Journal of International Business Research and Marketing
Volume 5, Issue 1, November 2019, Pages 26-31
Raising Money on Crowdfunding Platforms: The Components of Success
1 Urszula Mrzyglod, 2 Marcin Skurczynski, 3 Joanna Adamska – Mieruszewska
1 2 3 University of Gdansk, Gdansk, Poland
Abstract: Crowdfunding differs significantly from standard business and personal financing since it allows those in need of funds to collect small contributions from the community of internet users. This unique feature makes crowdfinancing popular throughout the world and supports growth in terms of the number of financed projects and the volume of financing. Nonetheless, the percentage of projects fully financed on crowdfunding platforms is relatively low. This raises a question about the origins of successful financing. Based on a unique dataset of 1,543 projects, we investigated the attributes of projects on one of the largest crowdfunding platforms operating in Poland: PolakPotrafi.pl. This paper aims to determine the key drivers of successful crowdfinancing. Along with standard statistical measures, we conducted logistic regressions. We found that the longer a project’s duration and the higher funding goal per supporter are, the less likely the project is to be fully funded. In contrast, active communication by the project’s founder increases the chances of successful crowdfinancing. We also found that crowdfinancing is not geographically neutral: conducting a project in Poland’s capital is correlated with a higher probability of obtaining funds.
Keywords: Crowdfunding, Crowdfunding Platforms, Business, Personal financing
One of the key challenges that entrepreneurs face is providing funding at an early stage of their business. Traditionally, financing is provided by banks, angel investors, venture capital funds or shareholders. Alternatively, for several years, entrepreneurs have benefitted from crowdfunding. In contrast to traditional funding, crowdfinancing is a process through which individuals or groups, operating both for profit and not for profit – can collect small contributions from a large number of internet users (Mollick 2014).
Crowdfunding differs significantly from the standard early stages business financing. First, it is based on small contributions donated by a large number of supporters. In crowdfinancing, the community of internet users, rather than a single investor, determines the success of the venture. Second, most of the biggest crowdfunding platforms use the donation- or reward- based model, in which the project founder promises rewards to the funders, instead of shares or interest. Finally, crowdfunding platforms offer individuals and organisations the opportunity to launch projects of any kind – not only business but also cultural, sport or social. The innovative nature of crowdfunding supports its growth. According to a Massolution report, the global crowdfunding industry reached more than USD 34 billion in 2015 (Massolution 2015). However, the percentage of projects that are fully funded remains relatively low. For instance, the success rate of projects on the American platform Kickstarter is 35.84% (Wang 2018), on the French platform Ulule 65% (Ulule.com 2018) and the Polish platform PolakPotrafi.pl about 50% (Adamska-Mieruszewska, Mrzygłód and Skurczyński 2018).The relatively low success rate and the unique nature of crowdfunding raise the question of what the determinants of success are.
The majority of previous studies on crowdfunding have analysed the determinants of success mainly on the basis of US and Western European data. This is mainly because these regions have a longer history of crowdfinancing and more advanced crowdfunding markets than emerging economies. In empirical researches, less attention has been paid to Central and Eastern Europe, though these countries have also experienced dynamic growth in the crowdfunding. We contribute to the empirical literature on the subject by providing the first in- depth empirical study of crowdfunding success predictors in Poland. We chose to study the largest Polish crowdfunding platform, PolakPotrafi.pl. Although the interest in crowdfinancing in Poland is still relatively small in comparison to other European Union countries, the growing use of the internet and social media especially by entrepreneurs has increased the potential of the Polish crowdfunding market. According to a survey conducted by the Cambridge Centre for Alternative Finance, the Polish market is one of the most active online alternative finance markets in Central and Eastern Europe. Among the platforms 205 surveyed by the Cambridge Centre in 27 European countries, 11 were located in Poland. The Polish market is ranked among the top 15 in Europe for its volume of alternative finance (equal to €4 million in 2014). In comparison, the survey covered only four online platforms from Estonia, two from the Czech Republic and Hungary and one from Slovakia (Cambridge Centre for Alternative Finance 2016). The paper is structured as follows. A literature review of empirical studies on the determinants of a project’s success and failure is presented in the next section. The empirical research design and findings of this study are discussed in Section 3, while conclusions and implications for further research appear in Section 4.
