Factors Determining Crowdfunding Success
by
Floris Remmerts de Vries
MSc. Business Economics, Organization Economics September 2014 Gruttersdijk 42 Bis 3514 BH Utrecht The Netherlands +31 6 17 36 00 38 florisrdevries@gmail.com Student number: 10670149
2
Table of contents
1. Executive Summary ... 3
2. Introduction ... 3
3. Literature Review ... 5
3.1 New venture funding ... 5
3.2 Business angel funding ... 7
3.3 Motivations for participating as a crowdfunder ... 8
3.4 Eight critical factors ... 10
3.5 Hypotheses ... 11
4. Methodology ... 12
4.1 Survey Design ... 13
4.2 Coding and analysis ... 14
5. Results ... 15
5.1 Summary statistics ... 15
5.2 The difference between funded and not funded projects ... 17
5.3 The relation between Score and Funding ... 19
5.4 The presence of a critical flaw ... 22
6. Conclusion and Discussion ... 24
References ... 27 Appendix ... 29 Appendix A ... 29 Appendix B ... 30 Appendix C ... 32 Appendix D ... 33 Appendix E ... 34
3
1. Executive Summary
This study identifies important factors which help determine the success or failure of
crowdfunding projects, and contrasts them with how these factors are used by business angels deciding to supply funding for projects. The study shows that the assessment score used by Maxwell et al. (2011) to assess business angels decision making, also holds predictive power in crowdfunding situations. A major difference seems to be that crowdfunders are more “forgiving” than business angels when it comes to rejecting a project. Eight critical factors are used to
examine how crowdfunders come to their decision. Not all critical factors, like relevant
experience and financial model, have significant effects in crowdfunding. This may be due to the fact that crowdfunders cannot easily assess them on Kickstarter.
2. Introduction
Business today consists in persuading crowds.
– Gerald Stanley Lee (1862 – 1944), American congregational clergyman.
The quote by Lee, that persuading crowds is central to doing business, may hold true today more than ever. A good example is the Dutch chain of bookstores, Polare, that filed for bankruptcy in February 2014. A few of the stores managed to stay operational on their own however, by raising investments up to € 250.000 through Dutch crowdfunding-platform Symbid.
Recently, the phenomenon of crowdfunding has gathered significant momentum. Crowdfunding allows people seeking capital to raise their required investment from a large number of small contributions. The investors in this case are usually common individuals without experience in venture capital, most crowdfunders are not professional investors. Projects that seek funding from the crowd are as diverse as its backers. A well-known example is the website Kickstarter, where anyone can post a project and ask visitors for funding contributions.
A commonly used definition of crowdfunding comes from Belleflamme et al. (2010) who state that “Crowdfunding involves an open call, mostly through the internet, for the provision of
financial resources either in form of donations or in exchange for the future product or some form of reward to support initiatives for specific purposes”.
Whether crowdfunding is here to stay or will turn out to be only a contemporary
phenomenon remains to be seen. Possibly the introduction of stricter liquidity demands for banks, which is predicted to make it harder for startups to acquire funding from institutional investors,
4
leads more people to consider crowdfunding as a serious option for funding their projects.
Although crowdfunding is a very popular way of funding new projects, in scientific literature the phenomenon has not attracted much research yet. In recent years a few very good defining and exploratory studies have appeared, but theoretical research into the mechanics of how this new way funding exactly works is scarce. This study contributes by trying to fill a little part of the gap in the literature. A first insight into how crowdfunders come to their decision to support a project and how this compares to more traditional ways of funding is made.
In a recent study examining the dynamics of crowdfunding, Mollick finds that funders of crowdfunding projects respond strongly to quality signals that signal preparedness of the project founder. The better funders perceive these signals to be, the higher the chance that the project will reach its funding goal (Mollick, 2014).
A similar dynamic is found by Maxwell et al. in their study on decision making by informal professional investors, often referred to as business angels. They identify eight critical factors that predict whether a venture will receive funding from a business angel. If the project is rejected in the first stage of the decision making process it is flawed in one of these factors (Maxwell et al., 2011). The elimination-by-aspects heuristic used by business angels to quickly evaluate a project seems to be a departure from a more fully compensatory decision model. This study aims to find out whether backers of crowdfunding projects use this same decision heuristic to quickly come to a decision whether or not to fund a project and whether the decision process of crowdfunders is in line with business angel decision making. For this study a survey is used that allows respondents to rate projects from crowdfunding website Kickstarter and lets them indicate if they are willing to invest.
I show that crowdfunders display similar patterns as business angels when deciding whether or not to fund a project. However, there is also a difference: where business angels reject a project that is flawed automatically, a flawed project could still have a chance of securing funding from the crowd.
In the next section a review will be presented of relevant literature on crowdfunding and business angel decision making. Section four outlines the methodology of the study. Section five presents the results, and section six finishes with conclusions, discussion and limitations.
5
3. Literature Review
One of the most critical resources for new ventures is obtaining the necessary finance (Gompers and Lerner, 2004). However more than 90% of startups seeking investment fail to attract funding (Riding et al., 2007). Most commonly investors evaluate a business plan presented by an
entrepreneur and decide, on the basis of the plan and communications with the entrepreneur, whether investment in the project is profitable. For venture capitalists the venture idea, the market, the management team and the personality of the entrepreneur are often cited as important factors influencing the decision to supply funding (Hall and Hofer, 1993; Robinson, 1987). In this respect crowdfunding could provide a new way for entrepreneurs to find funding for new ventures. Whether crowdfunding acts as a replacement for traditional sources of finance, or whether crowdfunding will supplement them remains to be seen (Mollick, 2014).
Crowdfunding efforts can, when very successful, lead to funding from more traditional sources, (Mollick, 2014). Crowdfunding has been used in this way by the Pebble smart watch, Pebble was rejected by venture capitalists, but managed to secure venture capital funding after one of the most successful Kickstarter campaigns to date. When demand falls short and a crowdfunding campaign is unsuccessful, most entrepreneurs engaged in crowdfunding fail quickly, their projects have a very low chance of becoming fruitful (Dingman, 2013).
3.1 New venture funding
Usually a new startup does not have the ability to access debt financing. They lack the needed collateral or stable cash flows to ensure interest payment (Berger and Udell, 1998). So instead of relying on debt finance, most new ventures seek to gain some form of equity finance.
Traditionally, the most common forms of equity finance for new ventures include friends and family, venture capitalists, and business angels (Schwienbacher and Larralde, 2010).
Since VC funding and banks usually only supply larger amounts of capital, entrepreneurs that require small amounts of capital traditionally used to rely on friends and family or own savings as seed capital for their business (Winborg and Landstrom, 2001). Also business angels are known to invest in small projects, supplying seed capital for new ventures. More recently, crowdfunding has become a serious alternative in providing seed capital for entrepreneurs. It allows entrepreneurs to seek funding for their project without approaching financial investors such as BAs (Kleemann et al., 2008; Lambert and Schwienbacher, 2010). Obtaining finance from
6
VCs or BAs is often perceived as a daunting task. An overview of the most common forms of debt and equity financing is presented in table 1.
Table 1. An overview of the different forms of equity and debt financing Equity
financing
Investor Main investments
Entrepreneurial team Entrepreneurs investing own
money, or money obtained through personal loans.
