That's not the whole list. We spend around $500/month if we include everything.

We used to host our own stuff to spend less, like having our own Mailtrain instance instead of paying for an email marketing solution.

But if you do too much of that, you'll find yourself spending a lot of your time doing sys admin stuff, instead of working on your product. This can kill you, especially at an early stage.

As we scale our[1] revenues (currently at $5k MRR), that $500/m cost will become a lower and lower percentage and be almost negligible.

[1]: https://mailbrew.com

Maybe it's worth waiting until you've scaled your revenue before committing to those outgoings? As someone else mentionrd, 10% of that is for scheduling meetings, with a team of 3!

If you have a lot of meetings with external parties (it's not about scheduling between the 3 internal) a good calendar scheduling tool is worth it in my opinion. When I was a sales engineer at a software startup we probably spent 10-15 minutes per lead sorting calendar times out. Considering only 1 in 10 leads turn into sales this was 2.5-3 hours per converted lead.

By having a self service scheduling tool we already filter for the non serious people "I'm not wasting my time finding time in your calendar" and improve our lead quality while reducing the back and forth nonsense that no one enjoys.

Might be offtopic but adding friction might seem like a good time balancing matrix but what you are doing is filtering not the worst leads but probably the easiest ones to convert. Those willing to go back and forth on time will build muscle memory and make a deal easier. The others are still cold.. they will take additional effort during the sales calls.

You may also need to hire more people as you scale. With API-based billing, this is fine as you can still allocate a certain expected revenue per user as a cost. With head-based billing, what used to be a fixed cost now grows linearly, yearly, with headcount.

Considering that the effective salary of 1 person/mo would be around ~15k, I think spending $200 to make people more efficient is absolutely worth it

Sir, where do you pay $15k for a startup developer salary?!

I've always heard that, for the US, when thinking about salary you want to about double what the employee makes when considering the employers cost.

This is overkill, add 35% to the top and you will be closer (depending on absence/sick levels).

Statutory allowances in the UK are higher than in the US, and in the UK for instance you would add:

- 12.5% national insurance + apprentice levy

- 3-4% pension

- 10% of days as holidays

- 5% of days as sick / other leave (dependent on amount taken in reality)

- 2% payroll fees / other

Note, this excludes training and equipment/office space.

I guess it depends when there you consider other costs like office space, equipment, utility bills, HR overheads, consumables etc.

Why add holidays and sick leave to an individual employee's costs? Most white collar employees would have an annual salary this is factored into already, unless you're having to hire temporary replacements for them perhaps (which would be more than a 15% cost, I suspect).

You are right, I usually do this to get to a worked daily or worked hourly cost (I work in consultancy) so I take holidays & sick into account to understand cost per worked hour / cost per worked day.

In reality it depends on the type of work you are doing.

Ah! Definitely makes sense in consulting/contracting :-)

That metric isn't just about actual payroll expenses, but also things like office space, janitorial services, etc.

Health insurance is another big one in the states.

This might apply to large organizations with lots of overheads and perks. But it definitely does not apply to startups.

In NL it's about 35%. Doubt other countries vary that much, probably a lot in the 25-40% range.

Depending on jurisdiction, there are taxes and such on top of salary.

Actually there is a more important comparison than salary: what benefit accrues to the company for 1 employee month.

When a company is undergoing fast growth, the return from one month of work can be startlingly huge. Every tool that enables better use of time is a multiplier on the income generated by that employee.

Also note, “sir” seems snarky to me at best and insulting at worst. Are you even answering a guy?

15k could be cost to the company, which also looks at employment taxes, office space, health insurance, benefits, hr/support overhead, interviewing, training and managing. A new developer earning 8k/mo easily costs the company 15k/mo.

He said per month.

The person at ~15k/month (which is a really good salary) wouldn't only be doing that. I maintain a lot of tools that would cost 1000x if we paid subscriptions per user @ SaaS and it probably takes me 1-2 days per month. We're at ~150 devs.

I do a lot of other work in the other days of the month including coding, automation, leading, etc. etc.

The company is Italian, 15k/mo is VEEERY far from the local reality (more likely 40k/year)

15 k / month? Lol. Perhaps in a very tech dense specific area in the US and then you have the rest of the world.

The problem comes when services start effectively 10x'ing or 100x'ing prices, like what postman did at the beginning of this year (bumping up the price for each tier and dramatically reducing what you get)

That $200/mo today could easily be $200K/mo a few years from now, and at that point it will be harder to switch off.

If ~$200/month materially effects your runway, what you're saying is true, but even super duper early stage bootstrapped startups have that kind of cash, in my (admittedly limited) experience.