2. Literature Review
Some recent crowdfunding literature concentrates on the motivations for crowdfunding from the project initiator’s perspective: Belleflamme, Lambert and Schwienbacher (2013); Gerber, Hui and Kuo (2012) and Hu, Li and Shi (2014). Harms (2007) and Ordanini et al. (2011) looked at the reasons investors support projects. Agrawal, Catalini and Goldfarb (2011), as well as Kim and Viswanathan (2013) studied the geography of crowdfunding.
One of the first studies on the success factors for crowdfunding was conducted by Mollick (2014), who used a dataset of 48,500 successful and unsuccessful Kickstarter projects from 2008 to July 2012. For the empirical study, the author selected 22,651 initiatives with a value of US $5,000 or more. According to Mollick’s final model, seven variables foster crowdfunding success: goal size, duration and spelling errors in the project description negatively impact projects’ success, while the number of pictures, images and videos in the description, the frequency of updates and the number of Facebook friends of initiator positively affects success. Cordova, Dolci and Gianfrate (2015) extended Mollick’s research through the use of a dataset based on different crowdfunding platforms. They used a dataset of 1,127 technology projects from 4 crowdfunding platforms: Kickstarter (259 projects), Ulule (91 projects), Eppela (9 projects) and Indiegogo (424 projects). In their empirical analysis, the authors ran two separate regressions, based on 410 projects from Indiegogo and Eppela, to explain fundraising success. As in Mollick’s study, they found that the size of a financing goal is negatively associated with a project’s success. The authors also identified four other variables that significantly impact a project’s success: the mean amount contributed to the project each day, the number of supporters, the mean amount contributed to the project by each supporter and the duration of the crowdfunding campaign. However, unlike in Mollick’s study, they found that duration is positively associated with success. Cordova, Dolci and Gianfrate (2015) did not find any significance in the number of updates from the project founder, comments posted by other supporters or videos included in the project description. According to the authors, also geography does not seem to increase the probability of success.
Müllerleile and Joenssen (2014) examined the 45,400 crowdfunding campaigns announced on the Kickstarter platform. The authors identified the positive influence of the number of comments, project updates and rewards (pledge levels) on the campaign success. Among the unfavorable factors, the authors proved the significance of the target amount (funding goal) and the project duration which stays in line with Mollick study (2014). Moreover, Müllerleile and Joenssen (2014) confirmed the relevance of the unique project’s website and the level of competition between projects on the starting day. As such, the authors provide an interesting contribution on the project comparative advantages. However it has to be mentioned that the authors do not explain how the website ‘uniqueness’ was assessed.
A complex research devoted to the project’s communication tools was conducted by Beier and Wagner (2015). The authors investigated the relevance of the number of updates, photos and video presentation on the project’s website on the success of the campaign. Moreover, the authors checked whether the additional social media tool such as additional website, Twitter account and Facebook page created for the project significantly influence the chances of the project’s success. Among the variables described above only videos and active communication (publishing updates) reached statistical significance and positively influenced the projects’ success.
Zhou et al. (2016) made an interesting contribution to the existing literature by studying 151,752 Kickstarter projects, including 69,784 successful initiatives. They conducted two logistic regressions. The first was based on control variables: financing goal, duration, Facebook friends, number of images, number of videos and pledge levels for rewards. The second included the length of the description, the readability index, the tone of the description and past experience and expertise by the project’s founder in crowdfunding. As in other studies, Zhou et al. (2016) found that the size of the financing goal and the duration of the crowdfunding campaing are negatively associated with a project’s success. On the other hand the number of Facebook friends, the number of videos and images and the levels of rewards are positively associated with success. Extending the studies by Mollick (2014) and Cordova et al. (2015), Zhou et al. (2016) noticed that the project description and the language used significantly influence the success of a project. According to their research, past experience with crowdfunding initiatives, as well as experience in backing other projects, also have a positive effect on crowdfunding projects.
