This project will analyze previous kickstarter campaigns in hopes of gaining insight into what factors lead to a successful campaign.
The purpose of this analysis on kickstarter campaigns is to assist my friend Louise with her own kickstarter campaign for her play. We hope to gain insight into factors that may have helped lead previous campaigns to success and emulate those factors in order to improve Louise's chances of success.
For the analysis of kickstarter campaigns based on launch date I began by creating a 'year' column in my kickstarter data. To create the year column, I simply used the year function to extract the year from the 'Date Created Conversion' column we created earlier. Now that I had a year column, I was ready to begin my analysis. I created a pivot table with outcomes as my rows and count of outcomes as my values. I added the 'Date Created Conversion' column for my rows and removed the Years and Quarters categories that were automatically added. Once those additional categories were removed the pivot table had the months of the year as the rows, and this was what I needed for my analysis. Once I had my pivot table created, I added filters for 'Parent Category' and 'Year'. I filtered parent category down to theater because Louise's campaign is to fund a play. When my pivot table was filtered the first thing I noticed was that summer months were the time where most campaigns began. And although the number of failed campaigns increased slightly during these months, the number of successful campaigns increased drastically, showing these months were the most likely to lead to success. I then created a line chart with months of the year on the x-axis and number of outcomes on the y-axis. With this chart I quickly noticed December was the worst month to start a campaign. December was the first month when the number of successful campaigns was almost exactly the number of failed campaigns. This trend makes sense to me. People tend to be more active and positive during the summer than winter. Winter also has some high expenses with the holidays, so it makes sense to me that people may be unable to contribut to kickstarter campaigns leading up and following the holidays.
For my analysis of outcomes based on goals I began by creating a new sheet with the defined goal ranges. Using a countif function I then found the number of successful, failed, and canceled campaigns in each goal range. I summed up the number of successful, failed, and canceled campaigns in each range to find the total amount of campaigns in the range. Now by simply dividing the number of successful campaigns in a range by the total number of campaigns in the range I found the percentage of successful campaigns in the range. I then repeated those steps to find the percentages of failed and canceled campaigns for each range. At this point I had the data I needed to begin my analysis. I created a line chart with the ranges on the x-axis and percentages on the y-axis to see if there was any trend as the goal amount increased. The first thing you notice is that the most successful campaigns are the ones in the lowest range, under $1000. You can also see that the two highest ranges are also the least successful. For the most part the data follows this trend of the higher campaign goals being less successful. The exception to this trend occurrs in the range of 35 and 45 thousand. In this range there is a spike in successful campaigns. I was unable to justify a reasoning for this spike in successes and it may simply be an anomoly in the data.
One of the major challenges I encountered in my analysis was interpreting the spike in success of campaigns with goals between 35000 and 45000. It may just be a strange anomally occurring in our data set, but other than that it is difficult to properly justify any explanation for this sudden increase when the trend was going down. Another possible challenge was the missing data point for canceled campaigns in October. It may be interpreted as missing data or another issue with the data instead of a zero.
- What are two conclusions you can draw about the Outcomes based on Launch Date? Summer is the time of year with the most kickstarter campaigns launched, as well as the most successful season for campaigns. Holiday season at the end of the year, especially December, is the slowest time of the year for campaigns to launch, and campaigns started in December are the most likely to fail.
- What can you conclude about the Outcomes based on Goals? Small campaigns with goals under $5000 are the most likely to succeed. Except for a range between 35,000 and 45,000, it appears as your campaign increases its goal, the likelihood of success decreases.
- What are some limitations of this dataset? One major limitation of this dataset is that we have no information on donors. I think some sort of demographics information would allow us to see trends in who is donating to kickstarter campaigns, and particularly to theater campaigns. Having information on who is donating would be incredibly helpful for Louise to know who to focus on for her kickstarter campaign. I also think being able to drill down further in the theater category would help our analysis. Having it broken down into genre or region/state of the country would provide great insight into how to run Louise’s campaign.
- What are some other possible tables and/or graphs that we could create? We could look at average donations to see if there is helpful information to be gained. A line chart like we did with outcomes based on goals would work. We would put ranges of average donations on the x-axis and percentages of outcomes on the y-axis. This could guide how we reach out to people by informing us if successful campaigns have lower average donation amounts from more people or higher donation amounts from a more limited number of people or if there is no indication either way. We could also do a scatter plot with average donation amount on the x-axis and the total amount that campaign gained on the x-axis. However, this scatter plot would only give total amounts gained and not answer whether the campaign was successful or not. We could have also made a pair of bar charts to see the numbers of outcomes when a campaign was a staff pick or when it was a spotlight campaign.