Small new ventures / projects
Friends and family An entrepreneur’s friends and
family.
Small new ventures / projects
Business angels Wealthy individuals willing
to invest in small projects.
New ventures / innovative projects
Venture capitalists Specialized investors
gathering money from non-specialists and invest this in projects.
More mature ventures
Stock markets Public invests through a
public offering.
Corporations
Banks Bonds. Mature ventures / corporations
Debt financing
Leasing companies Providing resources against
lease payments.
Mature ventures / corporations
Customers / suppliers Different forms of supplier
and customer credit.
Mature ventures / corporations
According to recent literature crowdfunding would overlap with both friends and family and business angels when it comes to funding new ventures (Mollick, 2014; Schwienbacher and Larralde, 2010).
It may be speculated that VCs and BAs follow similar decision patterns, since they both often evaluate an entrepreneur’s business plan to decide whether to supply funding. When deciding whether to invest, venture capitalists and business angels often have to rely on limited information about the project they are evaluating. Claims made by entrepreneurs are often hard to verify by VCs and BAs (Zacharakis and Shepherd, 2001). Therefore it is believed that making the funding decision is often based in large extent on a gut feeling, where the personality of the entrepreneur and the relations between an entrepreneur and a funder play important roles (Riquelme and Watson, 2002).
To entrepreneurs in the earliest stages of their ventures, BAs seem to play a bigger role than VCs. Research shows that VCs usually invest in more mature ideas since they enjoy shorter exit cycles and lower levels of risk (Sapienza et al., 1996). For this reason it seems that
7
Research has shown that a certain “passion” displayed by entrepreneurs are very important for future venture success. Proactivity and passion for work seem to have a positive effect on venture growth (Baum, Locke and Smith, 2001). However, Chen, Yao and Kotha (2009), find that passion displayed by entrepreneurs when presenting their business plan does not have a statistically significant effect on a VCs decision to supply funding. The preparedness of an entrepreneur on the other hand does have a significant positive effect on a VCs funding decision. Similarly, Maxwell and Levesque (2011), show that entrepreneurs who are able to secure funding from business angels display a larger number of trust building behaviors. Furthermore they exhibit a smaller number of unintentional trust damaging behaviors. Entrepreneurs who did display a large number of trust damaging behaviors, only can receive funding with extra control mechanisms in place.
3.2 Business angel funding
The important role of business angels in supplying finance for new ventures has been demonstrated since the early 1980s by for instance Wetzel (1983). But although there are a lot of business angels and they supply a large amount of capital for new ventures, attracting funding in this way has been notoriously hard for entrepreneurs (Duxbury et al., 1996).
The way business angels make their decision whether or not to invest in the idea
presented by an entrepreneur has been studied in multiple ways. Van Osnabrugge (2000) adopts an incomplete contracts approach and states that BAs place emphasis on reducing risk ex post investment. In studying the way BAs make investment decisions van Osnabrugge (2000) uses over 20 characteristics that BAs would consider when they decide whether to supply funding. Similarly Sudek (2006) identifies a list of criteria that BAs use when judging on an investment opportunity. Trustworthiness of the entrepreneur, quality of the management team, enthusiasm of the entrepreneur, and exit opportunities for the BA are identified as the most important factors. Both van Osnabrugge (2000) and Sudek (2006) employ classic decision
analysis to develop a model of how funding decisions are made. They survey BAs with a long list of criteria, and ask them to rate their investment decisions. This method however could contain some serious biases. Mainly, that they require BAs to self-report on their interactions with entrepreneurs ex-post their investment decision, could lead to biased results (Maxwell et al.,
8 2011).
Maxwell et al. (2011) on the other hand theorize that a behavioral approach in studying BA decision making could yield more accurate results. In their study they review interactions of business angels participating on the tv-show Dragon’s Den. They are able to show that fully compensatory decision models are not always accurate in predicting a project’s success with business angels. They find that BAs use a decision heuristic known as elimination-by-aspects instead of a fully compensatory decision model. More importantly they define eight critical factors that a BA would use when making a decision. When a new venture is seriously flawed in one of these eight factors, a BA decides not to supply funding. Their model has a very strong predictive power. However, the external validity of the study may be doubted, since the participants in the research are contestants on a tv-show.
3.3 Motivations for participating as a crowdfunder
This study contributes by investigating whether crowdfunders display similar decision heuristics as business angels. To establish a framework for investigating these dynamics, I first discuss some motivations crowdfunders may have to supply funding for a project.
Mollick (2014) distinguishes four main types of crowdfunding. First he notes the
existence of a patronage model, in which crowdfunders act as philanthropists, expecting no direct reward for their donation. Second there is the lending model, where funds are offered as a loan and the funder usually expects some rate of return. The third approach is reward based
crowdfunding, which is the focus of this study. In this form of crowdfunding funders receive a pre- specified reward for their contribution. Finally funders may be awarded with equity stakes in the projects they invest in.
Agrawal et al. (2013) note five incentives for funders to participate in crowdfunding projects. These incentives are; access to investment opportunities, early access to new products, community participation, support for a product, and formalization of contracts.
Access to investment opportunities applies mainly to equity crowdfunding (Agrawal et al., 2013), which is not the focus of this study. In this respect they cite Gubler (2013) who notes that crowdfunding offers ordinary investors an opportunity to get early access and participation for the next big idea. In his Wall Street Journal article Gubler pleads for creative regulation to stimulate and develop crowdfunding as a way to finance startups and nascent entrepreneurs.
9
The second incentive described by Agrawal et al. (2013), early access to new products, is a very relevant factor for funders participating in projects on Kickstarter. Websites like
Kickstarter and others demonstrate an enormous demand for new and innovative products, and the chance to be able to obtain them before they are released through traditional channels. This incentive is very relevant for reward based crowdfunding.
On a website like Kickstarter community participation is the third major incentive for funders to join. In this respect Kickstarter does not only act as a platform for people to supply funding, but also as a social medium for funders to connect with each other and with project founders. Funders not only value the product itself they invest in, but also the feeling of participating in an entrepreneurial initiative and being one of a select group of early adopters (Schwienbacher and Larralde, 2010). Agrawal et al. (2013) also mention the fact that funders seem willing to provide funding in return for recognition from project creators within the
crowdfunding community. In a study of the website iStockphoto however, Brabham (2008) finds that participation is not primarily motivated by networking and peer recognition. Monetary factors and the willingness to improve skills and have fun seem to be more important factors in participating on the website.
Simply showing support makes up the fourth major incentive for participating in crowdfunding. Many projects that receive funding from the crowd offer no or relatively small rewards to their funders. It seems that receiving donations and drawing on the philanthropy of funders is a very common phenomenon in crowdfunding (Lambert and Schwienbacher, 2010). Participants in crowdfunding projects do not just have extrinsic motivation for participating, but also an intrinsic motivation (Kleemann et al., 2008). According to Agrawal et al. (2013) this intrinsic motivation is not only derived from community participation, but also from showing support for new products, services or ideas. This incentive is shown to be relevant for the lending crowdfunding model, as well as the patronage and reward based models. In all cases most funders do not just expect a reasonable rate of return for their donation, but mainly seem to support projects they think support a social good (Mollick, 2014).
