A small introduction about me, this is my first post. I 've heard about mycroft from some time and i really value the project but I am not a backer or a buyer yet (for reasons I d rather not get into here).
While browsing the mycroft site I saw this thread and read it. Decided to weight in as a healthy funding model is imperative not only to the success of an open source project but that the people behind it don't have to make compromises from their initial "freedom specifications" in order to not have the project go down.
In my opinion a subscription model would be very hard to justify. Moreover from a customer perspective I prefer one-time fees as they help me manage my finances better.
The idea of an app store looked very good from the first time I read about it from gymboh32's post.
Google playstore is a famous example but I have reservations about how selling closed source apps would align with a open source projects (I remember reading for example an article about google abandoning development on some open source apps on AOSP and "giving for free" proprietary versions from the playstore - apologies for not remembering a source on this). That doesn't say it couldn't work. What I think people should also consider are cosmetic changes on the app store. For example having the voice of your favourite actor, or super hero (for kids) reply to you. Commercial deals with cuts for everyone involved can make this happen. The popularity of cosmetic products on "free to play" games means this can create a revenue, though an expert on the subject would be needed to determine how much of your operating costs it would cover.
Another very good idea has been mentioned by Neon_Samurai. Outsourcing infrastructure to the users would be very helpful for all involved and distributed nodes has a lot of advantages. However in order for this to be viable a lot of things need to be considered.
One thing, that I suspect is more important to early backers than the average consumer, is that with this way the company does not become a potential central point of failure (Revolv hub comes to mind).
Moreover keeping your data at your own physical location goes a long way in ensuring one's privacy
However one has to consider the financial viability of such a choice.
First and foremost for mycroft this means that they do not carry the burden of infrastructure cost but it instead is shared among the users.
Even updated packages and files could be shared in a distributed/utorrent like way to lessen the need of mycroft having to have a high upload server running 24/7. A lot of linux distributions prefer sharing iso images this way.
At this point I realise that hosting your own data requires a different hardware spec than the one mycroft already operates. Expert users can build it and install mycroft on top of it given the open source nature of the project. But average consumers would need to buy a different spec'd machine. Which means more sales.
An exception to all the above is that for users who don't mind having their data on a corporate server, or for users who opt in for a backup of their mycroft data, or for users who use a simplest version with no storage capacity you can have a subscription scheme (of various levels) to accommodate them. Since they are all optional (convenience here is the selling point and a lot of people spend their money to have it) more privacy cautious people will be able to opt out of it without losing anything from their customer experience, which is key.
Last but not least is the issue of leveraging the user data in order to create a better customer experience. Given the setup I mentioned the only viable way is distributed machine learning.
The open source nature of the project should be able to verify to people that their data is used in an anonymized way when distributed training/learning occurs.
Web of Trust or PGP cryptography solutions can also be used to verify the identity of nodes initiating requests for distributed machine learning to maintain the integrity of the system (although I am not really knowledgeable on this subject).
There are two big drawbacks to this suggestion:
I do not know how late in your business plan it is for such an idea to be adopted by mycroft if you are not already moving towards it. Potentially that could be a deal breaker.
Implementing such an idea is no easy task and the field is not mature enough as of this moment. It would likely require a feasibility study from your part (if there is not one already). I would be happy to contribute in that.
Disclaimer: I am not an expert, and I am not looking for a job either. I just really like this project and I would love to contribute to the extend that I am able. Feel free to message me for further consideration.
To sum it up I believe that premium services should mostly consist of cosmetic changes. People are willing to pay for them (as has been demonstrated in many online/pc/android game stores) and do not take a away from the main product. This revenue stream is unlikely to be enough and in order to compete with the likes of Amazon I believe the infrastructure cost should be shared with the users as much as possible. Peer to peer updates can be run and distributed machine learning approaches can be used to minimise the needs for services running on a central node as much as possible so operating costs are minimised. I would also like to believe that if the infrastructure is shared people will also feel part owners of the project and would be more willing to contribute to it and help make it successful.