How We built a $1M ARR open source SaaS | Hacker News Reader

How We built a $1M ARR open source SaaS

plausible.io

522 pointscaust1c106 comments

Pretty darn fun to be able to see the Hacker News front page effect in realtime since they have their own site's Plausible stats publicly available!

https://plausible.io/plausible.io?period=30d&source=Hacker+N...

What a great demo of the product :D

I'm surprised the baseline's so low? (Single digits)

That's incoming traffic from HN only. Remove that filter and it's closer to 5k visitors/day.

Ah! That makes a lot more sense, thanks, I didn't spot that.

(In case Plausible people are here: I found it a lot more obvious on the desktop site, in fact when I saw it there I thought 'oh it just doesn't show the filter on mobile', but then when I switched back saw it does, I just hadn't noticed it.)

good point, thanks! we had to shorten the top menu on mobile devices to fit all the important stuff so you only see "X filters" rather than the full filter as on the desktop

What a great link.

At the time of writing, Goal Conversions:

- Visit/blog 12800

- Signup 8

I knew anecdotally that HN doesn't convert but, oof, that's rough..

HN influences. It makes a wide and deeply technical circle aware. Maybe a week, month, year they suggest [or recommend to peers], etc. organic growth business like Plausible is about surviving long enough to keep growing your network effect. All these mentions and discussions add up. People remember.

exactly! our first "viral" moment was on HN too and we only got a few direct signups that day but we would never have made it to this point without being on the top of HN that time

All their marketing seems to be pitching the product against Google Analytics. But they don't spend any time differentiating themselves from other privacy-first analytics companies, of which there are many.

I mean, when they started in 2018:

> Uku’s first thought was, “Ugh. Can we just use something other than Google Analytics?”. This is how the idea for Plausible was born.

Surely they must have looked around and seen that there were already quite a few products doing privacy-first analytics? But yet they started anyway? And had success even though their marketing doesn't really state any reasons why the user should use them not the many other privacy-first analytics offerings out there?

I mean I'm happy for them that they're successful, but I think there must be more to it than alluded to in this blog post, both in (a) the decision to start in the face of lots of competition and (b) success despite not trying to differentiate themselves from the competition.

99.999% of their potential customer base is still using Google Analytics. I'm not sure the comparison to anything else matters at this point.

When you buy a Tesla you won't find comparisons about potential savings or benefits over other electric cars. Instead it is compared to the cost of ownership of a gas powered car.

I'm not sure. If you are convinced you need privacy-first analytics, and thus need to get off Google Analytics (e.g. because you've heard about Plausible) wouldn't you at least do a quick google to find other products (which might be better/cheaper - not saying they are, but you'd at least look?) And then you'd need reasons to choose one over the other.

And, when they started up, they must have googled and found other products and decide to start up anyway? I'd love to know more about that decision. Seems like it'd be difficult to justify, but obviously it worked out for them so maybe I'm just overthinking it.

Having lots of competition doesn't mean they have equal market share. It doesn't matter than there are lots of competitors if all of them are fighting for less than 1% of the market share. In fact, you don't even need to worry about them since none of them have figured out how to wrestle market share away from Google.

As for competition, you want to enter a market with competition. If there is no competition, then there must be a reason. Either a) there is no market at all, b) the market is not profitable or c) you are inventing a new market. I have seen many founders try to do c) because of your reasoning of finding something with no competition and fail.

Competition also shows you might have timing correct. If they had no competition except Google, then why would Plausible be the only ones to see the market opportunity? That's extremely unlikely and would indicate to me that Google still had a stronghold on the market.

I second this, it's better to enter a market with some competition and turnover, if you are trying to invent a new market, even if yhe need is real, you need to explain to people why they need this new wiered thing they never considered. unless you have tremendously deep pockets, you probably can't afford customer education at scale.

also you dont have any data to enable you to separate potential customers fron people who just want to waste your time because they are curious but wont buy. you dony know the buy cycle in that markrt, etc.

> And then you'd need reasons to choose one over the other.