The final incentive that motivates funders to engage in crowdfunding is the formalization of contracts. Whereas traditional sources of finance are often supported by professional services, funding from family friends appear in the domain of informal finance. Here a crowdfunding platform like Kickstarter acts as an official intermediary and formalizes these forms of informal
10
finance (Agrawal et al., 2013). Although there is no real estimation available on fraud in crowdfunding, formalizing contracts of informal finance could improve problems with fraud. Traditionally, funding from friends and family plays an important role in acquiring seed funding, or early-stage investment. Similarly, in crowdfunding early investors are friends and family of the entrepreneur (Agrawal, Catalini, and Goldfarb, 2011). Especially in these cases crowdfunding acts as a way to formalize contracts between new ventures and early-stage investors.This
incentive may be the strongest for funders considering to support lending or equity crowdfunding initiatives.
3.4 Eight critical factors
In order to demonstrate the use of elimination-by-aspects by crowdfunders a set of criteria should be specified that allow a crowdfunder to quickly reject the majority of possible projects. In this respect I follow Maxwell et al. (2011), who use eight critical factors in their study on BAs. The same eight factors have been used in an Innovators Assistance Program (Udell, 1989) and have been validated for predictive accuracy (Astebro, 2004). All eight factors include some criteria past literature shows to be significantly important when BAs judge possible projects for funding. A table shown in Business angel early stage decision making by Maxwell, Jeffry and Lévesque, 2011, clearly lists all eight factors, the criteria they include, and the relevant literature. A copy of this table can also be found in appendix A of this study. Below a table is presented listing the eight critical factors and their corresponding statements, as they are used in the survey.
Table 2. Eight critical factors and their corresponding survey statements
Factor Survey Statement
1. Ease of adoption Customers in the target market will easily adopt this product.
2. Product Status There is still a lot of work required before this product is ready for the market.
3. Protectability It is easy for people to copy this product.
4. Customer Engagement This product satisfies customer needs
5. Route to Market This product has a realistic route to market.
6. Market Potential This product has a large potential market.
7. Relevant Experience The creator(s) of this product has enough relevant experience.
8. Financial Model This product will make money.
The first three factors concern the product offered by a new venture; ease of adoption, product status and protectability. The second three factors are factors concerning the market a new venture operates in; customer engagement, route to market and market potential. The relevant experience factor concerns the entrepreneur, which can be of critical importance for ventures in
11
their earliest stages. Finally the financial model factor concerns financial aspects of a new venture.
Because elimination-by-aspects is not a fully compensatory decision model, it trades off speed for potential accuracy. It increases the chance of rejecting a project that could have received funding based on other criteria, but it also increases the speed and efficiency when reviewing a large amount of possible projects (Maxwell et al., 2011). The prediction is that when any of the factors mentioned above is flawed, a crowdfunder will pass on the opportunity to supply funding for the project.
3.5 Hypotheses
This study places the focus on a comparison between how business angels come to a decision about whether to supply funding for a new venture and how crowdfunders come to that decision. Past research has shown that there is a similarity in how more traditional funders and
crowdfunders assess entrepreneurial quality (Mollick, 2013). Furthermore Mollick (2014) also shows that crowdfunders respond strongly to signals of preparedness, which is line with other traditional funders (Chen et al., 2009).
Although research has addressed the way in which BAs form their decision, the same doesn’t hold for crowdfunding. The decision process of a crowdfunder is still somewhat shrouded in mystery. In this study I try to start addressing this issue by comparing the way crowdfunders decide to how BAs do it. Because of some similarities between crowdfunders and more
traditional sources for funding, like paying attention to signals of preparedness, it can be suspected that crowdfunder decision making is not that different from BAs.
These signals of preparedness seem to correspond with the eight critical factor framework Maxwell et al. (2011) adopt to investigate BAs. They find that BAs use elimination-by-aspects, a heuristic used to trim the number of available options and make the decision where to place their funding more manageable (Tversky, 1972). As with BAs it seems reasonable to assume
crowdfunders also do not use a full compensatory model, where higher levels of one attribute can compensate for lower levels of another one, when they are selecting a project for funding. Rather it may be speculated that crowdfunders too adopt some decision heuristic to quickly scan through a large number of projects, and only consider in detail a few projects with high potential.
12
and crowdfunders. Crowdfunding is as of today more a form of seed capital than a real supplier of serious finance for firms (Belleflamme et al., 2013). In order to secure BA funding an entrepreneur usually should have reached some milestones, while crowdfunding seems to be more attractive for new startups and acts as a way of mitigating the need of supplying an entrepreneur’s own finance and making use of informal contracts with friends and family (Agrawal et al., 2013).
Another major difference is that BA’s bring experience and expertise to a project. One of the most valuable aspects of receiving financing from BA’s is their personal involvement with a venture, and thereby the expertise they provide to an entrepreneur (Maxwell et al., 2011). This is an aspect that is lacking from crowdfunding, where entrepreneurs are reliant on their own
experience. Furthermore, since crowdfunding provides financing from a great group of funders, managing this group can often be more of a burden than a virtue (Mollick, 2013). Crowdfunding does supply an entrepreneur with valuable marketing information however. If a project is wildly successful this demonstrates a high market demand, since a great group of people are interested. This effect is obviously absent with business angels.
Because of these considerations I propose the following hypotheses.
Hypothesis 1. The higher a project scores on the critical factor assessment score, the higher the
project’s funding percentage will be.
Hypothesis 2. The chance a project will reach its funding goal is negatively influenced by the
presence of a critical flaw in the project.
Maxwell et al. (2011) find that when a business angels detects a critical factor of a project that is flawed, the project is automatically rejected. Their results show a very high accuracy in
predicting whether a project will receive funding after one or more of the factors have been established as a flaw. For this reason I propose the third and final hypothesis of this study.
Hypothesis 3. A project that displays a critical flaw will be rejected by crowdfunders.
4. Methodology
In this study an online survey has been used which lets participants rate five projects from the website Kickstarter. A selection of 60 projects from Kickstarter has been made, each time a
13
participant starts the survey he will randomly see five of these projects. The projects have a minimum funding goal of $ 10.000, as with the projects BAs evaluate in Maxwell et al. (2011).
4.1 Survey Design
Real crowdfunding projects from Kickstarter have been used in the survey. From the 60 selected projects 43% reached their funding goal for real on Kickstarter. This is close to the actual success rate of projects on Kickstarter (Mollick, 2013). In this way the sample of projects used in the survey match the real population as close as possible, in order to strengthen external validity. Basically the survey tries to simulate the “Kickstarter experience”, by presenting snapshots of real projects from the website as if respondents are surfing the website itself. This way their decision behavior whether or not to supply funding best approaches real crowdfunder behavior on Kickstarter. A summary of the projects used in the survey is shown in table 3.
Table 3. Overview of projects
This Study Maxwell et al. 2011
Number of projects 60 150
Number of funded project 26 16
Minimum goal $ 10.000 $ 10.000
Maximum goal $ 500.000 $ 500.000
From the 60 projects a screenshot has been made where any information concerning the actual funding of the project has been removed. This in order not to influence the decision of
respondents by giving them a reference point on which to adjust their funding decision. In order to test the hypotheses respondents were asked to rate five projects that were randomly selected from the pool of 60 on the eight critical factors. Each factor is represented by a question about the project. As in Maxwell et al. (2001) respondents could grade each factor of the project. Finally respondents were asked what they thought an average visitor of the website Kickstarter would invest and how much they would like to invest themselves. An example of a project as presented to respondents is shown in Appendix B.
