Building successful products shouldn't feel like a guessing game. In this video, we'll explore the differences between PostHog and Mixpanel, including pricing, futures, and of course, ease of use. You may think you know which one is better, but I'll cover some surprising facts that I think might just change your mind. Let's get started. Let's start with PostHog. This is the sort of open source upstart founded in 2020. So it's a relatively recent company. You know, Mixpanel was founded back in 2009 and if you look at the home page here it is clear that POSOG is positioning themselves for the developer audience the main tagline of course is for developers you know mixed panel amplitude historically have sort of positioned themselves either for product teams or marketing teams which on average tend to be less technical of course than most developers now POSOG also talks about products it's an intriguing take I think it's kind of kind of interesting actually you know they they have this different products here i think did the research for this video you know there were about 10 products uh and you know there's product analytics there's web analytics session replay future flags experiments surveys data pipelines data warehouse and ai engineering and we'll cover most of them in some level detail at least in comparison to Mixpanel of course is unused products they tend to talk about reports and futures uh but there's some equivalence and there's there's some overlap here between them now the other thing that makes PostHog really interesting is their open source component. So if you look at the GitHub repo here, you know, down here, we can scroll all the way down, you know, we have the same idea, the same products. And you can, of course, install this for free. You know, generally, there's a cloud version, which is what we're going to look at. And there's the open source version. So this has been part of the attraction for PostHog. One thing I'll know right away, though, is that even, you know, PostHog themselves say that this open source deployments can really scale to approximately 100,000 events per month. You know, 100,000 events per month is not a lot of events. Of course, this might be very early stage companies or products or startups. So something to keep in mind. Now, it may be possible that you might be able to scale this up beyond their recommendations. That's something that I don't think it's impossible. But just to give you a sense of where the open source falls into play. So if you are a large scale enterprise company or even just like a mid-sized startup the open source may not be a viable option for you nonetheless again you might want to dive into the technical details to understand exactly where this 100 000 limit comes from but this is just postdoc themselves kind of giving you constraints as to where the open source version can play that being said we are going to look at postdoc here in more detail right there's a lot here i think their their web page or documentation is very strong very detailed it might actually fall in line with that ethos of making really developer-friendly products. There's a lot the product can do. It's clearly this all-in-one solution, from product analytics to AP testing to future flags to data pipelines. But is PostHog doing too much? You know, are they just having too many products that spread too thin? Maybe. Before we cover that, though, let's look at Mixpanel in some detail. Mixpanel, of course, one of the original software companies in the product analytics space. Today, of course, they work with growth teams and marketing teams, and they can do much more than just product. So they've been around for a while. And I mentioned they don't really have products, but they have reports. And, you know, a lot of the major reports that you might see in PostHog are here. The segmentation or insights, depending on what you call it. Funnels, retention, cohort analysis, and so on. Mixpanel has been around for a while, which means this is a very mature product. 15 years in the making. And I think the maturity does translate in a few different ways. One is just the small details. You know, I find when you do analysis day to day, and you're trying to build things, there's a lot of shortcuts that eventually get added, that I think Mixpanel will have a little bit of an edge over POSOC. This might not be a huge deal breaker, of course, depending on level of sophistication, but it's just something to keep in mind. Now Mixpanel has also kind of come up on a different philosophy, you know, from basically like 2010 to 2020, the approach by many of these companies was to just be the best in class at what they do. So if you're going to do product analytics, be the best product analytics possible. In the last few years, definitely from about 2021 to today, 2024, when they record this video, there's a bigger push now to do everything. So of course, Mixpanel has session replay. They have a CDP, like a customer data platform, component. It's not as sophisticated as what Segment.com might offer, but it's there like a light CDP, There's, of course, data pipelines to take data out and take data in. And we're seeing the same thing with Amplitude. This video is not about them, but generally the space is going in that direction. So the POSAC approach is becoming the approach for many companies just to add more and more futures to a single solution. And of course, Mixpanel has been built for a non-technical audience. It's a generally technical product. Now, you need developers to implement it, but historically has been built for people who are in marketing or product marketing or just product in