Seeing which service has customers you have heard of as a proxy for credibility.

thanks for sharing! there's this saying in marketing that you should ignore anyone who's not much bigger than you so that's how we think about it. we focus on the tools with a large audience and ignore the rest as few people in our target audience would have heard about them. people tell us all the time how they didn't even know alternatives to GA existed before they found us and that tells a lot about the state of the market for millions of sites that use GA

I also started a privacy-friendly analytics SaaS without looking at the market. Sometimes it's better to just get started. Otherwise you probably won't start doing anything, as there are existing products most of the time. In my case, I was looking for a Go (golang) solution that I could embed in my website, as a library so to speak, and just turned it into a product later as I was looking for a new project to work on.

We're now at $1500 MRR and growing. I'm also opposing the position of "just being against GA" now and we try to differentiate more. It's almost impossible to get anyone away from GA who does do performance marketing. So I don't quite see how Plausible or other privacy-friendly products are a replacement. But most websites that use GA just don't have to, because they don't rely so much on ads or personal data to get value.

Our product: https://pirsch.io

They address this in the linked blog post: https://plausible.io/blog/the-analytics-tool-i-want

> When I looked around I was fully ready to pick up another tool and install it on Gigride immediately. > > Turns out there are some alternatives for Google Analytics but I didn’t find any of them compelling. Simple Analytics and Fathom are the closest to my ideal but they are a bit too barebones to be useful for my use-case. For example, it’s quite important to me what browsers versions my users are on – if I use a css rule that isn’t supported in IE7, how many users does it affect? Neither Fathom or Simple Analytics provide that answer currently. > > This seems like an opportunity to build the tool that I really want.

We capture browser version in Squeaky analytics at the moment, but we don't surface it cause nobody ever asked for it haha. If you want to try/use it I'm happy to visualise it for you!

I might be abusing the term here, but Plausible seems like the best example of a mittelstand[1] that sells SaaS.

They are not beholden to VC investors and can grow slowly, organically. According to this article, they were able to cover their costs by January 2021. By keeping their team small, they can remain sustainable for a long time and refine their product.

The biggest challenge for growth right now seems to be that customer support costs will scale linearly with growth.

[1] https://news.ycombinator.com/item?id=31327219

I believe the existing term of art is "bootstrapped business"

That's the current term of art. It used to be "lifestyle business," which has a more passive-aggressive feel to it.

Or ‘mom and pop’, which is even more derogatory (at least here in Silicon Valley).

Scaling customer support is the biggest fear, but I think it isn’t as bad as people think it is.

1. You can get a long way with extensive FAQ and give people ways to help themselves.

2. Existing customers don’t need the same kind of handholding as new customers, because they know the software pretty well.

3. It is up to you how much actual support you provide.

4. Customer support heavily depends on the kind of customers you have/want. If you provide something for free, you get a lot more stupid support requests than if you charge a modest fee.

this is right. I ran a fairly technical b2b SaaS product with hundreds of paid users and we had ~10 hours of support a month, most of which was driven by new features. If you keep the product simple and build your docs into the product, you can get by with a few hours of support a week.

You can also charge for different levels of support.

Customer support can definitely be handled or “minimized” by what I’d call “extreme thoughtfulness”. Design your application with the correct abstractions, be ready to redesign everything from the ground up every two years as requirements change, and constantly try to create tooling that will help customers and CSMs serve as much as possible themselves.

I usually apply to customer support/success role. My strategy to build a solid customer support system -

- Always incorporate gifs with documentation and guidelines

- Create video and step by step tutorials unconditionally

- Convert every customer interaction to an FAQ segment or a microblog

- Avoid writing books as documentation blogs. Adopt w3school like writing habits.

- Post your FAQs to stackoverflow. Two dummy accounts. One to post the questions, one to answer the question

- Add every meta information and phrasing variations possible to commonly asked questions

- Support/success staff should also be a part of the dev team so they can understand the product philoshophy and not just be a liaison of the dev.

TLDR: Make customer interaction as google-able as you can.

I suspect that support costs are actually sub-linear, as you scale you don't just add support staff, you start putting in extra stages to redirect support - better documentation, UX designers, community support. Support requests are a mix of unique challenges and common challenges, the common challenges can always be pushed towards a fixed support cost.

Since reading the thread on money-making side hustles on HN a few days back[1], I was able to finally brush off my procrastination and start twiddling with some of my side projects that have been lying dormant over the years! And the model Plausible uses (I think the right term is "Open Core"?) which is also used by the incredible Product analytics tool Posthog [1] aligns really well with how I want to build out my side-hustles. Open source, but also available packaged and hosted if you so prefer.