Since respondents are randomly shown five projects from a total pool of 60 projects, they were allowed to take the survey more than once, and rate more than five projects. A check has been performed afterward ensuring that no IP address saw the same project more than once. However, internal validity may still be harmed by allowing people to rate more projects than others.
14
4.2 Coding and analysis
For each project a respondent had to grade the eight factors. These factors were presented as a statement with which a respondent could agree or disagree, see table 1. As in Maxwell et al. (2011) respondents could grade the factors with A, when there was clear advantage associated with the factor and C if the factor represents a fatal flaw. A grade of B represents a score in between. Respondents could discriminate in more detail by selecting A, B, and C annotated with either a + or -. This way a more clear discrimination between opportunities was possible.
For analysis the grades are translated into scores and a gap is created between each grade category, so A+=10, A=9, A-=8, B+=6, B=5, B-=4, C+-2, C=1, C-=0. The gap was created to highlight the importance of a change in category, especially when switching to the “fatal flaw” grade (C). In this respect I again follow Maxwell et al. (2011).
For each factor the scores of all respondents were averaged, providing a factor score on a scale of 0 to 10. For each project all eight factor scores where averaged, providing a total score for a project, also on a scale of 0 to 10.
A series of regressions will be performed in order to test the relationships between the main variables of the study. The first dependent variable considered is Funded, a binary variable, which is 1 if a project reached its funding goal on Kickstarter, and 0 if the goal was not reached. The second dependent variable is Percentage, this variable also uses data from Kickstarter, and denotes the percentage of funding a project was able to reach. A funded project reaches at least 100%, but is able to receive funding up to a pre-specified deadline, allowing the percentage to go much higher. Any project that has a percentage lower than 100% is not funded, and will not receive the money donated by crowdfunders.
The main independent variable used in the regressions are Total Score, a variable that consists of the total assessment score of a project, as described before. When the score goes up, the prediction is also the percentage and the chance of reaching the funding goal will rise. Also the eight factors’ scores independently are used as variables.
The other independent variables used in both models are Goal, a variable that consists of the funding goal a project has set to reach. When this goal is higher, or even set too high, it is reasonable to assume the chance of reaching it will be lower. Updates, this variable consists of the number of updates a project posted during its funding period. The more updates a creator posts for its project, the higher the chance its funding goal will be reached. Facebook, this
15
variable consists of the number of Facebook connections a project creator has. The more connections, the higher the chance of reaching a project’s funding goal. Days, this variable consists of the number of days a project has been posted on Kickstarter. The longer a project runs to gather funding, the higher the chance of reaching its goal. And finally Average Investment, an independent variable that is an average of how much the respondents indicated they were willing to invest in a certain project.
Finally both the effects of Total Score and the presence of a Flaw will be tested on the
Percentage and the Funded variables. This way I can gather evidence on which of the main
effects has a stronger predictive power, and thereby which decision model is more likely to be present in crowdfunding. Total Score more closely resembles a fully compensatory decision model, while Flaw more closely resembles the elimination-by-aspects heuristic.
5. Results
5.1 Summary statistics
In total 64 respondents took the survey, giving a total of 320 observations. After inspection two observations were dropped out because they clearly showed a non-serious response1. So in total 318 observations were taken into consideration, which approximates five ratings per project. The ratings per project are aggregated according to the description in methodology, giving 60 project observations. As a robustness check some analyses have been performed with the 318 individual observations, when noted these results are presented in Appendix D. First a summarization is provided of the main aggregated variables for the 60 projects in table 4. As noted before the project variables Goal, Updates, Facebook, and Days contain project data from Kickstarter. The variable Total Score is the total assessment score of a project, factors 1 through 8 are the
individual factors and Average Investment is the average amount respondents were willing to invest in a project.
16
Table 4. Summary statistics of main variables
Variable Mean Std. Dev. Min. Max.
Total Score 5,44 1,04 2,93 7,69 1. Ease of adoption 6,26 1,47 2,20 9,40 2. Product Status 4,05 1,73 1,00 7,80 3. Protectability 5,13 1,97 1,33 8,20 4. Customer Engagement 5,91 1,51 2,40 8,60 5. Route to Market 5,86 1,49 2,60 8,17 6. Market Potential 5,38 1,60 1,40 8,20 7. Relevant Experience 5,78 1,10 3,17 8,00 8. Financial Model 5,18 1,66 1,20 8,60 Average Investment 48,07 140,73 0 1.008,33 Goal 48.925,73 71.082,30 10.000 500.000 Updates 4,73 5,59 0 29 Facebook 481,62 664,72 0 2.866 Days 35,13 10,70 20 60
Main variables based on 60 aggregated project ratings.
As can be seen from table 4, the scores of the different factors vary widely, suggesting that there exists a wide spread in the ratings different projects received.
Also the average investment respondents were willing to supply vary widely between projects. It would be interesting to compare the average investment of respondents to the average donations actually received by the projects on Kickstarter. However, since respondents were allowed not to invest, the Average Investment variable is not a good comparison, since it can be 0, while the average actual donation for a project is never 0. Therefore respondents were also asked to estimate how much they thought an interested crowdfunder would be willing to invest in a project. Table 5 shows the results of a T-test testing the difference in means between the Actual
Average Donation and the Respondent Estimated Donation. Table 5. T-test for the difference in means
Variable Mean Std. Dev.
Actual Average Donation 157,44 279,00
Respondent Estimated Donation 101,40 125,70
Difference 56,04 282,39
Degrees of Freedom 59
T-Statistic 1,54
T-test for the difference in means between Actual Average Donation and Respondent Estimated Donation.
The table shows that the null-hypothesis, that the difference in means is 0, cannot be rejected at a reasonable significance level. Since the difference in means between the actual average donation and the respondent estimated donation is not significantly different from 0, it shows that
respondents are actually quite good at predicting how much funding a project will receive, giving their ratings a strong credibility and external validity.
17
5.2 The difference between funded and not funded projects
It is useful to take a look at the data while dividing the projects from the survey into two groups; funded and not funded. As noted before, projects that don’t receive their funding goal have a funding percentage lower than 100%, while projects that did reach their funding goal have a funding percentage of 100% or higher. When looking at the total score of the projects, it is clear that the projects that reached their funding goal on Kickstarter seem to get a higher score than projects that did not reach their funding goal. For instance, the funded projects have a mean score of 6,16 , while the not funded projects have a mean score of 4,90.
An overview of statistics for funded and not funded projects is shown in table 6, a table with a breakdown of the individual factor scores is presented in Appendix C.
Table 6. Statistics funded and not funded projects. Funded (Percentage ≥ 100%) Not Funded (Percentage < 100%) Significance of the difference in means N 26 34
Mean Total Score 6,16 4,90 𝑝 < 0,01
Median Total Score 6,18 4,87
St.Dev. of Total Score 0,77 0,89
Minimum Total Score 4,59 2,93
Maximum Total Score 7,69 6,58
# Flawed Projects 13 31 𝑝 < 0,01
% Flawed Projects 50% 91%
Respondents Prediction of Success 65% 44%
T-tests have been performed to test whether the means of Total Score and of Flawed Projects differ significantly. A T-test for the difference in means between the Total Score of funded and not funded projects shows that the means are significantly different from each other at the 1%-level, the same holds for the difference in means of number of Flawed Projects.
