Currently, crowdfunding research, which is a form of digital infrastructure on the rise (Nambisan, 2017), is attracting the attention of more and more scholars as an alternative financial tool that has grown considerably in the last few years (Bernardino et al., 2021). Nevertheless, not all campaigns reach their funding goal. One of the main failures of entrepreneurs is the lack of knowledge about value creation in their products or services (Eiteneyer et al., 2019). A way to involve potential customers at an early stage, maintain the innovativeness of products and establish a financial base to implement a business model is by crowdfunding (Eiteneyer et al., 2019; Nambisan, 2017). Digital marketing strategies can raise awareness, communicate with the crowd, and encourage them to participate and invest their money in a particular crowdfunding campaign (Bernardino et al., 2021). Several studies have investigated digital communication in connection to crowdfunding (Eiteneyer et al., 2019; Kraus et al., 2016; Lu et al., 2014; Paolo Crosetto & Regner, 2014).

This paper summarizes the current state of research on crowdfunding success factors. By drawing on existing theories from the field of marketing and communication applied to crowdfunding, we discover new success factors in behavior patterns, particularly in communication, as well as recommendations for action for startups. The theoretical findings explained below firstly present the forms of crowdfunding and their success factors, move on to the role of communication tools, and finally discuss the role of networks in connection to social media. It is the first time these research streams have been merged; crowdfunding research is a comparatively recent phenomenon, so various definitions exist without full scientific acceptance (Kraus et al., 2016). Kraus et al. (2016) state that crowdfunding focuses on raising financial support from users of devoted internet-based platforms.

There are many kinds of crowdfunding, which most noticeably differ in the amount of consideration the investor is given via material or immaterial compensation (Kazai, 2011; Kraus et al., 2016; Lopez et al., 2010). These can be divided into reward-based crowdfunding, donation-based crowdfunding, crowdsourcing, and equity-based crowdfunding (Belleflamme et al., 2012; Giudici et al., 2012; Kraus et al., 2016; Mollick, 2014). In donation-based crowdfunding, donors do not expect material rewards for their contributions, but at most so-called social rewards such as acknowledgments (Giudici et al., 2012). In crowdlending, the investor offers funding and in return receives a payment agreed upon in advance (Giudici et al., 2012). The equity-based crowdfunding model turns supporters into equity stakeholders who expect to profit from the venture in the future (Mollick, 2014). Of those listed above, however, the most common model is that of reward-based crowdfunding, where financial backers receive material or immaterial compensation, such as access to a product before market launch (Mollick, 2014). This crowdfunding form is of particular interest for this study since a large number of investors are needed for a project. Thus, proper communication plays an important role.

Literature on success factors of crowdfunding projects has risen sharply in recent years (Dikaputra et al., 2019; Ferreira & Pereira, 2018; Kunz et al., 2017; Crosetto & Regner, 2014; Yeh et al., 2019). There are several ways of classifying the success factors, with Kraus et al. (2016), for example, arranging them according to preparation time and campaign duration. According to them, the way they are classified is one of the main factors for success, with campaigns by non-profit organizations, which focus more on attracting capital than on profit more likely to be successful than others (Belleflamme et al., 2010; Glaeser & Shleifer, 2001; Kraus et al., 2016). In addition, donations shortly after the launch of a campaign can have a strong signaling effect and can thus also be attributed to the success factors (Colombo et al., 2015). The target amount of the campaigns and their duration were proven to be negative on the Startnext platform in previous studies, i.e., the higher the target amount or the longer the campaign duration, the lower the probability of success (Crosetto & Regner, 2014). It is important to verify whether this also reflects the current situation, leading to the formation of hypotheses 1 and 2:

Hypothesis 1: The lower the funding target of a reward-based crowdfunding campaign, the higher the campaign's probability of success.

During a campaign, quality indicators include quality signals in the founder’s communication (number of videos and blog posts on the platform), a high presence of pre-selling rewards, and the number of Facebook Friends (Crosetto & Regner, 2014). The video on the crowdfunding platform is also of great importance, as the multimedia clip can have a significant impact on the crowd, tell the real story behind the project and introduce the starter or even the whole team (Wheat et al., 2013). By becoming acquainted with the founder, trust in the project increases (Boeuf et al., 2014).