One thing I was concerned about was about revenue that these "open source by default" products make, so glad that this post has come along just in time to alleviate some of those concerns (of course, as with most engineering side projects, the priority is to build something cool first, and if it ends up making some money, then all the better!) and it's always nice to see successful OSS-driven products in a crowded analytics space.

Congrats folks!

[1] https://news.ycombinator.com/item?id=31764696

[2] https://posthog.com/

I don't have a need for your product, but that site looks fingerlicking nice!

(I'm a PostHog cofounder - we get about 35% of revenue from (2) and 65% from (3) below)

There are three established ways to make money from open source if you're thinking about this.. my default recommendation would be the exact model Plausible uses today - (3) below, if you are just getting started and either bootstrapped or VC backed in today's funding environment.

1. paid support

Advantages: (i) no need to build any extra features or set up hosting for your customers up front - low risk of wasting time in case you don't get product market fit with your paid product (ii) probably a good way to get close to users, as long as you don't end up over-fitting to 1 or 2 edge case use cases (who are, likely, those most likely to need paid support) - however, this will happen naturally if you set up a slack/discord community and you help people out for free

Disadvantages: (i) you are no longer a builder, so you're not doing 100% leveraged work, so the margins are slimmer (ii) will take some time to get to revenue because you will need a reasonably popular project to get much demand (iii) this will be hard to fund if you want to do the VC route, there are exceptions though

2. extra, paid self-hosted features

description: you have an open source project, and users can buy premium features on top for a fee of some kind

Advantages: (i if you target enterprises who don't want to send data outside their environment who don't want to buy closed source SAAS products - they can try + buy much your product faster than your competition's

Disadvantages: (i) it's hard to avoid customers expecting support managing the instance (just because they're paying you, even though you "just" sold extra features) (ii) you have to build a lot since you need an open source product and a paid product on top - so it needs a lot of investment to do this without hurting the open source project's value (I'd strongly recommend you get VC if you do this option, as you'll probably find bigger order values but slower sales cycles as well as all the building work) (iii) many people think everyone is moving to cloud, but we've seen there is a ton of underserved market here (more than enough to get to $100M+ revenue in our market at least), and increasing legislation around user privacy is keeping pressure on here

3. paid hosting of the project

description: you have an open source project, and users can pay you to host it for them

If you are bootstrapping or even VC backed in today's funding environment, I'd recommend defaulting to open source project _and_ a cloud hosted version as your paid version, unless there's a good reason not to.

Advantages: (i) this is the fastest way to monetize (ii) for users without self hosting needs, this is likely a better user experience - less work to manage, so you may just grow a lot quicker initially

Disadvantages: (i) you need to manage the hosted version (ii) at a product level, you are potentially competing head on with closed source SAAS

It looks like PostHog uses a permissive license (MIT). For business model (3), how fo you defend against other companies taking your project and selling a hosted version themself? This was a problem for Elastic and AWS, iirc

There is a range of paid extra functionality that isn't MIT licensed (the extras that make up part 2's paid offer above), so if AWS hosts it "competitively" it'll be missing many of the extra features larger organizations would want to buy. If we didn't have these extras, we'd probably need a non permissive license or a belief that AWS competing like this would increase awareness so much it could be a good thing.

I've always wondered how you manage this (2) tech-wise.

Here are my contradictory hypotheses: - I don't expect to find the code for such features in the open-source repo - I guess you dogfood your own product and therefore host exactly what you publish in your open-source repo

How do you "plug" these closed-source features to the product? I've always been curious of this as I strongly believe in this way of doing business for a SaaS.

PS: amazing company ~you've built~ you're building, GG ;)

Posthog looks really nice, but given the focus on Kubernetes for deployment, and the number of services deployed (Postgres, Clickhouse, Zookeeper, Redis, Kafka, a worker, and the web app itself), I presume it's aimed at large sites?

For small sites there's a hobby deployment (https://posthog.com/docs/self-host/deploy/hobby) that costs $10/month to host (we can't get it below that at the moment), or we give a lot of free usage away on cloud for smaller sites. I'd stress we're product focussed (product analytics + session recording + feature flags + experiments), and plausible is web analytics (Google Analytics alternative), so choose accordingly!