As can be seen in table 6, there is a higher percentage of not funded projects that display at least one critical flaw, (for instance at least one factor score lower than 4). This does seem to suggest that projects that are perceived to possess a critical flaw will not receive funding from the crowd. However, the fraction of funded projects that display at least one flaw is still 50%. This would suggest that the third hypothesis, that the presence of a critical flaw always leads to rejection, should be rejected.
When looked at how many respondents estimated if a project would reach its funding goal, we see that 65% of respondents estimated the funded projects would make it. While only
18
44% of respondents guessed the not funded projects would make it, again this suggests a strong capability of respondents to predict the success of a project, increasing the external validity of their ratings.
Below is a graphical representation of all 60 projects, where the percentage they reached of their funding goal is plotted against their total project score.
Graph 1. Projects’ percentage of funding plotted against total score, including trend line.
The graph shows a trend line indicating a positive relation between the score of a project and its funding percentage. In the next subsection this relationship will be estimated with OLS
regression. For now we take a look at the correlations between the total score, the presence of a fatal flaw, the funding percentage, and a binary variable indicating whether the project has reached its funding goal or not.
Table 7. Correlation matrix of the Main Variables
𝝆𝑿,𝒀 1. 2. 3. 4. 1. Percentage 1,00 2. Funded (Binary) 0,52*** 1,00 3. Score 0,52*** 0,60*** 1,00 4. Flaw (Binary) -0,52*** -0,46*** -0,59*** 1,00 *** denote significance at 𝑝 < 0,01.
As can be seen in table 7, the total score of a project seems to be positively correlated to the funding a project receives from the crowd in a moderate to strong way. Also the presence of a
19
flaw in a project is correlated in negative way, suggesting this harms the chances a project will reach its funding goal. Correlation matrices for all variables studied are presented in appendix D. Looking at these first statistics there seems to be some support for the first and second hypotheses. The data seems to support the hypothesis that a higher project score leads to a higher funding percentage. Also the second hypothesis, that the presence of a critical flaw is negatively related to the chance a project will reach its goal, seems to be supported. However, the third hypothesis, that the presence of a critical flaw leads to automatic rejection by crowdfunders, must be rejected.
The relationship between the total assessment score of a project and the funding of a project is examined in more detail in the next subsection.
5.3 The relation between Score and Funding
In order to test the relation between the score of a project and its ability to reach the funding goal, first a probit regression is considered. The dependent variable of the probit models is Funded, it is a binary variable with a value of 1 when a project has reached its funding goal and a value of 0 if it didn’t reach its goal.
In total two probit models will be considered. In the first model the main effect studied is
Total Score, which is the total assessment score of any given project. According to the first
hypothesis, a higher total score would increase the chance of a project being funded. In the second model the main effect studied are the scores of the individual critical factors, 1 through 8 (for an overview of the factors, see table 1). Again a higher score for any of these factors is supposed to increase the chance of reaching the funding goal. The other independent variables used in the regression models are discussed in the Methodology section.
In order to further examine the relationship between funding and the score of a project, four OLS models are considered. In the first two models the dependent variable is
Funded, the same binary variable as used in the probit models. In the third and fourth OLS
models the dependent variable used is Percentage, this variable consists of the funding percentage a project was able to reach on Kickstarter.
Table 8. Regression models
Dependent Variable: Funded Dependent Variable: Percentage
Probit Model 1 Probit Model 2 OLS Model 1 OLS Model 2 OLS Model 3 OLS Model 4
Total Score 2,183*** (0,482) 0,294*** (0,044) 1,461*** (0,563) 1. Ease of adoption 0,431* (0,338) 0,083 (0,061) 1,020** (0,424) 2. Product Status 0,883** (0,403) 0,090** (0,041) 0,558* (0,297) 3. Protectability 1,067*** (0,413) 0,085*** (0,032) 0,668* (0,356) 4. Customer Engagement 0,975* (0,549) 0,089 (0,068) -0,617 (0,524) 5. Route to Market 1,617*** (0,553) 0,125* (0,071) 0,247 (0,472) 6. Market Potential -1,301*** (0,339) -0,034 (0,047) -0,007 (0,306) 7. Relevant Experience -0,248 (0,551) -0,052 (0,052) -0,048 (0,369) 8. Financial Model 0,158 (0,430) -0,063 (0,077) 0,124 (0,474) Average Investment 0,001 (0,001) 0,002 (0,002) -1,02E-04 (1,75E-04) -1,23E-04 (1,64E-04) 1,81E-05 (0,002) 0,001 (0,002) Goal (Per $ 10.000) -0,288** (0,144) -0,583*** (0,223) -0,005 (0,007) -0,009 (0,009) 0,009 (0,039) -0,015 (0,031) Updates 0,177* (0,104) 0,329*** (0,125) 0,017** (0,008) 0,009 (0,009) 0,182** (0,072) 0,116* (0,068) Facebook (Per 100 friends) 0,118** (0,048) 0,103** (0,053) 0,019** (0,008) 0,014* (0,008) -0,022 (0,038) -0,029 (0,046) Days -0,013 (0,021) -0,043 (0,038) -0,002 (0,004) -0,001 (0,003) 0,033 (0,030) 0,032 (0,034) Constant -12,353*** (2,701) -18,553*** (3,785) -1,247*** (0,259) -1,352*** (0,309) -8,043** (3,211) -9,799** (4,257) N 60 60 60 60 60 60 Adj. R-sq. 0,60 0,77 0,50 0,60 0,35 0,46 Wald Chi-sq. 23,37*** 53,94*** F- statistic 23,84*** 17,21*** 3,13** 1,87*
21
According to the R-squared statistics, the first model explains 60% of the variation of the dependent variable, while the second model is able to explain 77% of its variation. These numbers suggest a reasonable fit for the two probit models. In the first probit model the main effect, Score, is significant at the 1%-level, lending support to the first hypothesis. The second model shows that the first six factors are significant at various levels, while the last two are not. Probit model 2 suggests that the factors for relevant experience and the financial model of a project have no significant effect on the chance to reach its funding goal. This may be due to the fact that both criteria are not easily gathered from a Kickstarter project. Entrepreneurs do not always provide detailed financial information on their projects, and experience is only measured in the number of projects an entrepreneur has launched and/or backed himself.
Product status, protectability, and route to market have a significantly positive effect on the chance to reach the funding goal. Ease of adoption and customer engagement have a positive effect, but are only significant at the 10%-level.
The sixth factor, market potential, has a highly significant negative effect, suggesting that projects with a higher market potential have less chance of reaching their funding goal. This seems to be a very contradictory result. Later on I find that when the sixth factor is flawed this does have a significant negative effect on the chance for a project to reach its funding goal. Both results seem to indicate an inverse u-shaped effect for factor six. A possibility is that projects that have a very big and promising market potential might be too complex to handle for starting entrepreneurs. Another possibility is that projects with a very high market potential are perceived not to need crowdfunding to supply funding.
