GA4 vs Plausible vs PostHog: Choosing Analytics for Your Site or App

GA4 vs Plausible vs PostHog comes down to two questions: do you need to know what individual users do, and how much cookie consent friction will you accept to find out? For a typical business website I usually recommend Plausible, for an online shop that runs on Google Ads GA4 is hard to avoid, and for a SaaS product or web app with logins PostHog is the one that shows you what people actually do inside it.

The short answer

My rule of thumb is a single question: do you need to know anything about the individual user? If not, pick Plausible. If yes, and it happens after login, pick PostHog. Reach for GA4 mainly when your Google advertising depends on it.

Analytics is one of the jobs that only starts after launch. I've covered the rest in my guide to web application maintenance.

Web analytics vs product analytics

Web analytics answers questions about traffic. How many people visit, where do they come from, which pages do they read, and how many fill in your contact form? That's what most websites need, and GA4 and Plausible both handle it well.

Product analytics answers questions about behavior inside a product. How many new signups finish onboarding? Which features get used, and by which accounts? Are people still logging in four weeks later? To answer that, every meaningful action is stored as an event (a recorded action such as "invoice created") tied to a specific user. PostHog is built for exactly this. GA4 can technically do some of it, but it's awkward to work with, and Plausible deliberately leaves it out.

Before choosing anything, remember that your own database already knows a lot. New accounts, paying customers and orders per month can come straight from the database into a simple admin dashboard. Those are the numbers that run the business, and cookie banners and ad blockers don't touch them. Analytics earns its place on what the database can't see: where people came from, and where they gave up.

If the numbers will drive what you build next, my piece on prioritizing your product roadmap after launch picks up from here.

Cookie consent in the EU: where the three really differ

If you track visitors in the EU, EU rules generally apply wherever your company is based. The cookie rules come from the ePrivacy Directive and cover any storing of or access to information on the user's device, not just cookies. Analytics isn't strictly necessary for a site to work, so as a starting point it needs consent. The UK's PECR follows much the same logic.

GA4: consent, banners and gaps in your data

GA4 sets cookies and sends data to Google, so you need consent and, in practice, a cookie banner. Since the EU-US Data Privacy Framework took effect in 2023, transfers to the US can be lawful again under certain conditions. That framework only covers the transfer itself, though. You still need a legal basis, a data processing agreement and the rest of GDPR in place.

The practical problem is the visitors who decline. With the advanced version of Google's consent mode, GA4 sends cookieless pings for them, and Google can model their behavior with machine learning. But behavioral modeling requires, among other things, at least 1,000 daily users who accepted analytics cookies on at least 7 of the previous 28 days. Most small European B2B sites never get close. The result is that, by default, your GA4 reports only show the share of visitors who clicked "accept".

Advanced consent mode means Google receives pings from visitors who said no. Get that reviewed by a lawyer before you switch it on, or stick with basic consent mode, which only loads GA4 after consent.

Plausible: no cookies, but read the small print

Plausible sets no cookies and doesn't store IP addresses. According to Plausible's data policy, it counts unique visitors with a hash (a one-way code) of the IP address, browser and site domain, combined with a random salt that's deleted every 24 hours. That means the same person can't be followed from one day to the next, and the data stays on European infrastructure.

Plausible says you don't need a cookie banner, and plenty of sites run it that way. However, the European Data Protection Board's 2024 guidelines on the ePrivacy Directive make clear that the rules can also cover tracking without cookies, for example through JavaScript and pixels. The same guidelines say this doesn't automatically mean consent is required. My take: Plausible is the lowest-risk option of the three, but "no cookies" isn't the same as "no rules".

PostHog: your choice, your configuration

By default PostHog uses cookies for anonymous visitors, and PostHog itself recommends a cookie banner in that case. On the plus side, you can choose its EU cloud with servers in Frankfurt, and there's a cookieless mode that counts users with a privacy-preserving hash calculated on PostHog's servers. You can run it all the time or only for visitors who reject cookies. Session replay can mask inputs and text, and you should turn that on from day one.

In a web app with logins, it's tempting to assume your terms of service cover consent. They rarely do. Logging in doesn't change the rules for cookies and similar technologies on its own. And if you send names and emails to PostHog, that's personal data your list of sub-processors needs to include. My GDPR checklist for SaaS goes deeper on the technical side.

Which one I'd pick for your product

Marketing site or B2B website

Plausible. You want to know how many people visit, through which channels, and how many get in touch. Plausible handles that with goals on your contact form and on phone and email clicks. The numbers cover every visitor, not just those who accept cookies, and in many cases you can skip the analytics banner altogether. If you run Google Ads, GA4 can sit alongside it, behind consent.

