Most companies are trying to make more money. They usually do that by trying to get more customers, hiring sales teams, or even increasing their prices. One of my sponsors is doing the opposite though. They're actually trying to make less money.
I know, crazy, but hear me out. PostHog is killing it. I went into this video kind of excited to see what they changed. I'm going to go off stream and go change a bunch of things we're doing because I am really impressed. PostHog already was the all-in-one product analytics solution I chose to use for basically everything I'm building.
which is why I reached out to them to try and get them to sponsor the channel. And now they have, and now they're paying me to just nerd out about analytics stuff, which is so cool. And they sponsored this video so I can showcase how crazy their pricing mindset is and how cool it is that they're being fair with their awesome open source solution for everything you need for product analytics.
If you do sign up after seeing this, make sure you tell them that Theo sent you or use the link in the description or in the comments when you do. Without further ado, let's look at why they lowered their pricing. I love their little graphic guy.
We actually have a version of him that's like VO coded. Really cool stuff. More companies need to have fun with their branding. If you haven't seen the post hog site, it's really cute. It's really nice.
And like what other company has the balls to put this in the middle of their homepage? Like there's almost like the antithesis of something like Vercel or the classic linear template that everybody's been copying for a while. Also of note, post hogs entirely open source.
and you can go join and give them a star. They're at 20K. It should be even more because they're the best open source product analytics tool by a ton. Super cool.
Also really fairly priced if you want to use their stuff. Anyways, enough about why I love PostHog. More about why they're making less money.
TLDR. We've optimized our session replay pipeline and infrastructure, and we're passing the savings onto you by making session replay drastically cheaper, especially at lower volumes. And that makes them the cheapest session replay tool amongst competitive products. There's also a part two, which we'll get to in a second. So here is the cost comparison.
Also note, other channel sponsor Sentry is in here too. I consider Sentry and PostHog to be very different products where Sentry is trying to figure out what went wrong and why. PostHog is trying to figure out what went right and how often is it going right?
What are users doing? PostHog is more about the things users are doing on the service and Sentry is more about what crashed and why. And I don't think session replay is a particularly big part of what Sentry is focused on versus a company like LogRocket, where that's literally all they do.
And PostHog, which used to be a much more expensive thing here. Now PostHog went from being the most expensive session replay to the cheapest and by quite a bit. So if you just want to see what a user does in a session, this is a great way to do it.
And that's this price is for 25,000 recordings. So if you record 25,000 sessions, so 25,000 people visiting the site, and you recorded them doing that only 85 bucks, the egress fees you eat on different providers might even end up being higher than that because of how much data is being transferred. And if you think about this annually, it's a really good point. Annually, that's $4,100 saved. So if you're doing 25,000 recordings a month, and you were happy paying the old price, you just say four grand a year, and they're making four grand a year less.
Thankfully, their growth has been enough to balance it out. But it's pretty nuts that they chose to go from the most expensive to the least expensive because this is not where they want to make their money. Keep reading, though, because there's more interesting stuff here. Many tools don't actually publish their pricing. That's their choice, but we're pretty confident it's not because they're secretly a great value.
How else are they going to justify those huge outbound sales teams, huh? That sound familiar? Sound a bit cloudy? Yeah, Cloudflare is the worst about this. They love hiding their prices.
But like, try to find how expensive it is to optimize images on Cloudflare. There's like seven different prices you have to factor in, and then they don't tell you about two others you have to factor as well. Because if you want to do...
image optimization on Cloudflare, you have to run code in a worker to fetch the original image and then optimize it. Because if you don't write your own worker code to process it, then you have to allow all domains, which means anyone can optimize images through your account, which is a terrifying security hole. But that's just the Cloudflare experience.
Because either you don't care, you do it insecurely, or you do care. So you do it the right way. And then you end up racking up a huge bill, or they rate limit you until you have to pay a huge ransom fee. And they hide all of this in the process.
Talk all the shit you want about companies like Vercel. At least they show you where all of the costs are, almost to the point where it's more confusing because they show you literally every single number breakdown of what costs you what, where, when, and why. That's what AWS does.
And as much as I love to shit on AWS, it's one of the things they do right. Surprise bills, at the very least, can be reverse engineered because you can see exactly what caused them to happen. I have no idea how much money Cloudflare is costing me right now.
I don't even know how much it's being used because they... hide all of that data. They hide all of the pricing. They hide everything.
And it's awesome that PostHog is coming out and taking a stand against it. Makes sense the open source company is the one doing it. As they were saying here, we're not just cheaper than replay tools designed for developers either.