Contrary to the studies described above, Hobbs, Grigore and Molesworth (2016) conducted research devoted to the success factors of crowdfunding based on the significantly lower number of campaigns. The research sample consisted of 100 campaigns aiming to raise funds for the independent film projects. The authors included a range of variables grouped into four categories: operation of the campaign, network management, financial issues and the quality of the campaign. In comparison to other studies, Hobbs et al. (2016) provided an interesting coding scheme of the campaign’s quality assessment which encompassed reward and pitch quality. In the latter case, campaigns were evaluated with respect to the preparedness and exhibited passion of the project founder. Moreover, the authors ensured that data describing the network management referred to the exact time of campaign which justifies the smaller research sample. Basing on the discriminant analysis Hobbs et al. (2016) identified the ‘strong’ and ‘weak’ predictors of the crowdfunding campaigns. The first group consists of pitch and reward quality, money raised, the number of project contributors, Facebook shares as well as the number of updates provided by projects founder whereas the number of rewards, results of Google search, project’s Facebook friends, the goal and length of the campaign remain to the second group of predictors.
3. Empirical Research Design
3.1 Sample Characteristics
3.1.1 Data Collection
In the paper, we used data extracted from the Polish crowdfunding platform PolakPotrafi.pl since 2011 until 2nd September 2016. The sample consists of both the successful projects – the ones that raised the required funding, and the unsuccessful, that did not reach the required goal by the end of the campaign. Data was collected from the platform’s archive. In the case of unsuccessfull projects data had to be collected through the supporters unique web profiles. Initially, a list of 17,098 profiles was compiled. Identified users supported 2,380 successful and unsuccessful projects in total. The compiled dataset underwent a preliminary analysis. The projects with incomplete or incorrect information were eliminated. Eventually, the dataset includes 1543 crowdfunding projects with 826 successes and 717 failures. This complex procedure of gathering data on the unsuccessful project was necessary due to the platform’s information disclosure policy.
PolakPotrafi.pl is one of the largest crowdfunding platform operating in Poland. Established in March 2011, the platform uses donation- and reward-based crowdfunding. Like Kickstarter, the platform follows an ‘all or nothing’ model, meaning that supporters’ donations are collected only if the project goal is reached. Since 2011 till 2018 PolakPotrafi.pl obtained more than PLN 20.4 million (USD 6 million) in funding for 3,300 projects from 209,000 supporters (PolakPotrafi.pl 2018). The platform’s creators assign projects to one of nineteen categories: design, journalism, education, photography, games, food, fashion, travel, society, technology, art, dance, theatre, sport, comic books, books, music, video/film and events. The platform also uses the category ‘others’ for projects that cannot be easily assigned to one of the other nineteen categories. According to a study conducted by Adamska, Mrzyglod and Skurczynski (2017), between 2011 and 2016 the average success rate of the projects analysed was 50%; projects in the dance and music categories had the highest success rates. The mean amount donated by supporters was PLN 5,880 (USD 1,717), while the median was more than PLN 2,400 (USD 702).
3.1.3 Variables Uni-variate Statistics
Crowdfunding projects can be classified as a success if the initiators reach their funding goal and as a failure if they do not reach their goal by the end of the campaign. On the basis of the literature studies, we indicated several antecedents that might influence the success of crowdfunding projects. The variables are described in Table 1.
Table 1: Description of variables
We expanded the existing empirical studies by introducing variable amount per supporter. We argue that this relative ratio is more appropriate since it links the total target amount with the mean contribution given by supporter. Moreover, we propose the variable media representing the number of media notices about the crowdfunded project. This variable is justified because the presence in media increases the project’s recognition among potential supporters. Media notices included press as well as TV and radio news and these information was extracted from the project’s website on the PolakPotrafi.pl.