And it's not like you don't need what these services are providing, you'd just then have to do it yourself, and honestly I'm finding time to be more valuable than money right this very moment in the startup I'm working at...

~$200/month for a 3 person startup might seem high. now calculate the cost of a server to host that FOSS stuff and the time and energy to maintain it, backups, etc.

TBH i would rather put that time and energy in my startup instead of maintaining a FOSS stack for my company.

Do you think FOSS stacks are made of ice? That they would melt if you don't constantly resupply fresh ice from arctic excursions.

Surely you maintain your start-up tech on F/OSS, as most of the internet does, so just use the person who types 'apt-get update && apt-get upgrade' on your service to do the same on your local VM.

Or leave keep it local and literally never update it.

It wont melt.

> Do you think FOSS stacks are made of ice? That they would melt if you don't constantly resupply fresh ice from arctic excursions.

Yes, if you care about things like not getting oWn3d and having your customer data exposed.

> Or leave keep it local and literally never update it.

The mouth salivates at this prospect.

Yes, please set up your Jenkins server once and never look at it again...

> BUT a team of 3 at early stage project/startup level should be a bit more frugal and look for clever ways to save the ~$200/month

I think this is backwards. Instead the relevant question is how much can a team of three (x) achieve by aggressively outsourcing everything that isn't their expertise/value add

Even if you presume the FOSS versions were functionally equivalent, there is large cost to rolling your own (admin, backups, debugging). It’s precisely the 3 person teams where SaaS makes most sense. Larger orgs can amortize the relatively fixed cost over a larger user base.

How much time will you spend writing custom solutions? Why would you focus on anything other than your core businesses? Why would you want to maintain servers and OSS that has nothing to do with your core businesses? I don't think doing everything yourself just to save $200/month always pays off.

I know your point of view is the accepted mainstream view, but I disagree and I never did it in 15 years.

Maintaining a gitlab instance, a mattermost instance, a kubernetes instance, an elk instance on a cheap VPS provider gets you in a pretty good place for very little money. Your internal tools won't need much attention until you hit a higher number of employees.

Also with three employees you can still get away with a lot of free stuff (eg. a discord or matrix server for chatting, private GitHub repos)

Because your tech solution is not NOT YOUR business - its one aspect of it, and customized solution may instead of cloud one give you higher efficiency as it is tailored specifically to your needs, plus durability as you totally control it. Cost is totally not important in that regard.

Cost of cloud solution is not comparable to cost of employee - employees build what you want, while on cloud you adapt to the solutions offered and also have to do additional cloud-related duties not relevant to your problem scope (so you payed also for the stuff you don't need).

$200/mo across the time is like the cost of one developer laptop accidentally left under the dripping faucet of a hotel sink in Las Vegas. I know this stuff can add up and people should be careful about racking up recurring charges for stuff they're not really using or getting much value out of, but optimizing away $200/mo doesn't seem like a huge ROI.

If you're paying people on your team, the most expensive thing you probably buy is a person/hour of time.

Honestly, I am less concerned with the money and more concerned with the level of process overhead. This article makes it sound like they are spending a lot of time in email, notes, meeting minutes, etc. Which seems like an awful lot for 3 people who can just talk to each other. I feel like todo.txt is enough organization for three people.

> I've written on HN before that I'm TIRED of subscription apps, specially per user subs, for no reason!

This is my complaint too, especially when a "user" is defined in a ridiculous way. GitLab is a good example. Unless you're on the top tier plan, guests are considered users, so if you want to give someone access to a pages site that has authentication, you have to pay as if they're a developer. It's frustrating.

Our "solution" is to have 2 GitLab instances, one CE for projects used by guests, and the other one (paid) for regular users. Infuriating.

Don't forget per month/per user but can only pay annually! lol.

  > You'd be surprised how many great FOSS (free & open source 
  > software) are out there that you can deploy on your own VM 
  > and have unlimited number of team members and OWN YOUR OWN 
  > DATA!!!

I'd love to run some open-source stuff for analytics or error tracking or application monitoring. And we have, in the past. And they work pretty well. When they work.

If the VM disk starts filling up because log rotation wasn't set up right, or the ElasticSearch instance we only run for this service starts flapping, now I've gotta pay one of my devs to work on that instead of work on our product. And in your example, where I've only got three folks, that interruption could be ruinous.

$200/mo ($2,400/yr) is nothing compared to the financial and opportunity costs here.

One could look, at it as a form of dogfooding, in the broader sense of the indie SaaS ecosystem they're helping build.