Anyone using Hotjar for their website or web app can also save by switching to PostHog. This is cool too. Hotjar, we actually just talked about because I was covering an OAuth exploit that was discovered on Hotjar because they had a cross-site scripting attack that was possible for their analytics tool.
Fun. But also unlike Hotjar, PostHog actually supports session replays on mobile with their open beta for iOS and Android. Even React Native coming soon. So yeah, if you're a React Native person, be excited because they'll have that there as well. Yeah, pretty cool that you can do session recordings on mobile with the same project, the same web app, the same everything as you're doing on the web.
Dope. And yeah, these prices are pretty hilarious. Up to 5,000 recordings is free. Up to 25,000 is 5x cheaper.
250,000 recordings. It's only 543 bucks. That's crazy. That is crazy.
So why? Why would they do this? Well, they have a pricing principles page. They actually sent me the whole thing, which we can go through. But we'll do that in a second.
I want to read these two points first. The first point is that we should charge as little as possible while remaining margin positive. We don't believe in loss leaders or squeezing customers for the maximum that they're willing to pay, aka pricing to value.
Their goal here isn't to be a service that loses money in some places and makes it in other places like Cloudflare, or even with us with Upload where our free tier costs us a good bit of money. but our hope is that long-term, enough users will subscribe to the higher-priced tiers to offset that. And it's a tough balance to find.
The only loss leader that they have is their free tier, which they have optimized enough to be cheap enough that it's viable. If you compare it to other companies that I've worked with and been sponsored by, like PlanetScale, their free tier costs them too much money to maintain. And finding these balances are really hard.
And I know that these guys are working their asses off to do it right. They also called out that the true value of PostHog comes from using... all of their products together. We make money by onboarding customers onto more of them, not maximizing the profitability of each individual tool. That said, they have made it very clear that I can not only use but recommend that you start with and potentially only ever use their analytics tool.
And I'll be honest, that's the only one of theirs that I'm using right now. They have all these other things. Well, I guess I'm using web analytics and session replay too. But personally, I'm not using feature flags, A-B testing, surveys, data pipeline, warehouse, or AI engineering. I'm just using the product analytics.
I'm going to start enabling web analytics. I'm curious. And also the session replay, which I think I recently turned off because we weren't using it much. But you don't have to use all of the things to use PostHog.
The product analytics are so good that you can just use that and have a great time, which is what I've been doing for a long time now and highly recommend. People keep saying the pricing page is dope. Yeah, that's beautiful.
That is, yeah, that's really cute. As they call it here. Over 90% of companies use PostHog for free. You only have to add a card if you need more than the free tier limits, or the advanced features, or you just want to have more projects. So they're totally free.
You also click ridiculously cheap. This is so beautiful. And they show here, all the pricing is usage-based. Here is the exact cost of every single type of thing. So with analytics, it costs you 0.005 cents.
That's hilariously cheap per event. And it gets cheaper as you do more events. It gets to 0.005 cents.
Oh, oh, oh, nine cents. Once you have 250 million plus events, that's insane. So one divided by that, you'd have to have 111,000 events to be charged a dollar, 111,000 events per dollar. Once you get to the high tier, that's crazy stuff.
That's unbelievable. Even like the base tier. If we do the math for that quick, 20,000 requests to spend your first dollar.
Yeah. They're, they're not expensive and it's really cool to see. Also, if you decide you don't want to pay them, it's open source and you can go host it yourself.
There's a GitHub repo. You can go spin it up, connect to that URL instead with the SDK, and it all just works. So if you want to host it yourself and really maximize the costs, you can go do that.
Also, chat pointed out the first million events are free. This is unbelievable. It's hilarious.
If you compare this to something like, I was a big Mixpanel guy back in the day. I loved Mixpanel. I personally found post-hob to be the closest thing to Mixpanel in terms of the UX. $28 a month for 10,000 events.
I'm sorry. I'm sorry. That's so painful. Oh, God. Yeah.
And you can only have up to 20 million events monthly. That's not free of $20 million. That's the max you can have before you have to bump to one of these other tiers. And they still charge per fucking event because they like obfuscate all of these things.
Do you need more help? Yeah, you have to like, when you go to buy it, you have to right here. So a million events, which is, remember, a million events is free on PostHog.
$168 a month on Mixpanel. So I'm up to five mil. $612 a month. Best part is it's not usage-based.
You have to pick the right amount. And if you pick wrong, you either pay them too much. Or even better, you start losing events because your cap wasn't high enough. Great, isn't it?