general. And you're going to build queries in this sort of pseudo SQL interface that you can get around in multiple clicks, drag and drop things and so forth. So very solid product. Of course, you know, 15 years of making a very strong track record. There's lots of videos in my channel about Mixpanel. But of course, the question then becomes, is Mixpanel falling behind to upstarts like PostHawk? Have they sort of lost their edge and gotten complacent? So let's see how they both stack up across some of the major categories I want to look at here. And then from there, I'm going to give you some ideas for how to make a decision between these two tools. So we're going to do a side-by-side analysis of both PostHog and Mixpanel. I won't cover everything, but just the sections I think are the most important and relevant to our discussion here today. Here's what we're going to look at. You know, product web analytics, session replay, feature flags. Experiment survey, data pipelines, and a few other things. So let's get started with product and web analytics. Now, PostHog actually splits this into two, product and web analytics. It's not very common to do that. You know, typically we're talking about the same sort of event model underneath. You know, we're talking about events. on the web analytics side it's only really a handful of events it's really just page views maybe screen views if you're looking at mobile apps and then from there you can sort of build out sessions average session time Postdoc has a few things that are sort of very reminiscent of the Google Analytics world like bounce rate and then product analytics has to be anything that happens after a user signs up so this starts with a sign up or a login and then a user goes to your product and then engages whatever actions are available they may watch videos they may complete forms become subscribers make purchases and so forth. So I would actually combine this two into a single thing, because I think it just makes more sense. You're going to be using roughly the same reports. So on the postdoc side, there are some interesting things that has it, I think makes funnels doesn't have at all. So I think when it comes to the model itself, like, you know, you're tracking events with event properties and user properties, and you might call them differently in each product. I would consider them quite similar to each other. If you go down here to some of the reports, you know, the things that you're going to find on both tools. Funnels, graphs, user paths, retention, some form of stickiness, dashboards, those are all going to be there. Now the things that are kind of unique to PostHog, or at least relatively unique to them, some of the correlation analysis, you're not going to see as much. Mixpanel has a couple things that may do some of those elements, depending on what you want to do. So it may not be 100% different, just some small overlap. And then, you know, the HogQL, which is our PostHog's SQL own abstraction and language they have built, on top of the data. Now Mixpanel actually has something called JQL, which is their own kind of abstraction used on JavaScript, which is not as common today, just typically because if you want to use SQL on your raw data, typically you just take it to a data warehouse like a BigQuery, a Redshift or something and then do it there. But Postdoc has kind of built a way to actually interact with the data directly within the Postdoc interface. Again, it could be handy, could be not depending on what your existing structure or sort of data pipeline might look like. But I would consider them relatively equal in this manner on the product and web analytics portion. If I just scroll down to down here at the libraries, I do think Postdoc likely has a little bit more libraries than Mixpanel. Some are pretty standard. You're going to see them everywhere. The JavaScripts, the React natives, of course, native Android, native iOS. There's Backend, there's Ruby. Postdoc might have a couple more, but this again will depend on what your frameworks are for your product. And for most companies, you're working with something pretty standard, so it may not make a huge difference. But nonetheless, what I want to understate here is that from a technical perspective and perhaps from an ease of use perspective, they're pretty equal in this area, on this product and web analytics portion. Mixpanel, of course, again, some of the reports themselves might have a little bit more details, right? These are some of the dashboards, you know, they talk about engagement, something interesting too. I think, you know, you see Mixpanel talk more about sort of the outcomes you might get here, the retention, the loyalty, less so about the specific reports, but they're there. The funnels, insights or segmentation, the ability to slice things, to break things down, to then save reports and so forth. This will be here. And, you know, Mixpanel added some custom things that are sort of very special to web analytics, like be able to track session duration. Some things are not here. I think bounce rate stood out to me that postdoc tracks. Typically, bounce rate comes from the Google Analytics world and you have to calculate it, right, based on when the user actually stops or exits a page. So that's something that you might not have there. But I also think bounce rate may not be as important to you. Again, depending on what you care about. Next, we have session replay. Another interesting function that used to be done by tools like Hotjar or Full Story. And now a bunch of tools are bringing it into their own product suite. Session replays are pretty standard. You know, they