Fewer customers/income than your expenses
Say you built a SAAS with yearly pricing models, few customers subscribe to the yearly plan.
To cover your infrastructure/maintenance cost you still need more customers.
But the sign-ups are too less and your customer count is not growing the way you expected to and you decide to close and do away with the side project so that you would not incur additional charges on maintaining the product.
Considering few people have bought yearly plans, how would you shut down the service?
Would you wait for their subscription to get over and bear the loss or somehow refund the customers and close the service.
Edit:
As expenses I would also like to add paying to third party services, for example some external APIs, say for example goalserv charges around $150 per month.
Don't sell YEARLY plans when starting out! Period. It's a poor decision and doesn't do justice to those signing up. Unless you're 110% you will be able to honor those 1 year contracts, do offer them.
Yes, you can choose to refund them. However even when refunding those customers, you're essentially still screwing them over as they're likely depending on the service being available.
If you've attached your name to the product and might do something in the future, you might want to go out of your way to help those customers.
You could refund them the full year to show your appreciation, refund them partially but tell them you're giving them a couple months free to find alternative solutions, or just try to struggle through a year of service.
I agree with Andrew. Additionally, if you decide to shut down, besides refunding, you could make it easy for them to export their data (which, honestly, is now a GDPR requirement) or even suggest competing, similar products they could switch to. It's like when Google Reader shut down, and everyone mentioned "Hey, don't worry, there's Feedly".
If that were the case, I'd say you've probably made absolutely atrocious choices when it came to infrastructure. Very poor algorithm or business model choices could also be a possible cause.
My infrastructure costs me $5/month for hosting, $1/month for backups, and another dollar per month amortized if you count the domain name. None of those costs would increase at even 1,000x my current number of customers and even one paying customer is far more than enough to cover those costs.
To answer your question, though, I do have yearly plans and here's what I'd do if somehow I couldn't honor them: I'd refund whatever portion of the year they had paid for and I couldn't provide.
You're making a lot of assumptions here in your accusations...
Accusations? What are you talking about?
For a paid SaaS to be overrun by infrastructure costs, something has clearly gone seriously wrong. That's not an accusation, but it is hopefully an alarm bell to consider if you ever find yourself in that case.
I don't want to jump on a fellow alchemist, but we have no idea of the nature of their business. It could be resource-intensive.
Is what I was refering to. And then you go on to say what your costs are. $5 is what? 1GB memory and 1vCPU?
They might need much more than that for what they're doing, we don't know. I just found you jumped to harsh conclusions a little too quickly without knowing what they are actually doing.
Yes, and I've tested it under loads of over 1000 reqs/sec! It's not a particularly intensive app, but it's doing more under the covers than many content-based sites. My reaction was more general than how cheap VPS hosts have gotten, however.
I've also been an engineer at a unicorn, and a few startups in the US and China (where I have to say the resource costs / customer LTV concerns are intense vs in the English-speaking market). Even there, even nearly a decade ago, infrastructure costs were generally low compared to payroll.
I agree. This is why I also put out the possibility of a business model problem. Charging less for an expensive, resource-intensive service than its infrastructure costs would fall into that category. A less likely possibility is that OP's hypothetical involved dramatic cost efficiencies of scale that had been counted on but not realized.
My thought was that, an unnecessary alarm is probably a lesser harm for the OP and anyone else reading than an incorrect assumption that infrastructure costs should rival revenue for a paid SaaS. I definitely didn't mean it in an accusatory way though and sorry to the OP if you took it that way!