Wonderful user experience, especially if they're trying to upsell you on the annual. So if you buy 10 million events on the annual plan, 10 grand, and it turns out you actually need a 15 million, you don't have to cancel and redo the whole thing and you lose any discounts you got doing the yearly plan. It's, whoa, you're in chat saying gambling simulator.
Yeah, that's effectively how it feels. I just. Sorry to roast the competition.
I know it's not what I'm supposed to do as a sponsor. They did not pay me to go shit on Mixpanel. That was not part of the plan. They might even be mad about it and want me to remove it.
I don't care. I think it's important to understand the state of this industry because they're doing this significantly better than the other options I've explored in the past. So much better. It's actually hilarious how much better. Even at this price.
Let's say that you keep this price for that 20 million events. Worst it can cost you is a thousand bucks. But if you start getting the cheaper price, which you absolutely would, 20, one, two, three, four, five, six times that 590 bucks. That is significantly cheaper. That is hilariously cheaper.
Don't even consider Mixpanel at this point. It's sad. And like, yes, on paid show, you should go fairly explore the other options. But I have spent more on Mixpanel than PostHog has paid me. To be very clear, I'm still in the negative for analytics providers.
Yeah. Anyways. The whole loss leader thing is important, by the way.
Loss leaders are bad for customers and companies. Nobody enjoys a surprise price increase when a company decides it wants to make money now or seeing products get sunset because they're not sustainable anymore. We believe in products that are both cheap and sustainable.
We can do this because we're efficient, we're self-serve, and we don't do outbound sales. The closest thing they do to outbound is me here right now. They never had paid an influencer. They never sponsored anything. I reached out to PostHog because they were the best option.
I could have went and taken way more money from another brand that I didn't like as much. But my goal with the core T3 Deploy partner sponsors was things that I knew, I trusted, and I could recommend to you as the end user, knowing that I've had a great experience shipping them. PostHog was what I chose to ship.
I hit them up. We talked a bunch and we figured out what a sponsor deal could look like because both of us were new to this at the time. This is why I love them though.
They don't want to waste all of their money doing these crazy things they see other companies doing. They just want to be a simple, focused, fair option. And the closest thing they've ever done to some sketchy marketing thing is pay me.
If anything, you could give them shit for that because that's the closest thing they have done to not just being the simple, obvious choice. But if you do want to use them and you want my next renewal to go a bit easier, make sure you go to postdoc.com slash Theo because then they know I sent you. There's no discounts or anything.
We haven't figured any of that out yet. But at the very least, they'll know I sent you. Links in the description in the top comment if you're looking for it. Also, I love their writing style here.
And this is just part one? Yes. What's coming next? Part two. When?
Next week. It's out now. I'll read that in a sec.
So what should you do now? Figure out how to spend the money that you've saved. Here are some suggestions.
Donate to the Hearts and Middlesex Wildlife Trust in the UK, where the hedgehog population has sadly fallen by 50% in over the last 20 years. Max, our hedgehog in residence, would appreciate it. They're so cute and round and lactose intolerant. That's great.
You could also buy something from their merch store. Oh God, their merch is actually really cool. Like unironically, some of the coolest company merch.
My other home's a data warehouse. Great shirt. I have a hoodie from them that I actually love. It's so comfy. It's so comfy.
I heart product for engineers. Love it. So good.
Signed founder photo. Legibility not guaranteed. These guys are the goofiest dudes too.
I love them so much. If you have the opportunity to like listen to anything that they have to say or just see them at an event do it they're so entertaining they're just like british goofballs that love what they're doing i actually don't know what the part two is i haven't had a chance to read this one at all yet so we get to discover this together so we decided to make less money part two let me keep killing you guys want to know my biggest complaint about post hog honestly by far my biggest issue with them their blog is covered with things that get in the way when i want to make videos about it I just have to delete things off the page. I'm going to annoy them and they'll fix it. They might even see it in the video and fix it. But yeah, if you go to the Post Talk blog.
and it looks less like there's a bunch going on, I've had my impact. Last week, they slashed the cost of session replay. Today, we're announcing part two, a change to our pricing that makes tracking events for anonymous users 80% cheaper. Okay.
Okay, now we're talking. Now we're talking. We'll see a 30% decline in product analytics revenue from this change, but it will unlock several new use cases, such as web analytics at a massive scale. as well as tracking high volume backend events that were previously prohibitively expensive. This is a really important thing.
If you're not deeply involved in analytics providers like someone like myself is, you probably don't know that many of them bill based on the number of user identifiers that they have. So if I go to your site and I'm not signed in, they don't really know who I am. So a unique identifier is created to represent that session.