Figure 1.

Yeh et al. (2019) have classified the success factors into attraction and cognition-promoted aspects. The attraction-promoted and cognition-promoted aspects result from the categorization of information according to Figure 1 (Yeh et al., 2019). In the first category, the attraction-promoted aspects include the text description, images, Facebook Friends, and Crowdfunder experience. The more extensive the entrepreneur’s network on social media, the more likely a campaign will be successful (Ferreira & Pereira, 2018; Mollick, 2014; Yeh et al., 2019). Especially in the early stages of a project, social networking can raise the probability of receiving enough funding (Lu et al., 2014). To answer the question of as to precisely what targeted communication should look like, Schulz von Thun's Theory of Communication can be adapted to crowdfunding (Kraus et al., 2016). The original model has a sender, a four-sided message (1. Facts, 2. Self-revealing, 3. Relationship, 4. Appeal), and the recipient. Upon adapting it for crowdfunding, the project is equated with the sender, the four-sided message is the crowdfunding project [1. Description, pictures, videos, 2. Personal picture, 3. Networking, 4. Call to action (funding)], and the recipient is, in this case, the crowd (Kraus et al., 2016), who can be subsumed under the attraction-promoted factors.

Figure 2.

Hui et al. (2014) underlined that the campaign's success is determined by a mixture of information on the founder, facts about the project, customer relations and the call for funding (Hui et al., 2014). How exactly the networking and the customer relations aspects should be designed has not yet been sufficiently developed.

The second category of Cognition-promoted aspects includes Signaling factors, divided into Project response, Frequency of project updates, and an official website. In addition, the Kind factors which join the second category include funding other projects and the number of rewards that are promised to funders. The second category refers to signals that arise from social interaction, such as a frequent flow of information for updates and answers to financial backers’ questions to prevent information asymmetries (Yeh et al., 2019). Error-free spelling and regular status updates on the project are essential for success (Ferreira & Pereira, 2018). The Signaling Theory can be embedded into the theoretical background of the above-mentioned success factors of a crowdfunding campaign, which is an analysis of signals and the associated situations where they occur. In particular, it reduces information asymmetries between different parties to highlight the quality of the campaign (Kunz et al., 2017; Mavlanova et al., 2012). Project owners who keep their followers updated frequently can almost double the probability of reaching their funding goal (Kraus et al., 2016). Both the method and frequency of communication are essential to increase the likelihood of success. Xu et al. (2014) attribute a far more significant increase in the probability of doing well to thorough interaction with the crowd rather than to the design of the project site (Xu et al., 2014). The critical role of communication via social networks for a crowdfunding project is also highlighted by Mollick (2014), who adds that geography can impact the project's success, such as the business founder's closeness to the crowd and potential financial backers and his/her efforts to make the campaign appeal to them as much as possible.

The Communication Theory initially developed by Schulz von Thun and adapted for crowdfunding by Kraus et al. (2016) can be applied to Yeh’s classification of success factors (Kraus et al., 2016; Yeh et al., 2019). Aspects of media richness, such as text description and images, are subsumed under aspect 1. Facts. The second one, Self-revealing involves how the founder presents himself/herself, by way of a photo, for instance. One of the essential aspects which should be elaborated on is the third one, Networking. According to Signaling Theory, signal factors in the network can deliver any of the qualities that the campaign may possess to the crowd. Due to the crowd’s heterogeneity, the information flow must include all four facets of communication like in Figure 2, according to von Thun. Failing that, the prognosis of campaign success could have a negative outcome (Kraus et al., 2016). The interaction between the project owner and a potential backer can occur on any platforms, the crowdfunding platform itself, and on one or more connected social media channels (Hapsari & Sulung, 2021). This leads to the formation of the following hypotheses:

Hypothesis 3: The more blog posts shared within a reward-based crowdfunding campaign, the higher the likelihood of the campaign's success.