First, as we aim to verify the significance of the selected variables for the crowdfunding campaigns we summarise the number of successfully funded projects across their quartiles. We check for the statistical differences between the 1st and 4th quartiles and employ for three variables (time, supporters and amount per supporter) the Kolmogorov-Smirnov as well as U Mann-Whitney test. Second, based on the standard approach in the case of dichotomous data we conduct the binary logit regression which allows to capture the changes in the probability 𝑝 of successful crowdfunding campaign as a function of the selected antecedents (regressors). Since we obtained four potential specifications we compare the obtained regression results basing on the pseudo-R2 measure and the post-estimation tests. In the latter case, we conduct the standard post-estimation analysis concerning the correctness of regression specification (link test) as well as to the goodness of fit (Hosmer-Lemeshow test). Finally, having chosen the best model, we provide the standard measures of its predictive accuracy.
4.1 Basic Statistical Characteristics
Table 2 presents descriptive statistics for selected variables. The average campaign duration for the projects in our sample was 40 days: 37 days for projects classified as successes, and 43 days for those classified as failures. The projects had an average funding goal of PLN 10,200 (almost USD 3,000), and half of them had goals of more than PLN 6,000 (USD 1,750). The median funding goal was more than PLN 2,500 (USD 730) higher in projects classified as failures than in those that successfully gained funding. On average, successful projects were overfunded by 11%, while failed projects were underfunded by 91.7%. Our results also showed that the average funding goal per supporter was PLN 625 (USD 182.5), with a standard deviation of more than PLN 1,800 (USD 526.7). Such a significant difference is the result of the diversity of the projects’ size. The minimum funding goal in the sample was PLN 105 (USD 30.7), while the maximum was PLN 150,000 (USD 43,810). In our sample, on average, founder supported 2.20 other projects (variable backed).
Table 2: PolakPotrafi.pl projects, 2011–16
a.coef.var – coefficient of variation
Source: Computation in Statistica
The significance of project features like duration, the number of supporters or the initiator’s experience with crowdfunding did not differ significantly from that found in previous studies by Mollick (2014), Kuppuswamy and Bayus (2015), Hobbs, Grigore and Molesworth (2016) and Zhou et al. (2016). In contrast to Mollick’s findings on Kickstarter projects, PolakPotrafi.pl project supporters were less active and did not comment on projects as frequently. In our study, we investigated the impact of both author’s and supporters’ activity regarding the project. The first one was measured by the number of news (updates) about the initiative that the author uploaded on the project’s website. The supporters engagement was indicated as the number of comments regarding the project. The average number of comments on each project in the sample was 4.44, while there was on average 3.59 news and 1.3 media notices per project. In the area of author’s and supporter’s activity, the significant difference between successful and unsuccessful projects may be noticed. The distribution of the selected antecedents are generally strongly skewed and revelled the lack of normality.
Figures 1-3 show the difference between the frequency of founders’ and supporters’ activity in successful and unsuccessful projects. Over 22 percent of successes had at least one media notice, in comparison to 12.18% of failures. The supporters were significantly more active in those projects that managed to collect the targeted financing – almost 30 percent in comparison to over 17.7 percent in case of failures. Similarly, in almost 20% of successfully funded projects founders updated the project’s website more than five times, while in the case of failures only 3.7 percent were updated at the same level of frequency.
Although Agrawal, Catalini and Goldfarb (2011) indicate that crowdfunding platforms tend to reduce the distance barrier in funding projects, still the geography seems to play a role in determining the crowdfunding project’s success. The majority of projects submitted to collect money on PolakPotrafi.pl are initiated by authors from large cities. In our research, more than 63% of the projects were initiated by a founder from a large city with more than 250,000 inhabitants, and 21% by founder from Warsaw – Poland’s capital (Figure 4). The frequency of project’s author from Poland’s capital city differs significantly between successes and failures. Over 26 percent of successful projects’ was initiated by authors located in Warsaw, in comparison to 18.5 percent in the case of failures.