If I then sign in, a new session is created. So you're effectively being double billed. But now if I'm a scraper, and I'm just reading the site and quickly running the code to see what's on it, you're getting billed a ton for all the events being fired by those scrapers, even though they never sign in. So what's the solution? The solution I and many others have chosen is to not run any of these product analytics tools until you have signed in.
The recommendation I have made for a long time is to use a service like Plausible. Plausible is another open source analytics provider. This one's built in Elixir, which I personally love. But the difference with Plausible is that their focus isn't product analytics.
It's not seeing... what users are doing, how many users hit these thresholds. Like with something like UploadThing, which is, if you guys don't know, the upload file managing service that we built for full stack devs to make file uploading significantly easier and better.
Really proud of what we have with UploadThing. But if you want to know how many users have uploaded this number of files, it's hard to get that information if you use something like Possible. If you use something like PostHog, you can tag the events with the user IDs and then filter user IDs that have had this event happen enough times.
With Plausible, there is no concept of a user or an identifier. All they track is traffic. So if you just want to see how many visits your blog is getting, or how many times people go to your homepage and where they come from, Plausible has been a good solution for that. But if you wanted to see what they actually do, what buttons they click, what features they're using, all of that, Plausible did not work.
Also, their pricings, fine. It's not great. I'm spending quite a bit of money on them now because I'm too lazy to move things over. Not the cheapest thing in the world, but it's good. It's open source.
It's been easy to recommend for a while. I think PostHog took that personally though, because they're going after that now. By focusing on anonymous users and making them cheaper, they're making it so I don't feel like I have to restrict PostHog until you've signed in.
Because right now, PostHog doesn't trigger on upload thing until you've signed in. We just use plausible on the homepage. I might go change that after stream now.
Like no front, they're not paying me to say any specific thing. They just paid me to talk about this in general and get my thoughts. My thought is I'm gonna go make a bunch of changes to a bunch of my services, so.
Take that as you will. So what's changing? Until now, we charge the same for all events that we ingested.
And again, this was the problem because somebody scraping my homepage is not as valuable as somebody uploading a file using my service. And if those are priced the same, I'm just going to kill the first one. And they know that and they're changing that.
As they said, this is standard practice among product analytics tools because again, product analytics and web analytics are very different. Tools like Mixpanel, Amplitude, and Heap all charge customers the same for... each tracked user session or event regardless of the cost to them.
But they're changing this now by splitting the product analytics pricing into two different event types. It's identified events, which was the default for logged in users, which includes custom data like the user's name, email address, ID, ways to identify specific users so you can identify specific behaviors. As they say here, identified events let you create user specific insights and cohorts based on person properties.
These will continue to cost the same. So again, if you want to see how many users upload this file type. That's what this is for.
But if you just want to see how many users go to the homepage and navigate it, anonymous events are good for that. Events that only include non-identifying data, like referral data, such as UTMs, domains, like where did they come from? Did they click a YouTube link? Did they click a Twitter link?
How'd they get here? As well as things like device type and basic location information. So all of that will be included still, but nothing about who the user is, not their email address, not their IDs, none of that.
80% cheaper for most of our customers now, and these will soon become the default event. That's really exciting. Making it the default event is going to hurt their revenue even more, but good on them.
The balls to do that is insane. In reality, if you're tracking both anonymous traffic like website visitors and identified users like login customers, you'll likely use a mix of both event types. We've updated our pricing calculator to help you better estimate your exact usage and potential savings. Yeah, insane. This means website owners who send 10 million events per month would pay just $324 per month if they switched to only sending anonymous events, saving $9,000 per year in the process.
So when should you use anonymous events? You should use them for things for tracking users who aren't logged in, or for use cases where you just need the behavior. So visitors to your site.
marketing sites specifically. Shoppers on an e-commerce site where you don't need to know every detail of what users are doing. In fact, if you're well implemented enough, you might have built your own system for doing product recommendations that you're tracking yourself not through a product analytics tool. So what you want from the analytics isn't necessarily what specific products did a person visit. What you're looking for is what UI are they navigating and what buttons are people pressing or not pressing.
Makes a lot of sense to have that be anonymous. Also logged out users of mobile apps. If somebody's in an app and they're not signed in.
Getting product analytics from them, seeing how they're using the app can be really cool. Tracking API and backend events. Oh, that's actually killer for us.
That's killer. That's going to save us so much money. Oh, fuck. Okay.