Looking at OLS model 1, we see that the Total Score has a highly significant positive effect on the chance of reaching the funding goal. The regression suggests that for every standard deviation the average score increases, the chance of achieving the funding goal increases with 30%. Also the number of updates of a project positively influences the chance that the funding goal will be reached. The Average Investment respondents were willing to supply has no
predictive power however. OLS model 1 explains 50% of the variation in the dependent variable. OLS model 2 shows a breakdown per individual factor. The negative sixth factor is not significant in this model. The factors that do have predictive power are; product status,
22
account for a 9% increase in the chance to reach funding. The effect of route to market is even stronger, with 12,5%, however this result is only significant at the 10%-level. The R-squared of the model suggests 60% of the variation of the dependent variable is explained.
The third OLS model shows that the effect of Total Score on the funding percentage that is reached is highly significant. The coefficient of 1,461 seems to predict a rather extreme effect, suggesting that every point in total score accounts for 146% in funding reached, however this is mediated by the negative coefficient of the model, which acts as a hurdle that needs to be overcome. This model has a weaker fit than the previous models, since 35% of the variation in
Percentage is explained.
The final model, OLS model 4, shows a breakdown of the individual factors and their effect on Percentage. Ease of adoption, product status, and protectability, yield significant positive results. However, only ease of adoption is significant at the 5%-level. This model explains 46% of the variation in Percentage.
Judging from these results mainly the factors in the first two categories, product factors and market factors seem to be very important. The last two factors, concerning finance and relevant experience show not to have any significant results. This may however be due to the fact that these factors are very hard to judge on the basis of a Kickstarter project.
As a robustness check the OLS regressions have also been performed with the 318 unaggregated observations, where standard errors have been clustered either for projects or for respondents. These results can be found in Appendix E, and are in line with the results obtained from the aggregated ratings.
These results seem to support hypothesis 1, that a higher assessment score for a crowdfunding project has a positive effect on the funding percentage it will reach. In the next subsection the effect of the presence of a flaw will be examined in more detail.
5.4 The presence of a critical flaw
In order to establish support for the second hypothesis, namely that the presence of a critical flaw negatively influences the chances for a project to reach its funding goal three regressions are considered. The dependent variable is Funded, the same binary variable as in the first two probit models. For the main effect of this model a new variable Flaw is introduced. It is another binary variable with a value of 1 if the project contains one or more factors with a score lower than four.
23
The variable is 0 if the project contains no flawed factors at all. The rest of the independent variables of the model are the same as in the former models. First a probit model is considered to identify significant variables. In order to test the predictive power of the presence of a flaw, two OLS models are considered. The first uses Flaw as in the probit model, the second OLS model breaks this down in flaws per factor. Every factor is represented by a binary variable which is 1 if the factor is flawed and 0 if not. The results of the model are presented in table 9.
Table 9. Regression models
Dependent Variable: Funded Dependent
Variable: Percentage
Probit Model 3 OLS Model 5 OLS Model 6 OLS Model 7 OLS Model 8
Flaw -2,182*** (0,637) -0,426*** (0,143) -0,108 (0,161) -2,038* (1,124) Total Score 0,269*** (0,057) 0,986** (0,436) 1. Ease of adoption 0,009 (0,201) 2. Product Status -0,371*** (0,137) 3. Protectability -0,100 (0,147) 4. Customer Engagement -0,149 (0,244) 5. Route to Market -0,052 (0,157) 6. Market Potential -0,317** (0,128) 7. Relevant Experience -0,339 (0,235) 8. Financial Model -0,028 (0,250)
Average Investment -3,34E-04
(0,001) -1,88E-04 (2,09E-04) -1,32E-04 (2,60E-04) -1,26E-04 (1,66E-04) -4,74E-04 (0,002) Goal -0,222*** (0,067) -0,015** (0,007) -0,006 (0,009) -0,006 (0,008) -0,013 (0,037) Updates 0,160** (0,066) 0,024** (0,011) 0,022** (0,010) 0,015* (0,009) 0,144** (0,072) Facebook -0,021 (0,035) 0,003 (0,010) 0,009 (0,008) 0,019** (0,008) -0,033 (0,037) Days 0,001 (0,021) -0,002 (0,006) 0,002 (0,005) -0,002 (0,004) 0,028 (0,030) Constant 1,749* (1,013) 0,75*** (0,267) 0,552** (0,222) -1,001** (0,437) -3,416 (2,549) N 60 60 60 60 60 Adj. R-sq. 0,39 0,34 0,47 0,51 0,39 Wald Chi-sq. 16,46 F-statistic 9,15*** 9,28*** 23,09*** 3,05***
24
Probit model 3 shows that the presence of at least one flaw in a project has a significantly negative influence on the chance for a project to receive funding. Furthermore, a higher goal has a negative influence, while more updates for a project influence the chance in a positive way. According to OLS model 5 the presence of at least one flaw reduces the chance of reaching the funding goal with almost 43%, this result is significant at the 1%-level. The final model shows that especially the product status and the market potential have strong significant negative effects on the chance to reach funding.
These results show that the second hypothesis can be accepted, the chance of reaching the funding goal is significantly diminished by the presence of a critical flaw in the project. Although hypothesis 3 is rejected, since the presence of a flaw does not lead to automatic rejection (table 6), the results show that flawed project are in serious danger of not reaching their goal.
In order to see whether a fully compensatory decision model is used or whether crowdfunder are more likely to adopt elimination-by-aspects, OLS models 7 and 8 have been constructed. Model 7 shows that only Total Score has a significant effect on Funded, while Flaw has the right direction but is not significant. In model 8 both Total Score and Flaw have
significant effects on Percentage. From these results it may be suspected that a full compensatory decision model may be better in predicting crowdfunding behavior.
6. Conclusion and Discussion
This study examined similarities between the decision making process of traditional investors, business angels specifically, and the way crowdfunders make a decision to supply funding. The results show that some aspects of how crowdfunders assess projects are in line with how business angels come to their decision, suggesting that crowdfunders may act in a very close manner to traditional investors. However, whether crowdfunders use the same elimination-by-aspects heuristic Maxwell et al. (2011) show to be a good predictor for business angels decision making, cannot be thoroughly confirmed by this study.
The critical assessment score used in this study has a predictive power for a crowdfunding project’s funding percentage. A higher score leads to a higher funding percentage. Projects falling short in their assessment score will have serious difficulties securing enough funding to reach their goal. Crowdfunding projects that are perceived to display at least one flawed factor will have a significantly lower chance of reaching their goal. This is in line with the results Maxwell
25
et al. (2011) find. They show that if business angels perceive one of the factors to be flawed, the chance of receiving funding diminishes.
Although the presence of a flaw does decrease the chance of reaching funding
significantly, it doesn’t automatically lead to rejection. At this point my results deviate from the business angels. Maxwell et al. (2011) show that when a business angel perceives a factor to be flawed, the opportunity is rejected automatically. One could interpret this difference as
crowdfunders being more “forgiving” when judging a project. Using the terminology in Maxwell et al. (2011), crowdfunders seem to use a fully compensatory decision model rather than
elimination by aspects. Alternatively, when a project reaches out to the crowd to secure funding, it is easier to find a niche group of people that have enthusiasm for the project and are willing to overlook a possible flaw. Further research is needed in order to precisely establish the working of these decision making mechanics.
