of this work are as follows: setting up a detailed evaluation index system (EIS) [30][31][32] for WeMedia ads marketing, which covers three aspects: online marketing performance, supplier performance, and relationship benefit; determining the weight of each evaluation index through multilevel fuzzy comprehensive evaluation (FCE) [33]; building effect evaluation models and performing confirmatory factor analysis (CFA) [34]; and verifying the effect evaluation models by fitness test and proving the correctness and effectiveness of the model through experiments.
Starting with the online transactions in the context of the development of the Internet economy, this paper discusses the influence of we-media on ad marketing performance, with the aim of laying the basis for decision-making of small merchants facing the demand of the Internet economy, as well as small and medium-sized enterprises amidst the everchanging we-media marketing landscape. e research also attempts to provide a reference for industry-wide we-media marketing.

EIS Construction
e EIS for WeMedia ads marketing should be constructed by the specific-measurable-achievable-relevant-time-bound (SMART) principle. To realize the goals, this principle must be adhered to throughout the design of the performance evaluation indices for WeMedia ads marketing, making the EIS more reasonable and effective. Drawing on the literature and theories on Internet economy, this paper sets up an effective EIS for small merchants as well as small and medium-sized enterprises, which implement WeMedia ads marketing. Covering three primary aspects (i.e., online marketing performance, supplier performance, and relationship benefit), the EIS paves the way for the subsequent expert survey, factor analysis, and model construction.
e EIS covers three main dimensions, namely, online marketing performance, supplier performance, and relationship benefit. Based on the demand of Internet economy, the EIS for WeMedia ads marketing should reflect the effect, economic benefit, and social benefit of the current marketing model and be quantifiable, available, and representative. Under these guidelines, the specific EIS was constructed as follows: In terms of online marketing value, WeMedia exposure MP 121 refers to the number of exposures on top search; mainstream media exposure MP 122 refers to the number of exposures on media above the provincial level; and online search frequency MP 123 refers to the online trend index.
In terms of supplier sales performance, sales volume MP 211 refers to the number of purchased commodities; sales volume growth rate MP 212 is calculated by (the number of sold commodities in the current month/that in the previous month-1) × 100%; sales amount MP 213 refers to the specific turnover; and sales amount growth rate MP 214 is calculated by (the sales volume in the current month/that in the previous month-1) × 100%.
In terms of consumer source concentration, the number of source regions MP 221 refers to the number of cities for order placement; maximum regional sales MP 222 refers to the highest sales among all consumer markets; and minimum regional sales MP 223 refers to the lowest sales among all consumer markets.
In terms of after-sale service, return or refund rate MP 231 is calculated by the number of returned commodities, which are sold in the current month/the number of commodities sold in the current month; positive review rate MP 232 is calculated by the number of positive reviews on the commodities sold in the current month/the number of commodities sold in the current month; logistics time MP 233 refers to the mean duration from shipment to reception; and after-service evaluation MP 234 refers to the consumer evaluation of product quality.
In terms of consumer relationship benefit, the number of reviews MP 311 refers to the number of pushed articles or videos, or live broadcast links being reviewed; the number of sharing MP 312 refers to the numbers of pushed articles or videos, or live broadcast links being shared; the number of likes MP 313 refers to the number of pushed articles or videos, or live broadcast links being liked; and the number of forwarding MP 314 refers to the number of pushed articles or videos, or live broadcast links being forwarded.
In terms of enterprise participation, the number of online operators MP 321 refers to the number of online operators that engage in ads marketing through pushing articles or videos, or live broadcasting, and the number of suppliers MP 322 refers to the number of suppliers that engage in ads marketing.

Multilevel FCE Model
rough the analytic hierarchy process (AHP) [35,36], this paper determines the weight of each evaluation index for WeMedia ads marketing. Based on the demand of Internet economy, the complex problem of WeMedia ads marketing was interpreted and parsed to reveal the correlations between all the influencing factors. en, these factors were divided into multiple classes by the evaluation objectives. Each class contains many evaluation criteria. In this way, a model of clear hierarchy was established based on the above EIS.
After obtaining a complete, scientific, and quantifiable model, experts were invited to weigh and rate the importance of each index on a layer relative to each index on the superior layer, according to the judgement matrix formed through pairwise comparison between indices. e ratings reflect the relative importance between indices on the same layer. In the judgement matrix, each element e ab represents the quantified value of the relative importance between indices MP a and MP b on the same layer.
To assign accurate weights to indices, the first step is to normalize the elements in the judgement matrix E. e general term of the element e ab in the a-th row and b-th column can be described as where e ab belongs to the interval of [0, 1]. After normalization, the elements in E can be added by rows: e approximate solution of the eigenvector is the normalized result G(G 1 , G 2 , . . . , G M ) T obtained by formula (3). Let (E·G) a be the a-th element in the matrix formed by multiplying the judgement matrix E with the eigenvector G.
en, the maximum characteristic root of E can be calculated by To obtain scientific and reasonable evaluation results of WeMedia ads marketing, the consistency of the judgement matrix E should be tested, using the consistency index δ: e mean random consistency index ξ of the judgement matrix E can be obtained by looking up the table. e random consistency ratio η can be obtained by dividing δ with ξ of the same order: If η > 0.1, then the elements in matrix E are highly inconsistent. e matrix should be adjusted and updated to achieve a satisfactory consistency. If η < 0.1, then the elements in matrix E are highly consistent. e weight vectors can be equivalent to normalized eigenvectors.