The likelihood of success depends on the business founder's social capital and the ability to invest it correctly in accordance with his/her needs. An increase in social capital, which is the sum of resources derived from the network of relations (Nahapiet & Ghoshal, 1998), can be beneficial for the financial performance of start-ups or existing companies (Yezza et al., 2021). There are two main approaches for classifying social capital into dimensions or concepts (Sánchez-Arrieta et al., 2021), which are detailed below.

The first approach, the subdivision into a multidimensional construct (also called the resource approach), can be applied as follows: the structural dimension, which represents the connections between the social units, and the relational dimension, which includes the quality of the ties, usually characterized by trust or a similar understanding of norms and values (Eiteneyer et al., 2019; Nahapiet & Ghoshal, 1998). Lastly, the cognitive dimension describes the common understanding of the individuals in a network (Bolino et al., 2002).

Figure 3.

The second approach, the Ties Type approach, refers to the relations themselves and the quality of their relationship (Sánchez-Arrieta et al., 2021). The people determine the resources within the network, and they can connect to each other and share know how or wealth. Topology describes those within the network and the size of it, which all depends on preferences and the ability to develop one. The quality of the network refers to the nature of relations and how much potential is available. Relations can involve expectations, obligations, trust, norms, and the feeling of closeness, and all these aspects can, in turn, influence behavior regarding participation in a crowdfunding campaign (Eiteneyer et al., 2019; Sánchez-Arrieta et al., 2021). Trust, for example, can, in turn, lead to knowledge sharing and drive transactional behavior (Chiu et al., 2006).

Figure 4.

If Social Capital Theory is applied, social capital can be divided into internal (i.e. internal to the platform) and external (i.e. connected social media channels [Liao et al., 2015]). If this categorization is made for the abovementioned success factors, it becomes clear that much more research has been done internally on the platform, and if social media has been included at all, it is rather like a dummy variable (i.e. the account exists and may or may not be linked). In some cases, however, this distinction is not made. For example, Eiteneyer et al. (2019) have examined the connection of social capital with backer involvement in relation to product innovativeness (Eiteneyer et al., 2019). proving that founders’ social capital or that of their respective teams lead backers to contribute to the campaigns, either by supplying them with know-how in product development or offering product innovation (Eiteneyer et al., 2019). It is questionable as to what extent social capital leads backers to participate in a campaign. The following diagram shows how theoretical relationships are interconnected with regard to quality signals:

Figure 5.

Social media can be an additional means of communication that can make use of social capital (Lu et al., 2014). For this reason, the many factors of social media channels must be examined to see which ones influence the likelihood of a project’s success. Even though Facebook and Instagram have plenty in common, they differ in the way they function. To differentiate their impact on the success of a crowdfunding campaign, the following hypotheses arise:

Hypothesis 5: The more subscribers there are on the linked Facebook page, the higher the campaign's probability of success.

Hapsariet et al. (2021) have found that linking to a Facebook account does not necessarily increase the likelihood of success. This suggests that the number and quality of interactions (activity level) and the network itself are more important for predicting success.

Hypothesis 10: The mere inclusion of a link to the Facebook page on the crowdfunding platform for the campaign contributes to the campaign's success.

Startnext, like many other crowdfunding platforms, allows the inclusion of a link to social media platforms on the campaign page. In turn, the link to the campaign page can be integrated on each platform (Xu et al., 2014). The characteristics of social media, including the fact that it is highly dynamic and has an extensive reach, can be actively used to achieve benefits for one's own campaign (Xu et al., 2014). Furthermore, paid advertisements are an option on social media sites, which immediately provides the link to the campaign webpage, whereby specific, predefined target groups can be addressed at certain times, and there is significantly less wastage in communication (Xu et al., 2014). Age, gender, and interests can be defined, and the geographical framework means that previously gained knowledge about regional advantageousness through regional affiliation can be applied directly. In addition, the trust factor, which according to Social Capital Theory, helps determine the quality of the relationship, can be strengthened by using a social media account (Cumming et al., 2016). A social media presence is a strong indicator that fraud has not occurred because potential backers can quickly track further activities on the account; fraud is much more difficult to commit and is also easier to detect when it does happen (Cumming et al., 2016). However, a linked social media profile is not always a quality signal. If it has a really small number of followers, it can deter potential backers (Mollick, 2014). To the best of our knowledge, there are no studies to date that go beyond the mere effect of the link to a social media account and its contribution to the success of a campaign. Instagram, in particular, has been largely left out of the equation, and only the effect of Facebook and Twitter has been considered so far (Kraus et al., 2016; Crosetto & Regner, 2014).