Source: Own compilation
In Table 3 we summarize the results of the comparison across the quartiles of the three variables: amount per supporter, time and the number of supporters. As the mean value of variables within quartiles is not explanatory, we check for the number of successes along the quartiles and report the success rate. In the last two columns, we present the p values of the selected difference tests (Kolmogorov-Smirnov, U Mann-Whitney).
Table 3: Univariate sample statistics and difference tests
UM-W: U Mann Whitney, K-S: Kolmogorov-Smirnov
Source: Own computation in Statistica
The obtained result indicates that the success rate is significantly higher for the projects with smaller values of the amount to supporter ratio. The opposite pattern holds for the number of supporters and project’s duration (time). Overall, this statistical analysis provides the first evidence of the importance of the selected antecedents in explaining crowdfunding campaigns’ successes.
4.2 Logit Regression Results
In Table 4 we report the final results of the logit regressions. The first specification (model 1) encompassed following variables: amount per supporter, time, the number of supporters and the capital city. All variables, except the capital city variable gained statistical significance. In the models 2 and 3 we concentrated on the variables that are proxies for founders’ and supporters’ involvement. In both specifications 2 and 3 we confirmed the significance of the news, media and comments variables. Moreover, in model 3 we checked the importance of the number of crowdfunding projects previously supported by the project founder on the platform PolakPotrafi.pl (variable backed).
Table 4: Regression results
|Variables||Model 1||Model 2||Model 3||Model 4|
|time||-0.26*** (0.0069)||-0.015** (0.0064)|
|news > 5||(0.0033)||1.66***||1.61***||1.30***|
|media||0.085*** (0.0268)||0.87*** (0.2677)|
*** p<0.01** p<0.05 *p<0.1 note: the coefficient standard deviation is given in brackets
Source: Own computation in Stata
As we report in Table 4 the levels of pseudo-R² for both specifications 2 and 3 were relatively low. Moreover, not all of the specification tests for regressions 1 to 3 were satisfied. Thus we argue that variables describing the involvement of PolakPotrafi.pl users do not significantly raise the chances of success of the campaigns.
Therefore we had to examine other specifications, and finally, we argue that the best model includes the following variables: amount/supporter, the capital city, time and news. All variables remain statistically significant, and the results of the post-estimation tests confirm the appropriateness of specification 4 (Table 5).
Table 5: Post-estimation check of the logit model
Note: a. reports χ2 stat.; b. reports the level of z-stat
Source: Own computations using Stata
The level of the pseudo-R2, as well as the measure of the predictive accuracy of 89.95%, indicates that the specification 4 fits the underlying data. In the final specification, the amount per supporter and the duration (time) are negatively related, whereas the capital city and the frequency of updates (news) on the project website positively affect the project’s success rate.
The growing interest in crowdfunding in Poland and other Central and Eastern European countries justifies the relevance of the undertaken research problem. Moreover, the preceding studies devoted to the Polish crowdfunding market did not sufficiently cover the problem of success predictors. One of the reasons for this scarcity of the research is the difficulty in obtaining large datasets. Our study fulfils the cognitive gap in this research area. On the base of a unique dataset of 1543 projects extracted from the largest Polish crowdfunding platform PolakPotrafi.pl, by using logistic regression model we examined the potential antecedents of funding success. Among the selected eight antecendents we confirmed the significance of the four variables, namely: amounts per supporter, time, capital city and news published by the project’s author. The amount requested per one supporter and the time negatively impact projects’ success, while the frequency of project’s updated and the author’s origin from the capital city, positively influence crowdfunding’s success.The obtained results are generally in line with the existing literature regarding the US crowdfunding market. On the contrary to the US market, the Polish project’s participants are less active on the crowdfunding platform, and their involvement is not an important factor of the campaing’s success.
