We ended up with a case of tunnel vision: We poured all our resources into making sure that our prototypes were ready to go into manufacturing - because the project was already so far over due, we focused single-mindedly on a single milestone: Getting the final version of the software and hardware ready, so the manufacturing phase could start.
In the process, we took our eyes off the bigger picture, and we lost overview of how much the manufacturing itself was going to cost.
We only discovered quite how fast we were running in the wrong direction once we had a full and final prototype (towards the end of last year), as this was the first time we discovered that our estimates around the Bill of Materials (i.e. all the parts that go into the product) and NREs (non-recurring engineering costs) were way off.
By the time we discovered what the exact situation was around the end of December 2014, the damage had already been done: The invoices for development costs had all been paid, and there wasn’t enough left in our Ada war fund to be able to deliver on our Kickstarter campaign.
In retrospect, we then made a poor decision in not announcing our predicament to our backers immediately: We hoped that we would still be able to salvage the project. We tried to get additional pre-orders, we spoke with our bank, we spoke to potential investors, and we looked at all other sources of funds we could think of. Unfortunately, it wasn't enough, and we decided in late February that we had to cancel the project completely.