How Nano can solve subscription fatigue

People’s bank accounts are draining without them noticing. A dozen small subscriptions add up to a steady outpour of money, often including some payments to services you don’t remember using.

Businesses love a subscription model since it provides them with a steady and predictable stream of money. It sometimes makes sense for the consumer as well, but often does not and leads to annoyance.

Yet businesses often see little alternative, since payment processing costs mean paying per article or per movie watched adds a lot of costs.

There are two primary ways in which Nano can improve the experience here.

1. Allowing customers to take back control.

When alternatives are offered that don’t force you to use a subscription and instead allow you to pay per use you have a choice. Perhaps the package deal works best for you, but perhaps you’re far better off paying on an actual usage basis.

Tweaterdotdk and I created Nano-GPT which allows anyone to use the latest state-of-the-art chat and image models.

These would normally often be locked behind subscriptions (such as with ChatGPT at $20 a month), or require someone to own a strong GPU to run the model themselves, or force users into giving up private information.

Using Nano-GPT there is no subscription. Users pay per query. Need help debugging a program you wrote? Feed it in, get feedback for a couple of cents. Have some questions about data formatting? Pick your model and pay a few cents.

Want to generate an image to go with an article you’re writing? Generate an image using a model of your choice for perhaps $0.05. Want to get some feedback on that same article? Pay a few cents and GPT-4, or Gemini Pro, or Llama-2 will tell you what you can improve.

We use Nano for Nano-GPT because it allows for the smallest possible transaction sizes. There are no fees to take into account, at all.

We could give someone $0.05, and they would be able to deposit it on our website and use it for prompts immediately. Someone could deposit $1, use that until it runs out, or just use whatever they need and then withdraw the rest for free.

There’s no subscription, no sneaky siphoning out of money. It’s very transparent in that you pay exactly for what you use, and if you stop using and withdraw the rest of your Nano we’ll say thank you for your usage and we hope to see you again some other time.

In the case of our website we allow people to deposit using other crypto as well, via swap services, but all of these come with fees, confirmation times of more than a second, and make it impossible for us to offer full withdrawal at any time at zero cost.

Nano-GPT is just one example. A similar system could be set up by Substack, newspapers like Financial Times, or any other article-based service.

When it comes to Substack, I’d often gladly pay $1 to read an author’s article, but am unwilling to sign up for a $5 a month subscription.

FT has a lot of awesome content, but so does Bloomberg, the Economist, and many others. I don’t want a subscription to all of those, I want to read what I want to read and let that be it.

2. Expanding the market for existing services.

While offering pay-per-use models can be seen as directly competing with subscriptions, they can in many cases be complements.

Right now I don’t have an Economist subscription. It’s not worth it to me for the few articles I actually want to read and I would likely forget to cancel my subscription.

The Economist’s current revenue from me is therefore $0.

If I was able to pay say $1 per article that I want to read, I’d likely read at least a few articles per month, so their income from me might be $5 a month. That’s not as much as a subscription, but it’s $5 of revenue they would otherwise not have had.

Using Stripe or most other traditional payment processors for $1 payments hardly makes sense since ~30% of the revenue would be eaten up by fees, but 1 Nano sent is 1 Nano received.

Would offering pay-per-use mean subscription revenue declines? It definitely could.

While I might not think it worth it to pay $20 a month but would pay $5 a month for 5 articles, there might be people that currently read 15 articles per month yet pay $20 at the moment.

Alternatively, I might notice that the articles are really good and that I want to read more than I thought, and end up going for a subscription after all. It all depends on to what extent people think the subscription model makes sense in their particular case.

Companies implementing this would need to strike a balance between the price of individual access and subscriptions, and it would likely take some trial and error to figure out.

I expect that in time the businesses that do so using Nano will end up having a competitive advantage over those that don’t.

As more services start implementing such models using Nano the network effect spreads and offering pay-per-use access using Nano becomes even more attractive, which benefits all of us.

For businesses interested in trialing such a model, take a look at Nano-GPT and perhaps try it out. If your business is large I’ll gladly help figure out how to make this work for your service, and provide help on the practical implementation.

Given how tired people seem to be getting of subscriptions and the outpour of money that comes with it it seems worthwhile to experiment with these sorts of models.