“We’re intentionally small, profitable and sustainable.”

This here ^^. Too many fall in the trap of agiling themselves to death. Instead, find the right people with the right skills and dont force process. Communicate well and frequently and you’ve got yourself a well oiled team. Current client has successfully managed to daily, retro and refine themselves to a drop in productivity by at least 20%. My invoices are coming in nicely, despite wasting days on meetings that could have been an email or a short call with the relevant stakeholders, but i’ll fire the client first chance i find one that actually wants me to work and not play “fun” t-shirt planning pokers.

Same goes for me. It's fucking insane we waste do much time

>“We’re intentionally small, profitable and sustainable.” This here ^^. Too many fall in the trap of agiling themselves to death.

How do you know they don't use agile? Being small, profitable and sustainable is not necessarily != Agile.

We're currently on the 500K views/mo plan and are more than happy customers.

They're affordable, ethical, and the product does exactly what I want and nothing more. I simply love Plausible and I'm happy to see them picking up steam!

Plausible's website metrics are public and you can see the impact this article has on their traffic (due to front page of HN, I presume).

Traffic goes from approx 4K visitors/day to almost 20K visitors today (so far).

https://plausible.io/plausible.io

Let’s not forget that these guys are extremely good at marketing.

One of the guys is extremely good at marketing. They only started making substainable money after the marketing guy joined.

What you see here; this article, the technical writing, getting buddies to hype on twitter, forums etc. is marketing directed towards a technical audience (who are the buyers of this product). I think they are both extremely good at it.

The writer of this article acknowledges in the very same article that they were hired as a marketer, over a year after Plausible was started (see Feb 2020). This article is part of their job marketing towards a technical audience.

> I think they are both extremely good at it.

Look at their revenue charts and the part where the revenue starts to grow is about a month after the marketer joined. One may have learned a trick or two but the other is extremely good.

It's like how when working on a team with someone who is amazing everyone on the team are able to do their job better and improve. This doesn't make everyone on the team amazing, this just means the amazing person made everyone better.

I have been using Google Analytics for years and it has become really crappy. Especially the latest release where you can't even see your daily traffic anymore. Everything is lagging behind, you can only see the actual data 2 days after the fact. On the other hand I have 3M+ page views, which would induce a $129/month Plausible bill. That's simply too much, especially because I have a very low CPM.

Might be worth self hosting Plausible https://plausible.io/docs/self-hosting

I don't understand how AGPL protects them from cloud providers hosting their stuff. The cloud providers could:

- Host the code unmodified, which is within the terms of the licence

- Host the code modified, and release the mods, which is within the terms of the licence

Surely the worry is they will just host and charge for it for less then Plausible can, which AGPL doesn't stop (to my knowledge)?

It could be that the "mods" of a different company are their whole IP. That is why they'd avoid AGPL.

And competing might not offer enough of a margin. Integration is hard, even if they save on the codebase itself.

Interesting. The case in thinking of is ElasticSearch, where AWS just hosted it themselves and it's easy to deploy, works with CloudFormation and billed the normal way. AGPL wouldn't have changed any of that.

This is one of the best blog post on company building I've read. Actually the only one I've loved! (All I've read are VC garbage)

I love posts like this because they help set expectations. I'm building a real-time SaaS (among other things) and I've got a lot on my plate.

I also use plausible.

This is very inspirational, it’s my dream to build a business like that.

It’s also great to see that it is still possible to build sustainable businesses without the more shady practices and growth hacking you often see, without millions of vc funding, without aggressive hiring. Just a small but strong team and a great product.

It looks like it was a long journey, I applaud the perseverance, and congratulate them on their well deserved success.

Can you comment on the impact that Elixir has had for your platform? Usually reports of technology decisions are biased towards the positive, so sharing any challenges that you've experienced along the way would be really helpful. Do you think that Elixir was the right tool for the job and if you could start over again, would you choose it?

I use plausible analytics to measure traffic on my observable notebooks. I did try with GA but it didn't work with Observable's iframe sandboxing. Because plausible is open source, I could fiddle with it a little and could get it to report the top level page user facing URLs instead of the iframe's. Very happy I can get traffic stats from a site I don't own and very happy I don't need user consent because it's GFPR compliant. It tracks just what I am interested... My traffic, and with a little config it excludes my own page views. Great product, perfect for my use case I wish them well.

https://observablehq.com/@endpointservices/plausible-analyti...