typically work on the web. They capture the DOM. And then as a user moves, the DOM gets captured over and over again. And there's some software that kind of stitches together. So you get this video-like experience. It's not an actual video. It's not an MP4. Nonetheless, both of them have session replays. So this will be available here. I think the session replay functionality is pretty standard across all tools, if I'm honest. You know, you want to look at the events users are taking, the mouse path, the scroll path. Once you have session replays, you can sort of build heat maps on top of it. So this is pretty standard. Again, you may even have a hot genre full story. You may not even care about this. It is nice once you kind of have to start to centralize your data in one place. So you can open the profile of a user, see all the actions, the events they took, and then play a session replay off the same place. So there could be some convenience there. And again, a lot of the proc analytics tools are moving the space amplitude also added session replay recently so this is becoming a pretty standard future again for both of them i will consider them pretty equal next we have future flags just the idea just the ability to be able to roll out features to specific groups and kind of see the performance of those features before you roll them out to bigger groups again something that really makes sense from a development perspective or even maybe from a product perspective this is something PostHog has Mixpanel on does not has this at all built in you could do it with an external integration if you so want to for example here's a tool called launch darkly they have they handle the future flags for you and then they can integrate with Mixpanel so they basically send data to Mixpanel and they can build reports inside Mixpanel for it again another thing that if you don't have something your tool for it very handy to have it built in in the all-in-one of post hoc if you do have it maybe just a matter of being able to integrate it with whatever tool that you care about next we have experiments this is a b testing primarily on web and here really PostLoc is actually not competing with Mixpanel who doesn't have A-B testing experiments anymore. But it's really more competing with the WVOs of the world so you're able to of course report an A-B testing experiment but also build them out I think they have a beta future right now for building no code A-B testing experiments you can build them with code so very handy this in itself is actually quite powerful you know typically you might need a dedicated A-B testing tool but you get this within the within the same sort of solution of PostHog. So very valuable to have A-B test and build them with the same product. Mixpanel has something called experiments, but it's actually not A-B testing experiments in the ability to design an experiment and launch it, like, you know, like say Google Optimize when it existed. This is actually just about being able to report A-B testing data. So typically, if you have an A-B testing solution like a WVO, you would then integrate a Mixpanel and send data to Mixpanel in a specific format down here. This is sort of a special Mixpanel.track event and as long as Mixpanel sees this data then it can kind of give you a special A-B testing reports that look at statistical significance and a few other things that make sense for this data. So again, you're not really running the experiments you're just simply reporting on them. I have worked with some teams recently where they actually find it quite straightforward to build the logic to show different experiments so if you have two different pages they can sort of build the logic to show one page to 50% of users and another to. Another 50% of users but they do find it quite tricky to build reports. So I think this is why Mixpanel has added this. They assume that some teams are going to be able to just deploy their own AP test experiments, their own logic, and just simply send the data to Mixpanel and then Mixpanel can do all the reporting for you. Again, different approach, definitely a little bit more work than the all-in-one solution that Postdoc might offer. Next, we have surveys. Again, Postdoc has the ability to launch surveys within the app itself. You know, what you expect in surveys, different question types, templates, things like that. Mixpanel doesn't have this at all, so you will need to integrate it with some kind of third-party solution like a SurveyMonkey or something else. Again, another sort of plus side on the POSOC side. Next, we have data pipelines. This is the sort of the CDP customer data platform functionality that many tools are trying to add now. This is really the realm of a segment.com. All of these tools, PostHog, Mixpanel, Amplitude that have added their own CDP. It's really a light version of it. You know, segment.com, I have 300 plus integrations. In here, we might see 25, 30, 40 tools. It's growing, of course, but it's really a game of integrations. So the data pipelines aspect is typically broken down to a sources and destinations element. so you can bring data into a tool and then take data out. I will consider them equal in their sort of CDP capabilities. And it actually really depends on what integrations you care about. So you may have pretty standard tools. You know, you use Salesforce or HubSpot and you have Facebook ads and Google ads. In that case, you're probably going to be fine. But if you have something that's a little bit more obscure or a little bit more special, you really want to double check are your integrations here supported in the way you expect them. You