This study aims to develop upon the knowledge about success factors for a crowdfunding campaign and connect theoretical concepts of the Social Capital Theory, the Signaling Theory, and the Communication Theory by Schulz von Thun and elaborate new assumptions within a conceptual, theoretical framework. The goal is to discover which derived key factors in the communication behavior of founders are most likely to lead to a campaign’s success.

To elaborate this, data has been cumulated from the largest reward-based crowdfunding platform in Germany, called Startnext. This, just like global counterparts such as Kickstarter and Indiegogo, represents a platform on which transactions can be carried out, with the necessary legal framework being done behind the scenes. The funding volume in February 2022 reached around €130,370,000 with 1,888,000 registered users and 13,121 successful projects. The starter can choose to run a campaign for a minimum of 10 to a maximum of 90 days (Crosetto & Regner, 2014). Every campaign page consists of a video, a funding target, the campaign’s duration, the project’s description, a pinboard, a blog section, and the reward levels, which backers obtain when the funding target is reached. A campaign’s success can only be achieved if the funding sum is reached in the period chosen by the founder and backers will also only receive the promised reward in this scenario. If the funds raised are not paid out owing to a campaign’s failure, all payments made up to that point will be refunded. During the study, a total of 169 campaigns were observed, and a Web Scraper gathered the corresponding data. Canceled and non-launched campaigns were eliminated from the data set. The data collection period covered from 6^th January 2020, the date of the first campaign to 27^th April 2020, when the last campaign ended. The following data was extracted daily throughout the campaigns: pinboard posts, blog posts, funding, Instagram subscribers, Instagram posts (with content, number of likes and comments), Facebook subscribers, Facebook "likes," and Facebook posts (with content, number of likes and comments). In addition, unique contact information, rewards, and funding periods were sourced. Based on the Signaling Theory, the Theory of Communication by Schulz von Thun, and the Social Capital Theory, eleven hypotheses were developed to address the research question, namely, which key factors in starters’ user behavior lead to campaign success, with a focus on the use of network effects and considering the communication aspect. The following hypotheses in Figure 6, which are derived from the theoretical part, are analyzed by the gathered data set.

Figure 6.

Numeration	Hypothesis
H1	The lower the funding target of a reward-based crowdfunding campaign, the higher the campaign's probability of success.
H2	The shorter the duration of a reward-based crowdfunding campaign, the higher the campaign's probability of success.
H3	The more blog posts shared within a reward-based crowdfunding campaign, the higher the likelihood of the campaign's success.
H4	The more pinboard entries a reward-based crowdfunding campaign has, the higher the campaign's probability of success.
H5	The more suscribers there are on the liked Facebook page, the higher the campaign's probability of success.
H6	The more "likes" there are on the linked Facebook page, the higher the likelihood of the campaign's success.
H7	The more followers there are on the linked Instagram page, the higher the probability of the campaign being a success.
H8	The more posts there are on Facebook, the higher the campaign's probability of success.
H9	The more posts there are on Instagram, the higher the campaign's probability of success.
H10	The mere inclusion of a link to the Facebook page on the crowdfounding platform for the campaign contributes to the campaign's success.
H11	The mere inclusion of a link to the Instagram page on the crowdfounding platform for the campaign contributes to the campaign's success.

The dependent variable is campaign success, defined as reaching the funding goal. Thus, this variable is measured on a binary scale and assigned 1 if the funding target was reached, and 0 if it was not reached, in the selected period. If the initial funding amount set by the starter is not reached, the contributions made up to that point will be returned to the backer, and the project will not receive any money. If this occurs, the success of the campaign becomes negated (=0). If the funding goal is accomplished by reaching the initial set amount of money during the campaign period, the campaign is considered successful, and the starter receives the funding for his/her project (=1). The independent variables are classified in Figure 7, labeled “dichotomous” or “metric”, with the abbreviations used in the study, assigned to the hypothesis under evaluation:

Figure 7.