Looks like it runs on Elixir and ClickHouse, neat.

I’m also a happy customer.

There must be a lesson here in doing one thing well.

Thousands of people choosing to pay when there is a massively more powerful and fully featured free alternative.

Web analytics have also been around for decades, and even today there is no shortage of competition.

I think it really demonstrates how much more important execution is than an idea. Being the “simple” version of X or X for “niche user” can be a viable strategy rather than coming up with some revolutionary product idea. It also lends itself to competing as a small team, since larger competitors can't afford to pay close attention to those smaller verticals.

I’m a happy Plausible customer for https://allthecode.co - the privacy aspect is amazing but the killer feature for me? Actually understanding my website data and knowing how to find where traffic comes from and where is goes without needing to search for hours on end. GA is a disaster of UX now and this is wonderful in comparison.

> With all the new trials and customers, and an increase in larger websites trying our service, Plausible became much slower to use. (...) We moved from our PostgreSQL database to ClickHouse

Had the impression they were monitoring up to a couple hundred websites at that time. Probably small ones.

I'm surprised Postgres was slowing them down. Maybe the queries or the database were not well optimized?

If you need customizable reports (filtering and aggregating by any dimension), you cannot avoid scan over non-aggregated data. This is where you cannot optimize Postgres queries. Postgres becomes slow starting from just single million records.

Everyone talks about revenue -- but what about profits? What's more interesting to me are companies that make a profit after expenses.

This kind of indie companies are financially extremely efficient. I reckon between 80% and 90% of their revenue is profit.

It's a cool service, especially for those who don't need the more advanced features of Google Analytics.

The new Google Analytics is a mess to use, but at least it's no longer limited to ~"10M" hits for free each month.

Now please use that money to build more features!

I tell everyone to use Plausible AND a more in-depth events tracking platform -- you'll regret it otherwise if you ever want to track churn rates.

Why should they? They obviously got a winning formular.

What makes it such that the majority of such stories are in SaaS? Or are the SaaS success stories just the loud minority since they are very dependent on content marketing?

Not trying to rude but yes, this bootstrappers space is very noisy about MRR/ARR and there are very few of them I heard constantly. It apparently it is how it is .. to be done this way, otherwise only private channel sales is not going to be enough. If there are more successful bootstrappers, it's going to get worse (flooding environment where people hangout).

And most of SaaS that makes lots of money are in shovels business (during gold rush). Or probably so called "tools that help you make more money" (finger cross). Be it analytics, marketing content generator, social media post schedulers. Oh my.. I don't love marketing tools but they make shit ton of money of out B2C trying to reach millions of people.

getting people to subscribe at 100$ per month is an easier sell than asking (let's say 1 yr churn) 1200$ once

So, if an open source company changes their license, the code that existed before the license change still exists under the previous license right?

INAL but think the license is version controlled along with the code

Yes, the OSS licenses of the older versions of the software cannot be revoked (it's in the license texts themselves)

How did you build your landing page, it's very nice

How was your acquisition strategy for your first customers? Content marketing?

Looks like TailwindCSS.

Huh turns out plausible.io is blocked by my pihole. What’s the case for unblocking this flavor of analytics?

How did they eat for the first 2.5 years before they started paying themselves?

What are they using for creating the static content pages from the .md files?

Great read! Thanks for posting

> We’re a completely independent, self-funded and bootstrapped team of four

$250k/yr each, +/- payroll taxes, etc. etc.

Yeah, the cost of running the company (software, server, marketing, sales, etc..) is important. ARR never shows the true picture.

I haven't read much of the article, but they claim early on that they have no marketing spend. Do they say later in the article what their sales, saas, and server costs are like?

They don‘t have a sales team, so no cost. So it‘s barely sone SaaS they use by themselves (e.g. slack etc.) and server costs.

As they are running very lean I expect at least 80% margin.

They could scale to $10M without probably any extra staff, or maybe an extra SRE? It isn't Twitter: They don't need an army of bloat-adding squads.

7+ billion people can only dream of this kind of money.

That doesn’t include hosting costs. So likely closer to 150k/employee of take home pay.

Not in a million years would it take $400k to run Plausible. It would be a fraction of this.