want to bring data into Polsog? Is it a source? You want to take data from Mixpanel into some third-party tool? Is it a destination? Here is Mixpanel, right? The same thing. They call it integrations. And it's just the ability to send data to destinations or, of course, bring data in through sources. And again, you have to go to the list to understand if the integrations you care about are here. There's also a question about data warehouses. And the standard ones are always supported out of the box. The BigQuerys, the RedShifts, the S3s. And those are pretty sort of straightforward destinations for these data pipelines. We're going to look at the price and actually have the data pipelines in the last portion here before we move on. But again, this sort of light CDP concept, pretty equal for both of them. But definitely a lower functionality than you might expect from a segment.com. Next, we have this concept of data warehouse. It's actually a little bit different from a data pipeline. So if you're thinking, you know, we have a BigQuery, we want to take our data out, that's actually data pipelines. Postdoc gets actually built what they call the data warehouse. And it's actually be able to bring data in from different places. You know, there's examples here for like HubSpot or Postgres or Sendus or Stripe. And then you sort of link data within this custom data warehouse they have set up and then run SQL on top of it, also within the Postdoc product. So the first thing I'll mention, of course, Mixpanel doesn't have any equivalent to this. So right away, the second thing I'll mention is that this is a little bit unusual. It's kind of like a cool functionality. I haven't seen it too much. Typically, if we're talking about data warehouse, we're going to go with like a BigQuery, a Red Chip or an S3, bring everything in there through different ways you know different ETL processes and then any kind of analysis or queries where you can build them a data warehouse and then find a way to visualize that perhaps like a Tableau a data studio looker or something else so the fact that Postdoc has kind of gone on and kind of built their own kind of almost like alternative to a data warehouse it's intriguing again it will depend on what you have right now what are the limitations or differences compared to just doing this on yourself for like a BigQuery. Redshift approach What are some of the sources? How well does this sort of hog QL compare to just regular SQL, just vanilla SQL, and so forth? It's intriguing nonetheless, right? And I think, you know, they build it out and just I wonder how practical it is when there's a very sort of very standard way of approaching data warehouses. So not wrong or right, it's just a different way of handling data warehouses, which again, depending on what your team prefers, may be what you're looking for, or it may be a little too custom or a little too specialized compared to just doing something straight off Redshift or BigQuery. Finally, let's look at pricing. So let's build it out together just so you see some of my assumptions and how this might look like. So down here, Postdoc has this calculator for estimating pricing across different things. I think it's quite handy. So let's build a few things. So first, let's include product analytics and we're going to take 2 million events here. Then we have website analytics. And again, this is anonymous data. So we're going to have another, let's say 3 million events of anonymous users. And then I want to take some API events. This will be sort of backend events that we might have for the most important actions that we want to track. Typically backend events are only for identified users. You can, of course, do anonymous data, but again, generally that's kind of what we use backend for. So let's do one and a half million events. So this is six and a half million. Then we're going to add a couple add-ons that might be handy here first let's take the data pipelines because we want to take this data out at the very least like a big query like a redshift then let's say we're going to take group analytics and this is actually the ability to create companies or groups alongside you may have events event properties and users are those events are attached to but if you're like a b2b company you may have a group itself they want to attach multiple users to and this is what group analytics allows you to do again very common concept access in Mixpanel, Amplitude, and a bunch of other tools. And let's look at a few other things. So Mixpanel has some replays. The plan we're going to look at actually has 20,000 records a month. So let's just match it just so we're even here. So 20,000 recordings. We're not going to take feature flags because Mixpanel doesn't have it. They don't have experiments, of course. We're not going to look at. Neither surveys or data warehouse. We're not going to take any of that because, again, they don't really exist in the Mixpanel world. So we're up about $870 here a month. I'm sure there's some annual discounts if you want to go that way. But just you get you get a sense of what this case study might look like here. So Mixpanel, we start by choosing the events we want. So it's kind of go something similar here, you know, we chose six and a half million events. Again, there's no really distinction here between anonymous or identify events, which postdoc actually has, which is interesting, we just have a number of events that we split out in any way possible. Typically, this means that if we have limits of events, and we want to stay under those, then we start to figure out how we can slice and dice it. Maybe we don't track all