A limitation of the research presented in this paper is that only projects from the website Kickstarter are used. These projects use a reward-based crowdfunding model, meaning that a donation offers some small reward. Business angels on the other hand usually make a decision to fund a project on an equity basis, they basically buy a stake in the project. This equity model also exists in crowdfunding. It would be interesting to compare the decision making of crowdfunders of these equity projects with business angels. Possibly some of differences may disappear for equity based crowdfunding.
Another limitation of my study is that respondents were asked to rate projects and decide whether or not to invest on a hypothetical basis. This could harm internal validity, since
respondent may not always provide answers seriously. Future research could focus on an economic experiment, providing monetary incentives for making investment decisions.
Despite the above limitations, this thesis provides a number of lessons for people seeking funds through crowdfunding. First, factors concerning the product: ease of adoption, product status, and protectabilty, are important predictors for crowdfunding success. Also the market factors:
customer engagement, route to market, and market potential, are of importance to crowdfunders. With market potential however this could go both ways, for example a market potential that is too high could harm crowdfunding success.
Second, crowdfunders seem to be more forgiving on relevant experience and the financial model of the project. On a website like Kickstarter these dimensions of a project are often hard to
26
assess. Moreover when information about these factors is provided by the project founder, a crowdfunder has no way of verifying the truth of the claims made. Despite these facts,
crowdfunders seem willing to accept this and give project founders their trust on the basis of the six other factors mentioned before.
Finally, crowdfunding platforms may benefit from providing guidelines on how entrepreneurs present their projects with respect to product status, protectability etc. Also they could inform crowdfunders how to best assess these factors when deciding whether or not to fund a project.
27
References
Agrawal, A. K., Catalini, C., & Goldfarb, A. (2011). The geography of crowdfunding(No. w16820). National Bureau of Economic Research.
Agrawal, A., Catalini, C., & Goldfarb, A. (2013). Some Simple Economics of Crowdfunding. Innovation Policy and the Economy, 14.
Astebro, T. (2004). Key success factors for technological entrepreneurs' R&D projects. Engineering Management, IEEE Transactions on, 51(3), 314-321.
Baum, J. R., Locke, E. A., & Smith, K. G. (2001). A multidimensional model of venture growth. Academy of management journal, 44(2), 292-303.
Belleflamme, P., Lambert, T., & Schwienbacher, A. (2010, June). Crowdfunding: An industrial organization perspective. In Prepared for the workshop Digital Business Models: Understanding
Strategies’, held in Paris on June (pp. 25-26).
Belleflamme, P., Lambert, T., & Schwienbacher, A. (2013). Crowdfunding: Tapping the right crowd. Journal of Business Venturing.
Berger, A., & F Udell, G. (1998). The economics of small business finance: The roles of private equity and debt markets in the financial growth cycle. Journal of Banking & Finance, 22(6), 613-673.
Brabham, D. C. (2008). Moving the crowd at iStockphoto: The composition of the crowd and motivations for participation in a crowdsourcing application. First Monday, 13(6).
Chen, X. P., Yao, X., & Kotha, S. (2009). Entrepreneur passion and preparedness in business plan presentations: a persuasion analysis of venture capitalists' funding decisions. Academy of
Management Journal, 52(1), 199-214.
Dingman, S. (2013). Canadian’s smartwatch startup matches record $15-million in VC funding.
The Globe an Daily Mail.
Duxbury, L., Haines, G., & Riding, A. (1996). A personality profile of Canadian informal investors. Journal of Small Business Management, 34(2), 44.
Gompers, P. A., & Lerner, J. (2004). The venture capital cycle. MIT press.
Gubler, Z. J. (2013). Inventive funding deserves creative regulation. Wall Street Journal, January
31st 2013.
Hall, J., & Hofer, C. W. (1993). Venture capitalists' decision criteria in new venture evaluation. Journal of Business Venturing, 8(1), 25-42.
Kleemann, F., Voß, G. G., & Rieder, K. (2008). Un (der) paid innovators: The commercial utilization of consumer work through crowdsourcing. Science, Technology & Innovation
Studies, 4(1), PP-5.
Lambert, T., & Schwienbacher, A. (2010). An empirical analysis of crowdfunding. Social
28
Maxwell, A. L., Jeffrey, S. A., & Lévesque, M. (2011). Business angel early stage decision making. Journal of Business Venturing, 26(2), 212-225.
Maxwell, A. L., & Lévesque, M. (2011). Trustworthiness: A critical ingredient for entrepreneurs seeking investors. Entrepreneurship Theory and Practice.
Mollick, E. R. (2013). Swept away by the crowd? crowdfunding, venture capital, and the selection of entrepreneurs. Venture Capital, and the Selection of Entrepreneurs (March 25,
2013).
Mollick, E. R. (2014). The dynamics of crowdfunding: An exploratory study. Journal of Business
Venturing, 29(1), 1-16.
Riding, A. L., Madill, J. J., & Haines Jr, G. H. (2007). 13 Investment decision making by business angels. Handbook of research on venture capital, 1, 332.
Riquelme, H., & Watson, J. (2002). Do venture capitalists' implicit theories on new business success/failure have empirical validity?. International Small Business Journal, 20(4), 395-420. Robinson Jr, R. B. (1987). Emerging strategies in the venture capital industry. Journal of
Business Venturing, 2(1), 53-77.
Sapienza, H. J., Manigart, S., & Vermeir, W. (1996). Venture capitalist governance and value added in four countries. Journal of Business Venturing, 11(6), 439-469.
Schwienbacher, A., & Larralde, B. (2010). Crowdfunding of small entrepreneurial ventures. SSRN Electronic Journal.
Sudek, R. (2006). Angel investment criteria. Journal of Small Business Strategy Vol, 17(2), 2007. Tversky, A. (1972). Elimination by aspects: A theory of choice. Psychological review, 79(4), 281.
Udell, G. G. (1989). Invention evaluation services: A review of the state of the art.Journal of
Product Innovation Management, 6(3), 157-168.
Wetzel, W. E. (1983). Angels and informal risk capital. Sloan Management Review,24(4), 23-34. Winborg, J., & Landström, H. (2001). Financial bootstrapping in small businesses: examining small business managers' resource acquisition behaviors. Journal of Business Venturing, 16(3), 235-254.
Zacharakis, A. L., & Shepherd, D. A. (2001). The nature of information and overconfidence on venture capitalists' decision making. Journal of Business Venturing, 16(4), 311-332.