The 150k/year also accounts for overhead such-as taxes, HR, cash reserves for rainy day, software licensing etc.

i guess they might as well pack it in and go work at a MANGA

Maybe in dollar terms, but assuming the business keeps growing: they have sustainable income, they don't need to worry about getting laid off, they don't need to play politics with leadership, and they can turn the business into passive income by hiring someone to take over.

At this stage they are a lifestyle business, its not hard to grow it even more if they get investments. If they can sell the business over $10 million, they already surpass what most corporate employees will earn in their career. That's a very attainable valuation in 2022 with a PM fit product.

And slave away for the rest of their lives?

I'm sure they're not killing themselves with work and I'd trade equity in plausible with any job

Revenue != profit

Anyone who things $250k is a normal salary has probably never left one of 20 or so top US tech cities.

Where I live, Sydney, it is impossible to get that in tech outside of executive management (which is not coding), rare positions at unicorns, and HFT C++ type jobs.

This is why so much of the best talent from Sydney, including a very talented former colleague of mine, are lured to Palo Alto and nearby cities.

Yeah it is tempting! I don’t think you even need to be “the best” right?

Are all 4 developers? it looks way too much. What's left for other parts of the business?

I say this as a founder of a $1.5 mil/year SaaS

It‘s one developer, one marketer and when I remember correctly two for support.

I like Plausible, and ethical analytics services in general. I'd certainly use them over Google Analytics. But it does frustrate me that Plausible (and others) take the stance that because they are doing what they can to preserve privacy, they have an absolute right to collect telemetry about users.

This includes things like CNAME cloaking, and adding a local JS proxy script so that visits can be sent back to Plausible's servers to make it harder to block for the user. The user has expressed a clear preference for their visits not to be logged, and Plausible (to satisfy site owners who want every visit logged) have done whatever they can to circumvent that.

I get it - it's a business, and making sure the site owners are happy is a big part of making money. But it grates that the whole thing is supposed to be about privacy while ways to get around privacy preferences are baked in.

What aspect of your "privacy" do you believe Plausible is violating? None of the (very minimal) data they collect is even linked to you in any way. There are no cookies or any other persistent identifiers. Raw web server logs (which you cannot opt-out of, for obvious reasons) are more intrusive to your "privacy" than Plausible.

Is it a "strawman" argument where people try to get you to respond to a point you haven't made?

No, it's a legitimate question. There is literally nothing you can do to stop a website from logging your visit, so I don't understand your argument here. Visit logging (what Plausible does) and tracking (what Google Analytics does) are two very different things.

Plausible is a third party that logs visits for analytics purposes. An end user expresses their preference (eg, with some sort of blocking browser extension) that the site doesn't send details off to a third party. Then the analytics service provides an easy way to work around this preference, and if that's blocked again then they provide another way, etc. They explicitly work around the end user's choice.

Why does it matter what the reason is for the end user's preference? Or if the data is being stored in a way that's currently difficult to deobfuscate? It's ironic that the whole push is "end user privacy", ie something that benefits the end user, but multiple workarounds are offered when the end user (for whatever reason) doesn't want their visits logged on a third party.

Saying that using an ad blocker equals a user expressing their preference to not have their visit logged to a third party is a pretty weak connection.

As long as none of your personal data is involved, it's not your data. It belongs to the website operator, and they can do with it what they please.

(Why is it these kinds of discussions get snotty so quickly?)

I don't know how else an end user can express their preference.

IP addresses are PII under the GDPR with enough context - although honestly I don't want to go down that horrible rabbit hole.

But sure, sure, the site operators can do whatever they please. It's just Plausible banging on about being privacy friendly and ethical seems a bit ironic and is frustrating to see.

Edit: The DNT flag is explicitly ignored by Plausible as well: https://github.com/plausible/analytics/discussions/646

Plausible is correct in ignoring the DNT flag, as it does not apply to the data they collect. Have you read the DNT policy that the EFF created, which is linked in that issue? It goes further than what the W3C defined, and it still does not apply to the data Plausible collects. In fact, it explicitly allows the collection of more data than Plausible actually collects.

You're really stretching definitions here, and have yet to make an actual argument about how Plausible invades your privacy. I'll just end my involvement in this discussion here, because it's clear to me that it will not be productive.

Okey dokey

Reply on news.ycombinator.com