The crux of the matter is that I find some SaaS completely invaluable, like watching session replays (= finding bugs/why people buy), accounting, and recently, using AI to prospect/lead gen (and doing actual content myself, at conferences, via phone). Many times hosting is free on startup plans on cloud, but if you don't create a new company entity, you often cannot access cloud credits for the smaller projects (sensibly), or you are paying OpenAI, so you are looking at $100-$500 for the tech too. I also feel they are kind of unoptional now: Sales response rates are 3-4x lower than what they were in 2011, because of AI prospecting/the gradual train of spam and investors focusing on tech. So you need to counter this with tooling.
The thing is each of these services hit minimum plans these days of say $50-$100 for decent usage. And with 3-4 projects, this means I am paying the majority of my margins to them. I think back to 2011, when I had a short term startup, and with the market being less crowded (and luck), was able to generate revenue without any of this tech. Our only running cost was servers and accounting (which was cheaper). The problem then was always distribution, but now I know how to do that, the problem is actually the fixed costs. Nowadays I feel like if you don't have a good enough UI/quality (i.e. use session replay/pay users for feedback), you just can't get high enough revenue and retention. So I buy service after service for each project (as since 2022, more and more limit to one domain per plan etc or have worse free tiers than they used to), but it feels like I'm no further along than I was in 2011. Yes I have tried to self-host supabase and highlight.io, both were too hard to do for each site. Other software was self-hostable, but not the ones I'm paying for! (without all sorts of config for email auth, etc)
Come to think of it, will we ever see a service like "Once" cover these kind of things? Because SaaS boilerplates sort of solved that issue for me. It's now the "improvement and listening to users" part that I am forking out money for.
With things like session replay or distributed tracing, you might find that the best thing is to simply do without.
If your issue is paying for big-3 cloud companies, you could migrate to running on a cheaper provider like digital ocean, linode or hetzner, and not use expensive value add products. You could use a single DB or compute cluster for multiple projects and so on.
In terms of organization, you don’t need to necessarily run multiple businesses. You could run one business with multiple product lines. It might be easier to keep on top of things by approaching it that way.
Assuming what you sell is a SaaS itself, you need to keep in mind that a good SaaS can easily have COGS of only 10-20% of revenue. They can then funnel a big chunk of their profit into marketing, grow, and if they're doing what you're doing, crowd you out of the market.
So you need to get as close as possible to that magic number of being able to provide the service at a cost of only 10-20% of revenue. You can do this by lowering costs or by raising prices (if your market won't tolerate higher prices, try and retool the service for a market than can). Get your COGS down and you will be able to reinvest the surplus cash you're generating into trying different marketing channels, with growth you will grow out of the fixed expenses problems that tend to come along with a micro/hobby business.
If you are saying that the software you buy to run your projects is higher than what your make from these projects it means that they are a hobby and not a side hustle or business.
Shut down the projects that you lose money on, keep the ones that are profitable.
Go through the list of vendors you pay and see if they are really needed (for example, do you really need session replay if you have 10 customers?)
Can you use a shared database for all of them, can you put them all on a dedicated server instead of pricy cloud resources etc.
If you are seperating them for accounting reasons, maybe chat to an actual accountant just need to do it like once a year about that and they may have a different solution for you. There may even be an accountant in your niche that could recommend a different product stack for account that might suit your needs better.
I could think of a few ways to handle that creatively but would want to check with a professional, like spin up a seperate company whose only purpose is to sell those SaaS products to you at exactly cost...
Same with the DB, maybe look at an older idea, like sticking a Postgres DB on a cloud, or even using a managed postgres DB. There's generally a reason people converge on a single cloud and that is egress fees.
A single managed DB instance in GCP or Amazon with all your services in the same cloud means you are not paying egress fees and you are not paying whatever markup Supabse us adding.
The accounting thing - yes maybe I should look at that.
You're absolutely right about DB and other aspects like that. We should do that for postgres.