Pricing model for resource-intensive SaaS?
I am starting work on a B2B SaaS product in the data analytics space. Folks in this space seem to expect a free tier, or at minimum a free trial, to establish value before buying.
The trouble is, my product will use quite a bit of computing power--tens or hundreds of CPU minutes each time it is used. A typical user session could easily cost $5 in compute. As a bootstrapped indie hacker without VC funding to splash around I am nervous about non-paying users running up my hosting bill without offsetting revenue, especially since there is no guarantee they'll ever convert to paying customers.
How have other IHers built resource-intensive products? Do you run free/trial customers at a loss, or do you require a paid plan up front?
Requiring a paid plan up front likely increases my effort per sale and limits the incoming customer funnel, but it does limit the downside risk of runaway costs. Is this just the price of bootstrapping?
Could you offer a demo that uses less resources like don't use fresh data/have a hardcoded data set just for the demo?
Another question, based on expected usage and how much your computing costs are, can you be profitable? I haven't seen anything nearly that expensive compute wise for what I'm assuming is a short user session.
You might be right--a limited demo could be the best way to offer a trial. I like your idea of demoing the product with hardcoded results. There are also a few rare special cases for this kind of analysis that turn out to be quite cheap computationally, so maybe I could offer just those to trial users.
On profitability: yep, getting the unit economics right will be critical to success. Definitely something to solve before launch. On the other hand, this is a challenge everyone will face so finding a model that works could end up being its own sort of moat.
Some ideas...
Can you run it partially? Hence provide a free glimpse of what to expect to establish trust in what value it would provide? Establishing data connections upfront may provide easier onboarding later (since the potential customer is already "technically" invested)
Can you provide a demo with fictional data? (with the heavy computation only done once)
Instead of requiring a paid plan, you may experiment in providing a paid one-off test to a lower price?
Lastly, with a more strict lead qualification, it may be worth that customer acquisition cost (CAC) if it increases the likelihood of a new user with a lifetime value that soon pays back the CAC. (of course - cash flow is the tricky thing when bootstrapping...)
Yeah you're probably right, this is the way.
Absolutely, and maybe videos demonstrating how a user would interact with the analysis results.
Possibly! Certainly worth a try.
This would be a solid fallback if I can't figure out a way to build a compelling demo--or as you say if a customer is a particularly well-qualified lead.
I would highly limit the scope. Like having a free trial but with only XY CPU minutes, or processing files within certain size, etc.
Yep. The trick will be to find ways to show value without giving away the store.
We had similar issue with https://www.qapop.com/: generating report with SEO data cost few $.
When we launched the MVP, we did not have any free trial and we had a sales led approach. You had to pay upfront to get a report, but you were speaking with a human before paying. This help us validate the idea and justly investing more in product development.
We developed a free trial which doesn't require the SEO data, but already show some value for the tool. That helps us to get higher volume of users to gather feedback. And hopefully it will help for the product hunt launch.
We have on the backlog to create a demo report with SEO data for people to play around.
So my 3 advices would be:
Thanks @FabianMaume. This seems like solid advice. Maybe a one sentence summary could be "try to get momentum without giving away resources for free."
For our product I think we could manage all three of these options (sales led early on, cheaper trial version, hard-coded demo). Being a bit more sales-led at the start is probably not a terrible idea for other reasons, too. Talking through the solution in detail with customers up front probably helped keep a lid on early churn, too.
I think the key here will be finding a way to communicate value to prospective customers without needing the expensive version of the tool, much as you did with qapop (which is a good looking product by the way, props!)