Kickstats: Launch Day vs Last Day

Kickstats: Launch Day vs Last Day

Photo by Aw Creative on Unsplash


Can you predict a project's chance of success, based on its funding level at any point during the campaign?

That's the question I've been trying to answer for the past few months. It makes sense that if a campaign reaches 90% funding on Day 1, it's basically in the bag to succeed. But what about 10% by Day 1, or 20% by Day 3? Is there any way to make an accurate prediction of that campaign's success?

I don't think it's much a stretch to say "Yes, you can predict a campaign's success based on it's current funding level." In fact, it's probably one of the easiest and most obvious predictions to make, but I'd still be interested in seeing what they actually are.

Predicting success by current funding

Well, after several months of collecting, cleaning, and analyzing Kickstarter campaign data, I can safely say that we can, indeed, predict success based on a campaign's current funding level... but we still don't have enough data.

Although initial results have been promising, my true goal is to have a table that Kickstarter creators can go to look up what day their campaign is on and what funding level they're currently at, and the table will show them a general prediction of their success. Of course, that creates a lot of table cells, which means we'll need a lot of data before it becomes reasonably accurate.

Unfortunately, we're not there yet, but I'm very excited to show it to you once it is ready!

Comparing the first and last campaign days

With the prediction table not being ready yet, I started thinking about other ways to analyze that we may have enough data for now. Other things like whether or not there's a relationship between how well your launch day goes compared to how well your campaign's last day goes. I was hoping they would be fairly well correlated, but as it turns, they're not even close...

Here's a scatter plot of launch day performance vs last day performance for 2020 campaigns that were 30 days long. To try cleaning this up a bit, I filtered out extreme cases (campaigns in the bottom or top 5%). As you can see, the filtering didn't help much - it's still a mess!

Kickstarter campaign funding received on first vs last day

Just to drive this point home a little more, lets zoom in a bit, by limiting the x-axis range to match the y-axis. There really is no correlation there!

Kickstarter campaign first day vs last day

I did notice one thing about the first graph, however. The fact that the y-axis (last day performance) goes from -20% to 120% while the x-axis goes from 0-710% seems interesting. What that at least suggests to me is that a Kickstarter campaign's launch day is probably more important (i.e. you get more funding) than the last day.

To be honest, it's not all that surprising that your launch day is likely more important than your last day, but what the above also shows is that we may have enough data now to quantify how much more important it is.

Well, it may not be anything special, but here is a chart comparing Kickstarter campaign's funding levels for their first day vs their last day.

Kickstarter launch day compared to last day

This chart includes all campaigns that started and finished year-to-date 2020 that were 30 days long and did not cancel earlier than 12 hours before the scheduled end of campaign. Altogether, that included almost 3000 campaigns. I limited to only 30-day campaigns because it stands to reason that the length of your campaign could easily influence the relationship between your launch day and last day.

As for how to actually read the chart, it's a standard box plot - although it may not look like it. If you look at the box on the left, you'll see it has a shaded box region, a line in the middle of the box, and a "whisker" on top. The bottom and top of the box are the 25th and 75th percentiles, respectively. The line in the middle of the box is the 50th percentile, otherwise known as the median. The top of the whisker is the 90th percentile. These points tell you how many campaigns were at that funding level or lower by the end of their first day. 

The box plot for the last day reads the same way, except that it's been smooshed even more. For that one, the median is just above 0%, and the 75th percentile is around 7% funding.

There are bottom whisker on these plots as well, but they're so short you can't even see them.

What this shows us, as the graphic states, is that your Kickstarter campaign's first day will likely see much more funding than your last day does. How much more? Expect maybe somewhere around 2x more funding on your first day than you'll get on your last day, although the variability there is quite high.


Write a comment
David Miller

David Miller

I’m having a hard time reading your first graph (but I also have a migraine). As you mention the scales affect that. I’d love to see the scales even and dump the outliers but that’s more of my lack of exposure to various graph types.

To me, from my reading of Stegmaier and from our own campaigns, only the first day or two and the last day or two make the difference. And you make that point to. Although, your second graph really surprised me (the lat day may only be a partial day, not sure how much that affects it.

Our Mini Apocalypse campaign seems to illustrate (with one odd “big” mid -campaign day) what you discuss and what I’ve always thought of as a “normal” Kickstarter. First day has to be big and the last two days will see last minute folks who convert their dollar pledges plus fence sitters jumping on board.

One thing I’d love to see graphed are Kickstarters that only run in a a month. In other words, they start and end in the same month. Seems like that causes a sense of urgency because I can’t wait and “check it out next month”.

I may have missed it, and will look at your posts, but shorter duration seems to do better and I wonder what’s ideal—10 days, 15, 18, 20, etc.

Thank you for the always provocative and well-written blog posts. You’re really good.



Hi David,

I like your suggestion about making the ranges of the first graph the same; it would really help to show just how little correlation there is between the first and last day. I’ll see what I can do and might even have something up later today.

As far as the second graph, though, most of the last days are right around 24 hours, since I defined “day” as “24-hour period” rather than an actual calendar day. Of course, many of them I had to round up or down, but for most part they’re pretty close to a true 24-hour period.

And you’re right about what days matter. It’s usually just the first 2-3 days and last 2-3 days that see any real movement in funding. I think that’s why the general advice is to have shorter campaigns, since lengthening them only increases the middle section where not much is happening anyway.

The ideal length is difficult to say, and likely hinges on a number of factors, but I actually wrote a post about how campaign length appears to affect success rates. You can read it here if you’d like:

You’re idea about campaigns that stay within the same calendar month is interesting as well. I may just save it for later :).

David Miller

David Miller

Thanks for the link, I somehow missed that one or forgot! Derr . . .

I’m digging in now, I love your analytics and perspective.

David Miller

David Miller

Here’s the Stegmaier link and I forgot about his payday thoughts. I subscribe to a Tuesday launch and Sunday ending . . .



I love Jamey’s posts! They’re actually what inspired me to start this blog – I wanted to see if his hypotheses ended up being correct when we actually analyze the data. So far, everything he’s said has been more or less accurate, I just need more data before I can write about some of the topics.

The big one I’ve been wanting to analyze has been what day of the week you launch. I actually have a slightly different perspective from some of the big names on that one and I’m dying to see what the data says. I have to be careful with it though. It’s such a hot topic that I want to make extra sure I’ve thought everything through before I draw any conclusions.

Also, I included another version of the first chart, where both axes have similar ranges. I hope it makes it a little easier to see that there’s just no correlation going on there at all!

Contribute to the conversation!