Abreviation	Variable	Measurement Level	Assigned Hypothesis
S	Success of Campaign	dichotomy	All
FA	Funding Amount	metric	H1
D	Campaign Duration	metric	H2
B	Blog Entries	metric	H3
P	Pinboard Posts	metric	H4
FF	Facebook Follower	metric	H5
FL	Facebook Likes	metric	H6
IF	Instagram Follower	metric	H7
FP	Facebook Posts	metric	H8
IP	Instagram Posts	metric	H9
LF	Link to Facebook	dichotomy	H10
LI	Link to Instagram	dichotomy	H11

External social capital is raised through the linked Facebook and Instagram channels with all factors corresponding to them. Communication with internal social capital includes platform-internal exchange.

The metrically scaled variables were examined in the descriptive analysis with the most important statistical values. In contrast, the nominally scaled variables were evaluated only with their frequencies. Figure 8, below, summarizes the most important descriptive statistics for the nominally scaled variables.

Figure 8.

Variable	Mean	Median	Std. Deviation	Variance	Skewness	Kurtosis
FA	9471.69	4500.00	32367.747	1047671023.556	10.842	128.991
D	35.70	34.00	12.135	147.264	1.190	3.749
IP	6.43	2.00	9.102	82.845	1.817	3.456
IF	882.99	109.50	5448.124	29682055.946	11.788	146.187
FP	5.98	2.00	8.513	72.473	2.247	7.604
FL	656.83	84.50	1618.279	2618827.940	4.154	19.503
FF	605.82	81.50	1400.023	1960063.912	3.709	14.139
B	0.83	0.00	1.967	3.868	3.938	18.779
P	1.83	1.00	3.457	11.948	5.408	35.417

In the descriptive analysis of the metric variables in Figure 8, outliers in the funding target are conspicuous. They are not adjusted because they reflect reality, where the targets have a high variance. Other values identified as potential outliers do not represent erroneous distorting values upon review, meaning that no adjustment takes place in this respect. Further tests for outliers are performed as part of the following binary logistic regression for nominal variables: Campaign success (S), linking Instagram account on campaign page (LI), and linking Facebook account on campaign page (FV). Frequencies have been examined with 66.7% of the campaigns being successful, 62.5% of them linked to an Instagram account and 59.5% to a Facebook one.

Figure 9.

		Success	Link to Instagram channel	Link to Facebook channel
Success	Pearson Correlation	1	.130	.214**
	Sig. (2-tailed)		.092	.005
	Sum of Squares and Cross-products	37.333	5.000	8.333
	Covariance	.224	.030	.050
	N	168	168	168
Link to Instagram channel	Pearson Correlation	.130	1	.263**
Link to Instagram channel	Sig. (2-tailed)	.092		<,001
	Sum of Squares and Cross-products	5.000	39.375	10.500
	Covariance	0.30	.236	.063
	N	168	168	168
Link to Facebook channel	Pearson Correlation	.214**	.263**	1
	Sig. (2-tailed)	.005	<.001
	Sum of Squares and Cross-products	8,333	10.500	40.476
	Covariance	.050	.063	.242
	N	168	168	168

When correlating the nominal variables' success and linking it on the campaign page to the Facebook or Instagram accounts, it is interesting to note that it is only the link to the Instagram account that is significant (see Figure 9). The metric variables in Figure 8, (FL) Facebook Likes and (FF) Facebook followers, have also been tested for their Pearson correlation coefficient, where there is a strong positive correlation, r=.827, p<.001. This may be because becoming a fan of a page automatically results in a Like. Here, users must first be active for this automatism to be reversed, resulting in the decision to exclude the Facebook followers variable (FF) in the binomial logistic regression as an input factor, which reflects the same aspect and explanatory approach due to the identical underlying network.

