So Escape the Nightmare has launched and we’re only a few days in but it looks like it’s going to be a close run thing. There were some tricky issues in the launch (some of which are still ongoing) and I’m sure at some point I’ll dissect those so that we can all talk about how they impact on things – but today I’d like to talk about ways of seeing the future and their impact both on me as a creator and on strategic decisions taken about the project as a whole. I’ll get onto the impact of a prediction later, first let’s talk about the method:

First go to the Kicktraq page and hit the “trending towards” button. Take a great big belly laugh, close it and look for another way to make predictions. That thing’s often wrong by as much as an order of magnitude.

I’ve heard a bunch of heuristics in the form of “If you hit 33.3% funding by day three your project will succeed” but these seem to be about as factually driven as suggestions like “Sacrifice a goat by the full moon and thy campaign shalt overfund to the third stretch goal.” There are some hard numbers in an older post by James Mathe, I’ll pull out those most relevant to my current campaign:

75% of successful projects have made more than 25% of their funding by day 3
98.7% of unsuccessful projects have not made more than 25% of their funding by day 3

It took me a minute to get why those numbers don’t sum to 100, when you get it you’ll see that these stats are a little less useful than they first appear – but it still seems that beating 25% funding by day three is a good indicator of health. Presently I’m 45 hours in and 22% funded so the odds of that seem alright. However none of those numbers can be twisted into a prediction of what will happen next – but I have a secret weapon.

I’ve done this before and kept relatively solid stats on what happened. Since my projects have a lot in common (using the same team, talking to the same existing audience, etc. ) they can represent a better model of each other than data pulled from KS databases at large. Sadly we suffer from small sample sizes, but you can’t have everything and that problem will correct itself with time. Hopefully 😉

My method is dead simple. For each completed campaign I calculate “On day X the campaign had reached Y% of the total it eventually hit.” Then I average the Ys for each day. Finally I plug in the actual value for the current campaign and see the total it will eventually hit if it’s following the average of the previous campaign’s curves. As a scientist I know that’s pretty sloppy and I should be doing some less naive curve fitting, but this achieves a good compromise between “giving the best predictions I’ve had” and “being quick enough that it’s not distracting to do”. Once the spreadsheet is set up it’s very quick – you just type in one number each day.

This generates one prediction each day, the average of the last few is the best prediction you’ve got at any time, but if the predictions are systematically altering in a particular direction that implies that the present campaign is going more or less well than the previous campaigns at this particular moment. This is handy because it makes it very apparent where there are differences in the nature of the long middle section of the campaigns – which often look unremarkable compared to the dramatic changes at the start and end of a campaign.

So what’s our destiny? Too early to tell at the minute, since we’re a couple of days in and so just have a couple of predictions. Using the method in the previous campaign generated early predictions that were out by 30-40% so anything it’s spitting out is probably correct to an order of magnitude but cannot be trusted more readily than that.

That’s unfortunate because the predictions are very close to our funding goal. If the method’s producing an underestimate then we’ll fund comfortably, if it’s an overestimate then it’s likely that I won’t get to make my horrific game about nightmares this year. That would be really sad since I’m very happy with the game itself, but even that wouldn’t be the end.

Which brings me onto what the point of making these predictions is: Having a good estimate of what will happen next lets you take appropriate actions. A project that will wildly overfund rewards speaking to manufacturers *before* it overfunds to have properly planned and budgeted stretch gaols to announce when it does. A project that will fail rewards being honest with your backers about the odds of success and looking towards using the Kickstarter as a platform to gather feedback about how to improve the game for a future launch and to gather some support together to get it done. A project in the middle ground rewards greater efforts to find ways to beat the progress curves of previous campaigns.

I also think that frank and honest communication with supporters is important, it’s a poor show to mislead people who’ve shown some faith in your ideas. I don’t want to be delusionally stating “Don’t worry, we’re nearly there!” when we’re not going to make it, it was also handy to be able to go “Your prediction for the future doesn’t match the data” when we had someone irrationally doomsaying in the last campaign. In this instance it lets me know that morale is going to be important. At the edge of the curve funding won’t happen until the last couple of days of the campaign, it’d be easy for new people to skip on past thinking it’s not going to fund and it’d be easy for backers to lose hope and withdraw (not necessarily their pledge, but just to stop talking about the campaign).

Knowing that there’s a chance and a good chance is important. It gives me the confidence to keep to my planned convention schedule and to keep on working with backers and finding ways to shout about the project. It’s much easier to fight if you know you can win 😉