Hacker News new | past | comments | ask | show | jobs | submit login
Ask HN: I've decided to shut down my side project! What now?
4 points by denibertovic 52 days ago | hide | past | favorite | 11 comments
Ledger [1] is a tool that reminds you via slack to terminate your EC2 instances (works with GCP as well).

I've built this last year. It didn't get any traction and I've recently decided to shut it down. I wrote a twitter thread [2] about it explaining why and listed some of the mistakes I made. And boy there were plenty (some of them I didn't even mention...like how the name is totally wrong XD).

I got a suggestion in the twitter thread that I should post it here and see if I can get any more feedback. Might as well get some more knowledge out of this failure, right?

It was also suggested by a couple of folks that I list the project on sites like microaquire and indiemaker but I don't know how realistic this is?! I mean why would anyone want to to buy a SaaS with zero users?!

[1]: https://ledger.initeq.net/

[2]: https://twitter.com/denibertovic/status/1466408868276314114

Do you have any current customer? Can you just unplug it without hurting anyone?

The price structure is strange. I think 50 VM in the free plan is too much, and $1200 in the next plan is too much. [This is extremely far from my knowledge area, so perhaps my opinion is very wrong.]

Also, from this post by patio11, https://training.kalzumeus.com/newsletters/archive/selling_s... the sell price is the annual profit multiplied by a small number, like x1, x2, x3, ... It depends on a lot of things, and people have sold companies without profit, but don't be too optimistic.

I don't have any users. A few tried it out but did no continue using it.

Honestly I can keep it running forever since it's running on a dedicated machine that has lots of resources and that I use for other projects as well. So it's not like it's costing me to keep it up. The burden is purely mentally (I just don't want to keep thinking about it anymore...I tried...I failed...right?)

I experimented with pricing a lot. The one you see now is the latest iteration. The idea was you pay 1$ per EC2 instance per month. So kind of like insurance right? The trick is that if you have say 60 instances you're paying 100$ still because the tier allows 100 instances to be monitored. I wanted the expense to be fixed every month because I've gotten feedback from folks that companies don't like variable cost per month for infrastructure...and it's already an issue that AWS is costing them a different amount each month....so it would be a hard sell to get to them to pay me a variable amount as well (small fish and all that).

Initially I didn't have a free tier at all. I only added it later as a desperate attempt to attract users. That's why historically I never gotten around to setting up a register flow without an invite link. :/ I don't know if 50 is the right number for the free tier but it definitely needs to be a larger number since companies/individuals that I target have a larger amount of servers. The ones that only have a handful don't really need this tool at all since they already mostly know all of their servers by name (so to speak). It's the whole "cattle vs pets" discussion. Anyway, that's where the limit to 50 comes from.

Thanks for the link btw. I read a bunch of patio11's posts but I don't think I've read this one.

In my opinion, if as another comment hinted there is a real need then either: a) you are unable to reach your target user or buyer b) you are unable to close a sale c) you onboarding experience or UX needs work (the little things needed to manage this solution day-to-day.

As the other commenter suggested, I would recommend finding buyers on Twitter and seek feedback before selling. It's easy to lie to yourself in customer interview so suggest reading The Mom Test to understand how to get meaningful feedback that is unbiased from your potential audience. You should also Tweet on your subject with a focus on the problem you are solving and why it is a problem, rather than your solution.

Thanks. I'm pretty sure it's a little bit of all 3 of those. And possibly other things you didn't list.

I have seen SAAS's on microaquire listed with no customers. In fact I've seen them listed with no customers and no product! Who knows if those sell though.

Worth at least investigating.

Another option would be to pivot what you've built. So you have the stuff for slack and aws wired up. Forgotten ec2 instance sounds like something people would care about but based on your results maybe not as much as thought?

I wonder if there is something else in AWS you could alert via slack that people would care more about.

Expiring domains or ssl certs maybe? Though I'm not sure people care much about that either.

In any case, neat idea. I bet this experience will pay off big later in weird ways.

My suggestion is to change the pricing to pay as you go. There is no commitment from customers, whatever they save you get some percentage of it. Say you turned off 50 unused machines than you take percentage of 50x1hr of pricing(for demonstration only you choose what you want to charge).

1. This will show clear value of your service as you are charging tiny percentage of what you saved.

2. Customers would be able to try without any upfront costs/commitment so it would be easier to sell.

Apparently the links are clickable only in comments:

[1]: https://ledger.initeq.net/

[2]: https://twitter.com/denibertovic/status/1466408868276314114

I think you should persist with it. Corporates still struggle with this problem. Twitter is you friend for this. Connect with VPs in charge of cloud infra of companies with relatively high number of employees and that lean towards tech.

Thanks. I can't help myself and still feel like there's something there... but I don't know how to work it. I mean if there was nothing there I wouldn't be hearing about new projects doing something similar. The difference is that they seem super complex and do a lot more. The idea behind my project was to be simple and perhaps be able to serve a subset of the market. But perhaps that was a mistake.

Is it possible to refactor it to use AWS Lambda so it's only using the compute power it needs, rather than maintaining infra? Aka, "serverless"

I suppose. Given that it checks the AWS/GCP api on a schedule not sure how effective that would be.

Then again, it's not like the hosting is costing that much. Actually I currently run it on a machine with lost of resources that is used for other things as well...so technically I could keep running it forever. That is to say the cost isn't the main driver for shutting it down. It's more a psychological burden at this point where I'm thinking "ok I tried, it didn't work time to move on".

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact