As crowdfunding marketing experts at Command Partners, we’re always keeping our eye out for new measurement techniques to better guide our marketing strategies. We have been longtime fans of Sidekick, and we were excited to have the opportunity to interview the man behind some great Kickstarter metrics. Sidekick is a website that shows the probability of success of live Kickstarter campaigns, computed from various features extracted from the Kickstarter website and Twitter.

1. What is your background – have you run successful projects before?

I am a backer (I backed 7 projects so far, mostly video games and hardware), but have not created any project.

2. What led you to create Sidekick?

Ever since I started using Kickstarter, I have been fascinated by the projects that “go viral”, and reach several times their goal (for example, the Oculus Rift project). I wondered if it would be possible to detect which projects are going to “explode” like that, and what amount they would reach. Moreover, I also wondered how soon it would be possible to make this prediction. So, I started gathering data, and began by focusing on a simpler problem: predict if a project will reach its goal or not.

The reason I built a public website showing the predictions was to give feedback to project creators as fast as possible, to allow them to react if things are not looking good.

3. Why is Kickstarter data so important?

Kickstarter is a representative example of a crowdfunding platform, that is used by millions of users and that allowed to raise nearly one billion USD. Thus, I am pretty confident that any findings extracted from Kickstarter data would also apply to similar platforms. Moreover, understanding the dynamics of crowdfunding is very interesting, as this mean of funding becomes more and more popular, but also because the analysis could be extended to other types of funding.

4. How can project creators use the data to their advantage?

First, they can use the stats page of sidekick to have an idea of the success rates in different categories, and how much money people ask on average. This can help them finding the right amount to ask. Then, they can search for similar projects, and see how they are doing, and what probability of success sidekick gives them. They can use this information to see what the successful projects have in common, and then apply these findings to the design of their campaign. Finally, once they have launched their campaign, they can check the feedback sidekick gives them, and react accordingly. For example, if the chances of success are low, they could invest in ads (either on Google or on social platforms) to promote their project and raise their chances of success.

5. How accurate are your predictions?

The accuracy varies over time. The very first prediction (about 4 hours after the launch of the project) is 76% accurate. Then, it constantly increases, as the predictions are refined when new data comes in. After 20% of a campaign, the predictions are 87% accurate. After 50%, it is about 91% accurate, and continues increasing until the end.

6. What factors do you use to gauge the success probability of a project?

I use a combination of multiple factors:

  • project attributes (goal, duration)

  • amount of money pledged

  • number of tweets, retweets, and replies on Twitter about the project, as well as the number of persons who tweet

  • number of backers, as well as how many other projects these backers have backed, and, among these other projects, what proportion has succeeded.

7. How often is the success probability updated on Sidekick?

The probabilities are updated several times per day, but I might increase the rate.

8. I understand that social media plays a big part in the analysis, but does this also account for large press hits?

No, I do not take into account factors such as the popularity of news articles, or the amount of press coverage. While this is somehow reflected on Twitter, it would require collecting data from several news sources, depending on the project category, which take a lot of time.

9. How accurate is Sidekick in a niche where there is less social activity, like Art or Children’s Books?

I haven’t looked at the prediction accuracy in individual categories, as the predictor in its current form does not differentiate between categories. Because of that, I expect the accuracy to be somehow similar between categories, but I do not have the data to support this claim at the moment. A planned extension of the models will take the category into account, though.

10. What do you believe the most successful product category is on Kickstarter?

To see that, you simply need to have a look at the stats page of Sidekick, to see that (surprisingly) Dance projects have the highest success rate, with more than 75% of successful projects.

11. What could all projects do better to help their chance of success?

I think that the most important part of any crowdfunding campaign is to engage its audience. You should start way before the launch, by preparing a teaser website and a Facebook page, already gathering some “fans” and generating some interest. Ideally, people should already know what your campaign is about before you launch it. Then, once it’s launched, you should make sure to keep the backers updated, by posting updates on the different medias, and interacting on the social platforms. Finally, you should make sure that your campaign is of high quality, with a detailed description (especially about the usage of the money), presentation video, and carefully design rewards.

12. Based on all of your experience and data, what is the single most important factor for a Kickstarter campaign to consider?

The most important factors is backers, of course 😉 ! To get backers, they need to be aware of your campaign, and to make them aware, you should be present on social medias. So making sure that you facilitate sharing your project (on your website, as sharing on Kickstarter is by default), and that you engage people on social medias, you maximise your chances of success. However, as good as your social media campaign is, it will not compensate a bad project or an incomplete project description ;), so make sure these are great, too!