the anonymous users, only the users who arrive on our pricing page or the signup page or something like that. So that's kind of ways to play around depending on what you have. So let's kind of go with that. Then we're going to have the data pipelines. Again, pretty equivalent functionality to what Postdoc has. And they have group analytics, pretty also equivalent functionality to what Postdoc has. Then we have the session recordings. Again, you know, Mixpanel includes 20,000 in the plan. This plan that we're looking at, this growth plan. And we did the same thing on the post hoc side. So we come up with $1,255.81. That's $1,255.81. Now, of course, remember that post hoc is around $870, which means, of course, the post hoc is lower than Mixpanel. And, you know, we do have some flexibility in the post hoc pricing, you know, depending on how much anonymous versus known or identify events we have. so that may even give you more flexibility. This is before even taking into account any of the. Annual discounts you might get with either tool. So this pricing advantage might be important, especially for early stage companies who are careful about spending, but really, you know, almost every company, you know, has got to be thinking about spending and how to reduce costs that you may spend on another tool, especially if you have other things in your tool stack, like a CDP, some kind of communications platform like a brace, and so forth. So being said, how do you actually make a decision, right? Some things exist in one tool and in others, maybe one tool is stronger on one thing, but not the other. How do you actually make this decision for your company? So let me share some ideas in the next section. So when we want to make a decision like this, like a postdoc versus Mixpanel, or even really anything, I mean, you could be looking perhaps postdoc versus Mixpanel versus amplitude or something else. I think it's handy to look at several things. First, we want to look at the stage of your company. If you're an early stage company and the open source actually fits you well because you have fewer than 100,000 users, honestly, that might just be the best fit for you. And that means you can start with PostHog. And if you ever cross over the 100,000 and you start to see issues, there's likely a straightforward way to transition to the cloud version with little downtime. So that in itself is a fantastic tool. If you're not early stage or perhaps a midsize to an enterprise company, then you're going to have more special requirements and you really want to start thinking about the long term. What does your stack look like so what does your martech stack look like right now do you have a cdp uh do you have existing a-b testing technology do you have email uh technology like a brace or a custom rio if not what kind of pieces do you see coming up in the future maybe you're going to need something like custom rio or like a brace and you're going to need something like a cdp so you start to make some decisions as to what the next six to 18 months might look like and from there you might realize hey you know it might make a lot of sense to have Mixpanel because it's going to integrate well with this two other three tools that we think we'll need or you know what this postdoc looks good we don't want to. Complicate our lives by having standalone tools so let's actually just take this one tool and use it as much as possible if we need something custom we'll build on top of it which you can so you you start to have a bit of a different approach. It depends where you kind of see yourself and what you see other competitors and similar companies doing to yourself. You know, generally in any kind of industry, you have a stack of tools that becomes common. You know, everyone's using Amplitude or everyone's using Mixpanel or everyone's using Hopspot. So, you know, you're kind of sticking around to similar tools. I don't know if tool technology is sort of the place to innovate when it comes for companies. You want something that's going to work that can allow you to handle the most common use cases that your industry faces. When it comes to communications, to product analytics, to data collection, data ingestion. So I want to think about that. Just like if you're going for a data warehouse. You're probably going to be choosing between BigQuery, Redshift, Snowflake, maybe one more, right? You're not going to come here and say, you know, we're going to use some brand new data warehouse. This is probably not the stage to do that. If you're enterprise level, you're a large company, then you're going to have so many specific requirements that it's going to limit the field for you. Do they have SSO? Do they have the right privacy requirements, the right legal requirements? Is this going to play well with our team? Do we have engineers who understand the languages that we might use here, and so forth? My general sense is that POSOC seems to be a better fit for smaller companies rather than larger companies, rather than the enterprise. They do have enterprise plans, of course. I think Mixpanel tends to be a better fit at the midsize level, perhaps a little bit larger. And you know if some smaller companies might want to use it thinking of the long term that might also be a good fit but also remember that Mixpanel doesn't have a lot of functionality that you might need so if you're early stage and you go with Mixpanel you might still need to find something for a b testing and for future flags and so forth so that's why i think maybe mid-size might be a better fit for Mixpanel next besides stage you also might want to think about what futures are really