- Adamska-Mieruszewska, Joanna, Urszula Mrzygłód, Marcin Skurczyński (2017), “Success and failures of crowdfunded projects in Poland”, in Finanse, Rynki Finansowe, Ubezpieczenia, Vol. 5(89), part 2, 415-426. Crossref
- Agrawal, Ajay K., Christian Catalini, Avi Goldfarb (2011), “The geography of crowdfunding”, in National Bureau of Economic Research Working Papers, No. w16820.
- Beier Michael, Kerstin Wagner (2015), “Crowdfunding Success: A Perspective from Social Media and E-Commerce”, Thirty-Sixth International Conference on Information Systems, Fort Worth.
- Belleflamme, Paul, Thomas Lambert, Armin Schwienbacher (2013), “Individual Crowdfunding Practices” in Venture Capital, Vol. 15(4), 313–333. Crossref
- Cordova, Alessandro, Johanna Dolci, Gianfranco Gianfrate (2015), “The determinants of crowdfunding success: evidence from technology projects”, in Procedia-Social and Behavioral Sciences Vol. 181, 115-124. Crossref
- “Crowdfunding Industry Report” (2015), Massolution, crowdexpert.com (22.01.2018).
- Gerber, Elizabeth M., Julie S. Hui, Pei-Yi Kuo (2012), “Crowdfunding: Why people are motivated to post and fund projects on crowdfunding platforms”, in Proceedings of the International Workshop on Design, Influence, and Social Technologies: Techniques, Impacts and Ethics, Vol. 2, No. 11.
- Harms, Michel (2007), “What drives motivation to participate financially in a crowdfunding community?”, Thesis Master in Marketing, University of Amsterdam. Crossref
- Hobbs, Jake, Georgiana Grigore, Mike Molesworth (2016) “Success in the management of crowdfunding projects in the creative industries”, in Internet Research Vol. 26.1, 146-166. Crossref
- Hu, Ming, Xi Li, Mengze Shi, “Product and pricing decisions in crowdfunding”, in Marketing Science Vol. 34.3, 331-345. Crossref
- Kim, Keongtae, Siva Viswanathan (2014), “The experts in the crowd: The role of reputable investors in a crowdfunding market”, in SSRN Working Papers, No 2258243. Crossref
- Kuppuswamy, Venkat, Barry L. Bayus (2015), “Crowdfunding creative ideas: The dynamics of project backers in Kickstarter“, in UNC Kenan-Flagler Research Paper, Vol. 2013-15. Crossref
- Mollick, Ethan (2014), “The dynamics of crowdfunding: An exploratory study”, in Journal of business venturing Vol. 29.1, 1-16. Crossref
- Müllerleile Thomas., Dieter William Joenssen (2015), “Key Success-Determinants of Crowdfunded Projects: An Exploratory Analysis”, in Data Science, Learning by Latent Structures, and Knowledge Discovery. Studies in Classification, Data Analysis, and Knowledge Organization, eds. Lausen B., Krolak-Schwerdt S., Böhmer M, Springer, Berlin, Heidelberg. Crossref
- Ordanini, Andrea, Lucia Miceli, Marta Pizzetti, A. Parasuraman (2011), “Crowd-funding: transforming customers into investors through innovative service platforms”, in Journal of service management VOl. 22.4, 443-470. Crossref
- pl, www.polakpotrafi.pl (20.01.2018).
- “The 2nd European Alternative Finance Industry Report” (2016), Cambridge Centre for Alternative Finance, Sustaining Momentum, available on: www.ey.com (12.02.2018). Ulule.com (20.02.2018). Crossref
- Zhou, Mi Jamie, Baozhou Lu, Weiguo (Patrick) Fan, G. Alan Wang (2016), “Project description and crowdfunding success: an exploratory study”, in Information Systems Frontiers, 1-16.