I'm even more excited now. Events generated by LLMs and error monitoring. Yeah. Yeah. This is great.
But if you're mostly tracking logged in users, you should continue to do so. Absolutely. So why are anonymous events cheaper? Their explain like I'm five version is that they cost less to ingest in process.
and now they're going to pass the savings on to us. So here's a slightly longer version. I'm actually curious.
When you call post hog dot identify, or you capture an event from an identified user, we have to do a bunch of processing to create, update, and merge person profiles in our backend. It's just one of the cool things. If you have a user that has like two different accounts, but they both are using the same email address, post hogs identify function will link those together correctly, which is really cool if they sign into mobile and web differently, stuff like that. But I'm sure it takes a lot of work on their end.
Yeah. They store that data in Postgres, which is separate from their events table in ClickHouse. If you're familiar with ClickHouse, it's a SQL compatible large data store solution that PostHog is largely built on top of.
So Postgres is being used for getting that user data because it's way faster. ClickHouse is not fast for querying, but it can handle gigantic queries much better. But since they have to bounce between Postgres and ClickHouse, this ends up becoming expensive for them to run. It's the most expensive and complex part of their ingest pipeline, which makes a lot of sense. But with anonymous events, they can entirely rely on the cheaper, faster ClickHouse tables.
For example, we use a session table to capture session properties such as the referring domain as well as the URLs visited. This lets us charge you much less for those events, saving you money and enabling many use cases that would have otherwise been cost-prohibited before. But why are you deliberately making less money?
I love these, like, they're just blunt and goofy. Like, all of their stuff just reads with this slightly sarcastic but very realistic British tone. It reminds me of, like, a Rich Harris from the Svelte world. It just, it's great. I would, God, I would kill to watch James from PostHog and Rich Harris have a podcast or just like an interview together.
That'd be so entertaining. I'll work to make that happen. But as I said here, they're deliberately making less money because it's the right thing to do for both our users and for our business.
Most of our competitors are inefficient. They employ huge outbound sales teams to grow their revenue. Their salaries and commissions for closing deals are passed on to their customers through higher prices.
I can tell you the amount of inbound I have gotten from different analytics providers, desperate. to get us on the platform, very real problem. And the price reflects this. There's like hundreds of people at each of these companies who do it.
They have to charge enough to pay their salaries. In contrast, PostHog is 100% inbound. They grow mostly through word of mouth and hopefully through you hearing about them through this video, and they charge based on the actual usage. We don't believe in loss leaders, so we make a modest positive margin on each event sold. What we charge is directly connected to what it costs us, not what we think we can get away with charging.
We grow our revenue through helping you grow. as well as onboarding you onto other tools like session replay, feature flags, and surveys. We think this is better than trying to squeeze you for every sense you have. You have a great experience, and we'll enjoy better retention as well as word of mouth for doing the right thing.
Oh boy, apparently there's a part three coming. I'll be sure to cover that when it happens. And a great, important last question. Would you rather fight one horse-sized hedgehog or a hundred hedgehog-sized horses?
And if you know anything about post-hoc, you know that they would never fight a hedgehog. Great stuff. This is awesome.
I went in excited, but this is even cooler than I expected it to be. If you want to learn more about how PostHog thinks about pricing and cost stuff, they have a bunch of articles about it. They have this pricing principles doc, which is how they think about pricing. I actually want to read a bit of this here.
Use PostHog for free if their hobby is your pre-product market fit. Experience the product before paying for it. Start paying when they're ready on their own with as few hurdles as possible.
Transparently pay for the value that they receive. So usage pricing. And paying for products so they only pay for what they're actually using.
Again, if you're only using a subset of PostHog's features and products, you should only pay for the things you're using. I know a lot of services make you subscribe to a crazy enterprise tier for hundreds of features when you only need two of them. They will never do that, which is awesome. And of course, make it a no-brainer to pick them over other competitors.
It is genuinely hard for me to think of reasons to use alternatives to PostHog at this point. I had an example earlier with Plausible, and they've even killed that. So here we are.
Yeah, you can check this out. I'll leave all these links in the description. They have a whole page on how they make money. what their actual margins and such look like, as well as this great article, enduringly low prices. We want our customers to spend their money on their engineering team, not on buying 10 software products.
That felt like a personal dig, but I get it. Yeah, great stuff. Let me know what you guys think in the comments.
Thus far, PostHog has been possibly the most positively received sponsor I've worked with. And I'm curious if you guys have used them too. Let me know in the comments what you think about PostHog and what you're using if you haven't tried them yet. Until next time, peace nerds.