I think this is probably a good reason. It probably really helps to see the billing, user management, support handling, emails, etc. from these other Saas products.

Vercel billing per-user is so head-scratching. I think it's a greedy billing strategy.

There's appears to be a mismatch between their costs and their billing, which either is true and someone will take advantage eventually which will force a restructuring of billing, or it's not actually a mismatch and we're all judging a situation with not enough information.

It is, but it’s funny no one complains about per-user billing for the other products. All of these other products bill per user instead of by bandwidth and storage.

I would switch to Cloudflare Pages for a serverless setup.

Can you mention specific self hosted apps that you think are great?

Depends on what apps you are already using and if there's a self-hosted alternative to it.

For analytics for example you can save a lot of money without sacrificing insights or privacy by using self-hosted userTrack[0] instead of GA + Hotjar/FullStory for example.

[0]: https://www.usertrack.net/

Overruse of subscriptions is like creating unnecessary dependencies. Eventually this transitions to a state that is incompatible with qualities such as small, fast, portable.

Exactly, eventually some of the Saas's will disappear.

"Adopt, adapt, invent"

This is questionable advice.

1) Any SaaS that actually provides values is probably worth 10-100x what it costs.

Of course, it doesn't mean they all provide value.

There's a mistake in your comment ('we stopped using them and didn't miss them') which is the assumption that SaaS doesn't provide value, which is obviously not the case.

Good tech is an amplifier - that's why it exists.

You're telling the farmer don't spend money on fancy tractors, toil the land with a sickle.

For every $ you spend, the idea is that you are generating ROI, ergo, you want to spend more.

If this 4-person team is hauling in $400K/year it's precisely because of those tools, without which they literally might not be able to do it.

So the advice there should be:

"Don't use the things that don't help, definitely use the things that do"

2) The mindset of 'FOSS is Free' is completely wrong.

Every tech has a 'total cost of ownership' and if you have to spend any time at all installing, fussing with, maintaining it - it's probably worth paying for a SaaS.

$5/month is literally 5 minutes of an employees time.

Think about that: if the SaaS saves you 5 minutes - it's worth it - OR - if the FOSS takes more than 5 minutes a month in over head, it's not worth it.

One of the many reasons that Windows dominates Linux (it's more complicated obviously) - is that Windows is somewhat easier to use and support out of the box. So if Windows saves a single support issue per year, per user - then it's worth it.

3) Data Ownership - like everything, the pragmatic reality of the situation matters more than ideology.

So many factors: security, integrity of host, does the data have 'long term value' (do you need granular data from 5 years ago?).

Do you really think that your home-baked file server in the corner hosting your shared documents is more secure, supportable, maintainable, lower cost than the SaaS?

Maybe - but probably not.

Summary: invest in things that are materially useful to your operations, and if they do 'help' they are probably worth quite a lot but have the self awareness to not get addicted to 'distracting' SaaS, and make pragmatic decisions about what data will sit in the cloud and why, and finally you can still push vendors to provide more 'controlled' data scenarios, such as co-locating services that they run etc.

My issue is that value based pricing sets up incentives against one of the coolest parts of software and digital goods- near zero marginal costs. Software is kind of magical in that once produced it doesn’t cost much to add a second or billionth user. Of course SaaS has more costs, but not nearly as much as the “charge 5% of value” or whatever.

So I think it’s foolish to not buy tractors and till the land by hand, it’s also foolish to think of a hammer as a value add so that the cost should be 5% of the benefit gained by using it instead of some reasonable profit on top of cost. I’ve bought some plumbing tools decades ago that are really useful, they are well made and should last decades more. Saying that I should pay based on value might make sense but isn’t a good user strategy in the long term.

So that's a fair point ... but 1) 'value based' pricing is inherently part of capitalism, and it's more or less a 'good thing' ... and 2) the upside in general is still pretty huge for most services. The fact that software has 'zero unit cost' is for the most part, built into the price.

It's the reason you can try 100 services for free before buying. The challenge is, to find things that really work for you and that do provide value.

Your summary is totally right. But first part of the comment feels too one-sided (pro-saas). So couple of balancing thoughts for each point.

(1). Tractor analogy seems too extreme - it's about skiping automation in farming. But if one skips saas, one still automates in house (the amplifier is the automation). Also, frequently "saas" feels less like "tech" and more like "outsourcing".

(2) Total cost of ownership applies to saas as well. Lawyer's time for contract. Developer\architect time to figure out documentation, so company would know what configuration to ask for. Communication being external also costs time.

(3) Solutions for data ownership exist, but it's one additional thing to worry about.