SaaS founders, what is your biggest technical pain point?
Troels Roennow
18 replies
I'm currently trying to understand what the biggest technical pain points SaaS founders have? For me it has been understanding all of the tech stack that I needed for my first product and how to make it work together.
What is yours?
Replies
Sagnik Datta@sagnikdatta129
I am keeping a tab of this thread, thanks for bringing this up. While not being a founder, I can empathise to the problem as we are in a similar situation ourselves. We are rebuilding our offering from on-premises to SaaS. We had to balance the initial objective (acquisition) we want to drive with a SaaS model and the best technical solution for scalability in the long run. The key result for the former objective to drive acquisition was to prepare the solution on the cloud. So we are taking a fork and lift approach. That's obviously not best on the long run and hence finding the most effective result for our next objective, which is to pick the technical solution stack. However, I have a high gut that we will just go by whichever will lead us to maximum activation from our users (going by standard AARRR metrics of SaaS).
Share
@sagnikdatta129 Don't worry - it is not meant to be founders only, I just assume that all who are developing products are founders, but that is obviously not true.
What stack did you end up going with for your project?
Pirsch Analytics
Hmm, there are so many problems to solve. One of them we fiddled around quite a bit is infrastructure. Container orchestration, Docker, root server or VM? There are so many choices and tools like Kubernetes are really intriguing, but hard to manage with a small team and relatively expensive in the Cloud (for a small startup without funding).
After re-evaluating we moved from Kubernetes to a single VM with Docker and Compose, which can handle a LOT more load than most will assume on a first guess and right now I'm thinking about trying Nomad, as orchestrators solve thinks like rolling releases and failover better.
@kugel Rolling releases and the corresponding infrastructure does require some fiddling and it is not easy to find good documentation. I am surprised that you found Kubernetes expensive? I am running my cluster at $20 / month which is quite acceptable imo. and have set up Gitops to do continuous roll out for https://indiemy.com. I might be able to shave off another $5-10 by going for an ordinary VM, but the benefit of Kubernetes definitely outweighs the price.
How did you find the learning curve of different container orchestration tools out there?
Pirsch Analytics
@troelsfr I haven't tried anything else yet. Where do you host your cluster? Did you set it up yourself? One of the major pain points I had was managing the Kubernetes cluster after we switched from Google Cloud to our own VMs on Hetzner. The GCE cluster did cost about $200/month depending on traffic, but the majority was the price for VMs, storage, ... It can get quite expensive rather quickly.
The self managed Hetzner Kubernetes cluster costs about $15 if you set up three nodes on the cheapest VMs (comparable to what we used on GCE), but the overhead is not worth it.
@kugel I am. currently running things with DigitalOcean. They are slightly more expensive than AWS, but not by much for dynamic allocation. $200/month sounds rather pricey so I understand why you feel the need to switch. Do you have some special requirements (like 200 GB of memory or something) or are these just ordinary VMs you are running in the cluster?
Slightly related to pricing: I looked at Amazon pricing yesterday as I am planning on starting 2 - 3 more projects over the next half year. I noticed that they give a quite significant discount if you reserve resources for 1 year.
Pirsch Analytics
@kugel that's understandable. At the end of the day, you are starting a business and that needs to take priority. If you don't know this already, there is a Docker image called Watchtower[1] which can be used for continuous deployment with docker-compose. If there is anything that I can help you with feel free to reach out.
I am currently building a course where Kubernetes is the deployment piece. Given that you building on Go and Vue, I don't think this is the right course for you (and it is neither finished yet) since the primary focus is on Django and ReactJS, but it may be that the Kubernetes part of the course is relevant to you.
1. https://hub.docker.com/r/v2tec/w...
For us it is also about choosing the right tech stack. There are simply too many ways to go.
@michael_sieb I fully agree. And there so many ways you could combine it.
@michael_sieb What stack are you thinking of going with?
For us it's at the moment to define the prorations (= SaaS resources assigned to a pricing model) and the pricing models. We have to understand the specific user needs, personas and user journeys in order to define suitable and acceptable prorations. It's very much at the center of the whole project, but it's an abstract and invisible thing - not only technical but also a technical thing in the backend.
@timz_flowers That is a quite interesting way of look at this problem. What is your current model for defining the pricing (if you solved it yet, that is)?
@timz_flowers Your view is first and foremost interesting to me because I never really considered this a problem. This usually means that there is something that I do not understand and so, if you know something that I don't I am quite keen to understand what it is :)
For me, the challenge is quite easy to solve: What I am building requires X amount of computational resource which will cost me Y and if I need the profit margin to be above 90%, then that roughly sets the pricing. However, I might have oversimplified things. Looking at your product, I can see why it might be harder to come up with a good pricing model as I suspect that it is more important how much resource you free up for your client than what it will cost to run you asynchronous video platform.
@troelsfr When you say it that way it sounds indeed easy. As a founder I've to make decisions, trade-offs between different evaluation systems and by doing so I create (most probably) irreversible path-dependencies. Do I want to have many users with low margin or few with high margin? The margins depend on scaling effects and these depend on the user-acceptance of product and price. All of this has to be set in reference with your budget, burn rate, growth expectations... Wow it's good to talk to you, this explains me some of my own situation.
@timz_flowers My preference is to have many users. And yes, this has been quite good exchanging notes on this. I probably should consider how to do price discovery a bit more instead of the simple model that I was planning to use.