Logistic regression is calculated to evaluate whether the current data set can provide a significant contribution to clarifying the query over which factors influence campaign success. The goal of logistic regression is to find probabilities for the occurrence of either of the campaign outcomes, i.e. success or failure. Since the dependent variable is dichotomous, a binary logistic regression must be calculated (Backhaus et al., 2018). The underlying binary logistic model is: π(x)=12+e−z(x)=11+e−(a+β1x1+...+βjxj)

The requirements for performing a binary logistic regression can be confirmed. Both the sample size and the classification of the variables are large enough. The dependent variable must be dichotomous, whereas the independent ones must be categorical or at least interval scaled, which is the case for the data set used. Furthermore, linearity can be assumed after it is tested by the Box-Tidwell method and then applying the Bonferroni correction (Box & Tidwell, 1962; Tabachnick & Fidell, 2007). Testing for multicollinearity has revealed a low correlation between predictors (r<60), which does not further affect the subsequent analysis (Backhaus et al., 2018). In the binomial logistic regression performed below, the variable success (S) is the dependent variable which needs to be explained. FA, D, B, P, FL, IF, FP, IP, LF and LI are the independent variables. This results in the following regression model:

S(Success=1)=11+e−(β0+β1∗FA+β2∗D+β3∗B+β4∗P+β5∗FL+β6∗IF+β7∗FP+β8∗IP+β9∗LF+β10∗LI)

In the model calculation, classification plots, the Hosmer-Lemeshow goodness-of-fit, the case-wise listing of the residuals, the correlation of the estimates, and the 95% confidence interval for exp(B) need to be reported. The procedure for variable selection is inclusion. In the case-wise listing, values which lie outside of two are indicated as outliers.

Figure 10.

	B	S.E.	Wald	df	Sig.	Exp(B)	95% C.I.for EXP(B) Lower	Higher
Funding target amount in euros	,000	,000	12,237	1	<,001	1,000	1,000	1,000
Duration of campaign in days	-,032	,016	4,033	1	,045	,968	,938	0,999
Number of blog posts	,092	,105	,765	1	,382	1,096	,893	1,346
Number of pinboard posts	,031	,080	,146	1	,702	1,031	,881	1,207
Number of Facebook Likes (end of campaign)	,000	,000	,057	1	,811	1,000	1,000	1,000
Number of followers on Instagram (end of campaign)	,000	,000	,135	1	,714	1,000	1,000	1,000
Number of posts on Facebook (during the campaign)	,153	,053	8,299	1	,004	1,165	1,050	1,293
Number of posts on Instagram (during the campaign)	,006	,030	,042	1	,838	1,006	,949	1,066
Link to Facebook account	-,107	,519	,042	1	,837	,899	,325	2,484
Link to Instagram account	,434	,490	,785	1	,376	1,544	,591	4,032
Constant	1,691	,635	7,097	1	,008	5,423

The regression model is statistically significant, χ (10) = 46.031, p < .001, with an acceptable variance resolution of Nagelkerkes R = .333. The classification is correct in 78.6% of cases, with a sensitivity t of 94.6% and a specificity of 46.4%. Three out of the ten variables in Figure 10 included in the model are significant: Funding target amount in euros (p < .001), Duration of the campaign in days (p < .05), and Number of Posts on Facebook (p < .005). The other variables in Figure 10, namely, Link to Instagram account (p = .376), Number of posts on Instagram (p = .838), Number of followers on Instagram (p = .714), Link to Facebook account (p = .837), Number of Facebook Likes (p = .811), Number of blog posts (p = .382), and Number of pinboard posts (p = .702), do not significantly affect the predictive performance of the model. A higher funding goal has a negative effect on the probability of success, with an odds ratio of 1.000 (95% CI [1.000, 1.000]), as does a longer campaign duration, with an odds ratio of 0.968 (95% CI [0.938, 0.999]).

One independent variable has been added to the model to check the results for robustness of regression. The author has added the age of the starter as an additional variable. The robustness check shows significant results that are consistent with Figure 10. Based on the results, starters’ ages have no significant effect on the success of crowdfunding projects. Therefore, it can be concluded that the results of the robustness check are consistently significant and aligned with the results in Figure 10.