important to you so a b testing may be completely relevant to you for example because you just don't don't have the volume for it but you really need the product and web analytics, or you really need the future flags, or you really need the surveys, whatever it might be. Now, if you only want, let's say the web slash product analytics, the session replays, and the data pipelines, honestly, either tool will work well. If you want something more specific, like a surveys, future flags, A-B testing, then you're probably going to lean more towards the PostHog. Third, we really want to look at integrations. So they have this sort of light CDP component, but again, it all comes down to what they support. So whatever you're thinking about, whatever the tools are common in your industry, are they there? If they're not there, then you really want to start thinking about what does it look like? You know, is there a custom way to pull data from a source or is there a custom way to send data to a destination? These are things that are becoming more common. Typically you write some code, you know, in JavaScript and you can pull data from an API and then send data to a POST API. So just want to think about a little bit of what this integration would look like. The name of the game in TechSax is always about integrations, you know. The only one solutions are common in the early stages, less common in the later stages because typically you would need something a little more sophisticated, you know. Eventually your communications is so complex that you need something like Brace or Customer RIO. So you really want to start thinking about integration from the very first day. Even if you're not using them right now, what does this landscape look like in six months or 12 months? Is it going to be easy to integrate things? Can we expect that are the tools we will use are going to be here? And if not, can we integrate with them somehow? So think about integrations quite heavily because again, the only one can work up to a certain stage and that stage may vary, may be debatable, but eventually it will become about integrations and you want to make sure that whatever you care about, will be there in 12 months. Lastly, let me tell you a few things that will help you make the best decision and make it easier to migrate wherever you have to in the future. One, you want to design a great taxonomy. I have spoken about this endlessly in this channel. This is the tracking plans that you see online. Just want to make sure you have all the events with all the event properties and all the user properties that you care about. Really spend a lot of time designing this taxonomy because this is your fundamentals or builder blocks for whatever tool you use. Second, you really want to get all this data out, the raw data into a data warehouse you control. So this might be a BigQuery, a Redshift, a Snowflake, whatever it might be. You might not use it. You might not do anything with it, but you want to have this integration and collect all this data. Because again, if you want to migrate to another tool, doesn't matter what it is, most tools tend to support the import of historical data if you have it. So if you have the data stored, it just matters where we're writing a script and then bringing that data into a new tool, relatively straightforward process. And third, you really want to build skills around working with data. So. There's some custom things that you might see in any given tool around how you build reports and how you slice them. It all kind of boils on to SQL at the end of the day. They're all kind of built in this sort of pseudo SQL approach. So the more that you understand data, how to slice it, how to break it apart, how to work with it, then the more that you'll be able to see different tools and say, OK, I get it. This tool thinks about the data model in this way. So if I just do that, I'll get the answers I need. And you want to do that within your company. At least a handful of people within your company become really good with data. So you become less attached to tools and just more flexible as you work with that raw data that you've been collecting from a great taxonomy you can sort of take it into any tool and then make the most out of it we talked a little bit about how to make the best decision for your team thinking about the stage of your company thinking about the products and futures you might need the integrations you might need in the future and of course some of the perhaps those long-term skills that are needed regardless of what tools you use and whether you choose post hoc or Mixpanel you're likely going to end up with a really solid tool. Now, my name is Ruben Ugarte. I'm an expert in data and MarTech. And before you go, I want to recommend one more video for you. Now, I don't actually have a tutorial video for PostHog yet. I think it's in the backlog for me. I think it'd be fantastic to see all the different steps to kind of get PostHog going, get the data in, show you some of the reports. So that's coming soon. But in the meantime, I do have Dart to Finish tutorial on Mixpanel if you want to see what that implementation might look like. I'll link that video here on the right and you can check that out. It's about, I think, 45 minutes to 60 minutes so it'll give you a great sense of what the Mixpanel competition looks like and stick around. I'm sure there'll be a video like that on PostHog in the near future because I think this is a really cool tool that I'd love to explore a little bit more and kind of see how it keeps playing and the strengths and the weaknesses of that. So, until next time, we'll talk soon.