Discrete Dynamics in Nature and Society
After the judgement matrix is prepared through pairwise comparison between indices, it is necessary to evaluate the relative importance between two indices on the same layer. N experts on Internet economy and WeMedia ads marketing were invited to fill out a questionnaire. e questions are about the analysis, weighing, and rating of the importance of each index on a layer relative to each index on the superior layer.
e valid questionnaires filled out by 20-30 experts were analyzed on the comprehensive evaluation assistance software YAAHP. Firstly, the proposed EIS was entered into the "Establishing Hierarchical Model" interface of YAAHP. en, the names, ages, majors, and titles of the N experts were imported by the "Group Decision" in the "Judgement Matrix" interface, and the weights of the experts were set to 1/N. By importing the data on the questionnaires filled out by the experts, the judgement matrix was updated according to the opinions of the experts. After the data of N experts had been imported, the random consistency ratios η were obtained for the N judgement matrices. If some or all η values are greater than 0.1, then the N judgement matrices are highly inconsistent; if all η values are smaller than 0.1, then the N judgement matrices are of satisfactory consistency.
Finally, the "Group Decision Results" was selected in the option "Weighted Geometric Average" to obtain the weights of all indices for WeMedia ads marketing. en, the weights of primary, secondary, and tertiary indices were denoted as e proposed EIS contains both quantitative and qualitative indices. According to the above judgement criteria for the goodness of WeMedia ads marketing evaluation, the qualitative indices were divided into five levels: strongly good, good, normal, bad, and strongly bad. e evaluation criteria of quantitative indices were further defined by the FCE.
3.1. ree-Level FCE. Firstly, MP abl (a � 1, 2, . . ., q; b � 1, 2, . . ., N; l � 1, 2, . . ., M) was combined with the FCE matrix E ab (a � 1, 2, . . ., q; b � 1, 2, . . ., N) composed of E abl to obtain the evaluation results of tertiary indices θ ab (a � 1, 2, . . ., q; b � 1, 2, . . ., N). When a and b are minimized, θ ab can be calculated by When a and b are maximized, θ ab can be calculated by e evaluation results of tertiary indices can be synthetized into the FCE matrix E a (a � 1, 2, . . ., q) for secondary indices. When a is minimized, E a can be calculated by When a is maximized, E a can be calculated by 3.2. Second-Level FCE. Firstly, MP ab (a � 1, 2, . . ., q; b � 1, 2, . . ., N) was combined with E a (a � 1, 2, . . ., q) to obtain the evaluation results of secondary indices θ a (a � 1, 2, . . ., q). When a is minimized, θ a can be calculated by When a is maximized, θ a can be calculated by e evaluation results of secondary indices can be synthetized into the FCE matrix E for primary indices, namely, marketing performance, supplier performance, and relationship benefit: 3.3. First-Level FCE. Firstly, MP � (MP 1 , MP 2 , . . ., MP q ) was synthetized with E to obtain the evaluation results of primary indices θ � (θ 1 ,θ 2 ,. . .,θ q ), that is, the multilevel FCE results in θ of WeMedia ads marketing based on the demand of Internet economy: e θ value is not a definite value, but a fuzzy vector. Based on the principle of maximum membership, if there exists θ R � max(θ 1 ,θ 2 ,. . .,θ q ), then the elements in the criteria set U � {U 1 ,U 2 ,. . .,U M } corresponding to θ R are the final evaluation results of WeMedia ads marketing PE.

Sample Statistics.
To judge whether the questionnaire is valid, the diversity of data was simply analyzed through descriptive statistics. Table 1 shows the descriptions and statistics of online marketing performance ratings. e mean expert ratings on the seven tertiary indices under online operation performance and online marketing value were all greater than 4, indicating that experts all have above the average perception of the indices of online marketing performance. Meanwhile, the standard deviations of the tertiary indices fluctuated around 1, suggesting that the index samples are consistent in terms of oscillation and that the experts are the same in the difference of perception. Table 2 shows the descriptions and statistics of supplier performance ratings. e mean expert ratings on the seven tertiary indices under supplier sales performance and aftersale service were around 5, with standard deviations of about 1. It can be inferred that the index samples to a certain extent mirror the supplier performance ratings and the experts differ slightly in perception. However, the mean ratings on consumer source concentration were all smaller than 4, with standard deviations greater than 1.5. is means the experts poorly perceive consumer source concentration from the index samples and differ greatly in perception.
Similarly, as can be seen from Table 3, under relationship benefit, the mean ratings and standard deviations on consumer relationship benefit and enterprise participation were greater than 4.7 and 1, respectively. Hence, the experts moderately perceive relationship benefit and differ slightly in perceptions.

