If you follow the GameMarx podcast, you know I’m pretty big on watching Kickstarter projects. You might also know I like to play with data. So it should come as no surprise I’ve built a database of Kickstarter project stats – 58,671 projects to be exact.
I recently gave a presentation using this data at the Technology Cooperative and will probably be doing this talk a few more times (it’s more than just stats). Initially I planned on keeping the data semi-private out of concerns on how Kickstarter might react. Recent articles such as this infographic and sites like KickTraq have so far been left alone by Kickstarter, so I feel comfortable in publishing my own efforts.
Kickstarter’s Hidden Data
Before I get to the details I need to point out that Kickstarter does not make it easy to get a complete picture. Search results and project listings on Kickstarter are biased toward live and successful projects. Some projects will not turn up in Kickstarter search results at all – for example no combination of keywords will reveal this project. Further, liberal use of meta “no index” tags on the site tell honest webcrawlers (like Google) not to index their content.
What does this mean? It means while I know about 58K of the 64K projects on Kickstarter, my results are biased to the positive. The net effect is I show 48% of projects are successful, while the true rate published by Kickstarter is 44%. The charts below are not exact, but can be used to get a general idea of what is a “best practice” for a Kickstarter project.
My efforts so far have been focused on how someone should structure a Kickstarter project to increase the chances for success. The basic approach is to take a question, like “does a video improve the chance of being funded?”, and compare it to the average success rate (in my case, 48%). If the new success rate is above 48%, then we have a winner! So the video question above looks like:
Projects with a video have a 52% success rate, while those without have a 35% success rate. Compared to the baseline 48%, I’d say you want to have video. Keep in mind this does not account for the subjective quality of the video, just if the project has a video or not.
The next question is “Does Project Length have an effect on success rate?”. When you launch a project you can choose to keep it open as long as 90 days. Here are the results:
There is clearly a trend in project length. After 45 days or so you start to do worse than the baseline, but there is a real sweet spot in the 8 – 34 day range. You may note I‘ve used a polynomial regression here instead of the more common linear. Everyone has the preferences, and I like a poly in this case to highlight the sweet spot that would be lost in a straight line. I use a linear when I want to forecast new values along the Y or X axis – and that doesn’t make sense in this case.
Next question: “Does Funding Goal have an effect on success rate?”
This is almost a perfect linear relationship – the more you ask for, the less likely you are to get it. Note on this chart I did my own amount grouping, as I wanted to see more detail in the lower goal tiers. Below 5,000 is grouped by 500, 5,000 – 10,000 by 1,000, 10K – 30K by 5K, etc. I played with a few other patterens, but the overall shape never changed. Bottom line, asking for more that $5,000 drops you below the baseline.
Next Question: “Do the number of Reward Levels have an effect on success rate?”
I was surprised by this one – I didn’t expect such a strong relationship. The take home here is you want to have 7 or more reward levels!
Next Question: “Does Description Length have an effect on success rate?”
I expected to see projects with short descriptions do poorly, but I was curious if long descriptions hurt a project. I’m glad to see they don’t, so be verbose! The counts are not character counts, but rather HTML counts, so they are a little inflated. The crossover point is 2000 HTML characters (let’s say 1500 real characters) or a few paragraphs.
Next Question: “Does FAQ Length have an effect on success rate?”
Kickstarter has an option for a project FAQ at the bottom of the description. The chart shows just having an FAQ helps, but most (48K) projects have no FAQ at all. Because of this, I’m going to say while it’s good to put some content in FAQ form, I’m not going to count it in my stats.
Okay, now to the real question based on the above – What is the success rate of a project that has a video, is 8 – 34 days in length, has a goal of $5,000 or less, has at least 7 reward levels, and contains at least 2,000 HTML characters in the description (again,keep in mind the positive bias)? 73%!
Now some caveats! These stats are for all project categories. I’ve already played with the data enough to know that the sweet spots will vary between categories, and between subcategories. As mentioned several times before, the dataset is biased and it’s not looking likely that will change unless Kickstarter releases the data themselves. Finally there are several ways to slice the data and while I feel pretty good in my approach others will disagree.
If you are planning to launch a project my advice is to consider the data above, but to then do your own reasearch. Find similar projects and get a feel for why some worked and others didn’t. Many aspects of a successful project cannot be queried easily from website data, so this is just one bit of information to consider and all of this is probably trumped by a good social media campaign promoting the project.
Oh, one bonus chart below. Here are the funding levels for the successful “ideal” projects as defined above:
Most of those projects raised 150% of their goal (I left off the sub $500 level because the numbers are wacky when you have a $20 goal). It’s not a guarantee, but it’s nice to know when you’re loved, you’re really loved.
Finally, if you wish to take a crack at analyzing the data (or verifying this post!) please contact me (you can find my contact info on the site). The data format is an MS SQL database (works with the free Express edition) and sorry, I’m not going to convert it for you, it’s a lot of data. I ask that you be willing to share anything you find, and I’ll update this post with a link to yours below.
Update: I've posted a Video Game specific version of these stats over on GameMarx.