29
Appendix
Appendix A – Table from Maxwell et al. (2011)
Factor Criteria Included Research
Adoption Product Interest Mason & Harrison, 1996
Landström, 1998 Ridingetal, 2007
Benefits Feeneyetal, 1999
Mason & Harrison, 1996
Innovation Mason & Stark, 2004
Cooper & Kleinschmidt, 1987
Product Status Status Mason & Harrison, 2002
Technology Risk Mason & Harrison, 2002
Development Risk Cooper & Kleinschmidt, 1987
Protectability Protectability Haaretal, 1988
Role of IP Sudek, 2006
Other Barriers Landström, 1998
Customer Engagement Market Validation Mason & Stark, 2004
Cooper & Kleinschmidt, 1987
Customer Engagement Mason & Harrison, 1996
Route to Market Operations Mason & Stark, 2004
Market Entry Landström, 1998
Distribution Partners Cooper & Kleinschmidt, 1987
Mason & Harrison, 1996
Market Potential Market Size Feeneyetal, 1999
Cooper & Kleinschmidt, 1987
Market Growth Mason & Rogers, 1997
Balachandra & Friar, 1997
Market Competitiveness Mason & Stark, 2004
Mason & Harrison, 1996 Clark, 2008
Relevant Experience Industry Experience Sudek, 2006
Landström, 1998
Management Ability Feeneyetal, 1999
Mason & Rogers, 1997
Team Experience Van Osnabrugge & Robinson, 2000
Sudek, 2006
Mason & Harrison, 1996 Landström, 1998
Team Record Mason & Stark, 2004
Financial Model Cash Flow Feeneyetal, 1999
Profitability Feeneyetal, 1999
Realistic Forecast Mason & Stark, 2004
Mason & HArrisaon, 1996 Clark, 2008
30
Appendix B – An example of a Kickstarter project with survey questions as presented to respondents
32
Appendix C – Statistics of funded projects with a breakdown per individual factor Funded N.Funded
N 26 34
1. Ease of adoption Mean 7,14 5,59
St.Dev. 1,062217 1,383619
Min 5,00 2,20
Max 9,40 8,33
2. Product Status Mean 5,00 3,32
St.Dev. 1,444807 1,59 Min 1,50 1,00 Max 7,80 6,50 3. Protectability Mean 5,36 4,96 St.Dev. 1,765787 2,12 Min 2,67 1,33 Max 8,20 8,20
4. Customer Engagement Mean 6,78 5,25
St.Dev. 1,139553 1,42798
Min 4,50 2,40
Max 8,60 8,00
5. Route to Market Mean 6,88 5,09
St.Dev. 0,841671 1,422921
Min 4,60 2,60
Max 8,17 7,60
6. Market Potential Mean 5,84 5,02
St.Dev. 1,765787 1,832551
Min 2,67 1,40
Max 8,20 8,20
7. Relevant Experience Mean 6,18 5,48
St.Dev. 1,034176 1,061748
Min 4,00 3,17
Max 8,00 7,80
8. Financial Model Mean 6,08 4,48
St.Dev. 1,303662 1,585074
Min 3,67 1,20
33
Appendix D – Correlation matrices for all variables
𝜌𝑋,𝑌 1. 2. 3. 4. 5. 6. 7. 1. Percentage 1,00 2. Funded 0,52*** 1,00 3. Average Investment 0,00 -0,07 1,00 4. Goal (Per $ 10.000) 0,00 -0,18 0,03 1,00 5. Updates 0,40*** 0,44*** -0,09 0,01 1,00
6. Facebook (Per 100 friends) -0,08 0,13 -0,20 -0,23* 0,24* 1,00
7. Days 0,05 -0,03 -0,07 0,03 -0,06 0,23* 1,00 *** p<0.01, ** p<0.05, * p<0.10. 𝜌𝑋,𝑌 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 1. Percentage 1,00 2. Funded 0,52*** 1,00 3. Ease of adoption (1) 0,45*** 0,53*** 1,00 4. Product status (2) 0,40*** 0,48*** 0,45*** 1,00 5. Protectability (3) 0,26** 0,10 -0,14 -0,30** 1,00 6. Customer engagement (4) 0,31** 0,51*** 0,78*** 0,37*** -0,03 1,00 7. Route to market (5) 0,41*** 0,60*** 0,68*** 0,62*** -0,18 0,70*** 1,00 8. Market potential (6) 0,22* 0,26** 0,63*** 0,29** -0,22* 0,62*** 0,66*** 1,00 9. Relevant experience (7) 0,34*** 0,32** 0,48*** 0,35*** 0,13 0,47*** 0,53*** 0,32** 1,00 10. Financial model (8) 0,45*** 0,48*** 0,73*** 0,46*** 0,07 0,79*** 0,76*** 0,69*** 0,46*** 1,00 *** p<0.01, ** p<0.05, * p<0.10.
34
Appendix E – OLS regressions for 318 individual observations
Dependent Variable: Funded Dependent Variable: Percentage
Std. Err. Adjusted for 60 clusters in Project
Std. Err. Adjusted for 64 clusters in Respondent
Std. Err. Adjusted for 60 clusters in Project
Std. Err. Adjusted for 64 clusters in Respondent OLS Model 1a OLS Model 2a OLS Model 1b OLS Model 2b OLS Model 3a OLS Model 4a OLS Model 3b OLS Model 4b Total Score 0,090*** (0,017) 0,090*** (0,013) 0,427** (0,173) 0,427*** (0,121) 1. Ease of adoption 0,022 (0,015) 0,022 (0,015) 0,197** (0,094) 0,197** (0,085) 2. Product Status 0,030*** (0,010) 0,030*** (0,009) 0,138** (0,062) 0,138* (0,080) 3. Protectability 0,010 (0,011) 0,010 (0,009) 0,148 (0,102) 0,148*** (0,057) 4. Customer Engagement 0,015 (0,018) 0,015 (0,014) -0,149 (0,143) -0,149 (0,113) 5. Route to Market 0,039** (0,016) 0,039** (0,016) 0,151 (0,118) 0,151* (0,091) 6. Market Potential -0,023* (0,014) -0,023* (0,012) -0,049 (0,080) -0,049 (0,082) 7. Relevant Experience -0,014 (0,014) -0,014 (0,012) -0,028 (0,090) -0,028 (0,108) 8. Financial Model 0,011 (0,017) -0,011 (0,014) 0,090 (0,117) 0,090 (0,079)
Average Investment -3,01E-05
(-5,78E-05) -2,60E-05 (6,34E-05) -3,00E-05 (2,00E-05) -2,60E-05 (1,83E-05) 3,01E-05 (3,29E-04) 1,33E-05 (3,08E-04) -3,01E-05 (4,00E-04) 1,33E-05 (3,96E-04) Goal (Per $ 10.000) -0,009** (0,004) -0,008* (0,004) -0,009*** (0,002) -0,008*** (0,002) 0,016 (0,037) -0,016 (0,034) -0,016 (0,015) -0,016 (0,015) Updates 0,033*** (0,008) 0,029*** (0,009) 0,033*** (0,004) 0,030*** (0,005) 0,263*** (0,070) 0,244*** (0,067) 0,263*** (0,028) 0,244*** (0,027) Facebook (Per 100 friends) 0,003 (0,009) 0,003 (0,009) 0,003 (0,005) 0,003 (0,005) -0,099* (0,051) -0,101* (0,052) -0,099*** (0,023) -0,101*** (0,025) Days -2,69E-05 (0,005) -1,17E-04 (0,005) -2,69E-05 (0,002) -1,17E-04 (0,002) 0,045 (0,033) 0,043 (0,034) 0,045*** (0,014) 0,044*** (0,014) Constant -0,184 (0,195) -0,160 (0,191) -0,184* (0,104) -0,160 (0,111) -2,708* (1,495) -2,858* (1,702) -2,708*** (0,810) -2,858*** (0,869) N 318 318 318 318 318 318 318 318 Adj. R-sq. 0,30 0,35 0,30 0,35 0,26 0,28 0,26 0,28 F- statistic 22,64*** 14,81*** 41,73*** 28,96*** 3,86*** 2,49*** 20,87*** 11,47***