4.2.
CFA. An effect evaluation model was constructed, and the key evaluation criteria were sorted out to further evaluate the degree of influence of the three primary indices (online marketing performance, supplier performance, and relationship benefit) on the WeMedia ads marketing effect based on the demand of Internet economy. Firstly, a second-order plane model was established for the secondary indices under each primary index. Next, a fitness test was conducted to evaluate the correctness of the model. en, the model was adjusted and optimized to generate more reasonable evaluation results. e second-order plane model in Figure 1 covers two planes: online operation performance and online marketing value. e evaluations of online marketing performance should be reasonable and necessary for constituting the two planes. erefore, four pattern analyses were carried out: first-order one-factor analysis, first-order two-factor uncorrelated analysis, first-order two-factor correlated analysis, and second-order factor analysis.
Since the evaluation plane of supplier performance is a two-order plane, supplier performance evaluation was split into three planes, namely, supplier sales performance, consumer source concentration, and after-sale service. e evaluations of supplier performance should be reasonable and necessary for constituting the two planes. Similarly, four pattern analyses were carried out: first-order one-factor analysis, first-order three-factor uncorrelated analysis, firstorder three-factor correlated analysis, and second-order factor analysis ( Figure 2). Figure 3 shows the second-order plane model for relationship benefit.
is paper divides relationship benefit evaluation into two planes: consumer relationship benefit and enterprise participation. Similarly, four pattern analyses were carried out: first-order one-factor analysis, first-order two-factor uncorrelated analysis, first-order two-factor correlated analysis, and second-order factor analysis. Figure 4 shows the second-order plane model for WeMedia ads marketing evaluation based on the demand of Internet economy.

Model Verification and Results Analysis
e correctness of the three effect evaluation models was evaluated by a fitness test. e test results on online marketing performance evaluation model are listed in Table 4. e second-order factor analysis had the lowest χ 2 , DOF, and χ 2 /DOF (67.352, 51, 1.402). Hence, a second-order plane is suitable for the evaluation of online marketing performance. e fitness test results on the supplier performance evaluation model are listed in Table 5. Second-order factor analysis and first-order three-factor correlated analysis had consistent fitness in terms of DOF and χ 2 /DOF. at is, when the model plane only has three latent variables, namely, supplier sales performance, sale service, and consumer source concentration, the two models have the same fitness (25 and 1.739). However, the second-order factor analysis achieved the minimum χ 2 . erefore, the secondorder plane is the most suitable for the evaluation of supplier performance. e fitness test results on the relationship benefit evaluation model are listed in Table 6. Second-order factor analysis and first-order two-factor correlated analysis had consistent fitness in terms of DOF and χ 2 /DOF. at is, when the model plane only has two latent variables, namely, consumer relationship benefit and enterprise participation, the two models have the same fitness (26 and 1.786). However, the second-order factor analysis achieved the minimum χ 2 (42.187). erefore, the second-order plane is the most suitable for the evaluation of relationship benefit. Table 7 shows the results of parameter estimation and convergent validity for the online marketing performance Discrete Dynamics in Nature and Society     Discrete Dynamics in Nature and Society evaluation model. rough second-order CFA, the composite reliabilities of the planes of this model were all between 0.7 and 0.95, and their SRMSSARs were all greater than zero. erefore, the model makes no offending estimate. e AVEs of the two planes were both larger than 0.5, meeting the required standard convergent validity. is further confirms that the overall fitness of the online marketing performance evaluation model is acceptable. Table 8 shows the results of parameter estimation and convergent validity for the supplier performance evaluation model. e composite reliabilities of the three planes of this model were all between 0.9 and 0.95, and their SRMSSARs were all greater than zero. erefore, the model makes no offending estimate. e AVEs of the three planes were all larger than 0.7, meeting the requirements for standard convergent validity and fitness.  Discrete Dynamics in Nature and Society Table 9 shows the results of parameter estimation and convergent validity for the relationship benefit evaluation model. rough second-order CFA, the composite reliabilities of the six latent variables (evaluation criteria) were all between 0.75 and 0.95, and their SRMSSARs were all greater than zero. e AVEs of the two planes were all larger than 0.5. erefore, the model makes no offending estimate and meet the requirements on fitness.
Finally, the error variances of the overall second-order plane model for WeMedia ads marketing evaluation were

Conclusions
is paper explores the evaluation of WeMedia ads marketing under the demand of Internet economy. Specifically, a three-dimensional EIS was established for WeMedia ads marketing, covering such aspects as online marketing performance, supplier performance, and relationship benefit. en, multilevel FCE was performed to determine the weights of all indices. en, effect evaluation models were constructed and subjected to CFAs. Finally, descriptive statistics were analyzed to judge whether the index samples are sufficiently diverse. e fitness, parameter estimations, convergence validities of the three models were obtained through experiments. e authors tested whether the composite reliabilities of the planes of each model are all between 0.7 and 0.95, and their SRMSSARs are all greater than zero. Finally, the error variances were estimated, which confirm that the proposed model makes no offending estimate and meets the fitness requirement.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
e authors declare that they have no conflicts of interest regarding the publication of this paper.