Online shop

GA4 is often the pragmatic choice here, especially if you advertise on Google Ads or Google Shopping and need conversion data flowing back to Google. Most shops already have a cookie banner anyway. If your store is small and lives on organic traffic, Plausible with revenue goals (part of its Business plan) may be enough, with your shop platform's own reports covering sales.

SaaS or web app with logins

PostHog inside the app, ideally on the EU cloud if your customers are in Europe. The questions that matter are about activation and retention, and only product analytics can answer them. Start with a short list of events tied to customer value, such as "project created" or "teammate invited", rather than tracking everything.

The setup many SaaS teams end up with

Plausible on the public website, PostHog behind the login. The website measures traffic without a banner, and the app measures behavior for users whose consent you've handled properly. If you'd rather keep everything in one place, PostHog in cookieless mode can cover both.

What the three cost in euros

The license is rarely the biggest cost.

GA4 is free for almost everyone. Google sells a paid tier, Analytics 360, but it's aimed at large enterprises. The real cost sits elsewhere: a consent management tool for your banner (usually a monthly subscription), the time it takes to set up and test consent mode properly, and the data you lose from people who decline.

Plausible's pricing page currently lists €9 a month for up to 10,000 monthly pageviews (shown as $9 or £9 outside the eurozone), and the price rises with traffic. There's a 30-day free trial. You can also self-host the free Community Edition, but then you own the server, updates and backups, and it lacks some features of the paid service. For a business website, the saving is rarely worth the hassle.

PostHog gives you 1 million product analytics events a month for free. After that you pay per event: the next million costs around $50 (roughly €45), and the per-event price falls as volume grows. Session replay and PostHog's other products each have their own free allowance. It's fair pricing, but the bill can grow if you autocapture every click. Decide which events you actually send, and keep an eye on usage.

The hidden cost is developer time. Plausible is up and running quickly. GA4 with consent takes longer, because the banner, consent mode and testing all have to line up. PostHog in a web app needs someone to decide which events matter and tie them to users and accounts. That's a small development task in its own right.

When each tool is the wrong choice

GA4 is a poor fit when your site has modest traffic and you don't run ads. You pay with a cookie banner and incomplete numbers for features you never use. It's also the wrong pick if nobody on your team has time to learn the interface. Many people find it heavy going, and reports nobody opens don't help anyone.

Plausible is the wrong tool when you need to follow individual users over time, see where customers drop off inside an app, or watch session recordings. It leaves those out on purpose. At very high traffic, the price can also start to matter compared with a free GA4.

PostHog is too much for a five-page business website with a contact form. It also demands discipline. Without a plan for which events matter, you end up with piles of data nobody looks at, and session replay can easily capture personal data you shouldn't hold.

Keep in mind that analytics isn't monitoring either. None of the three reliably tells you that your site is down or why a user hit an error. For that, look at uptime monitoring tools for web apps and a proper error tracking setup.

Next steps

1. Write down the 3-5 questions you want answered every month.
2. Decide whether those answers need user-level data. No points to Plausible. Yes, and after login, points to PostHog.
3. Check whether your advertising actually requires GA4.
4. Sort out consent and your privacy policy before the tool goes live, not after.
5. Review the numbers after a month and remove goals and events nobody uses.

If you're planning a new site, see how I build fast business websites that are easy to measure. For a web app or SaaS product, have a look at my work on web apps and platforms.

Frequently asked questions

Can I run Plausible and GA4 side by side?+

Yes. Both scripts can run on the same site without interfering, and many teams do this during a transition to compare numbers. GA4 still has to wait for consent, while Plausible can load straight away. Expect the figures to differ, and decide upfront which tool is your source of truth.

Why does Plausible show more visitors than GA4?+

Usually because GA4 only counts visitors who gave consent, while Plausible counts everyone. The gap depends on how many people decline your cookie banner. Some ad blockers also block both tools, and they define a unique visitor differently, so the numbers will never match exactly. Watch the trend over time rather than the exact figures.

Is PostHog's US cloud fine if my users are in the EU?+

It can work, but it's more effort. PostHog's own GDPR guidance recommends its EU cloud in Frankfurt for EU users and notes that the US cloud needs additional data protection steps. If you're starting fresh and most customers are in Europe, choosing the EU cloud from day one avoids a migration later and keeps your privacy policy and customer agreements simpler.

What about Matomo, Fathom or Simple Analytics?+

Matomo is an open source alternative closer to Google Analytics in scope, and it can be configured without cookies, either self-hosted or as a cloud service. It suits teams that want lots of reports and full control over their data. Fathom and Simple Analytics follow the same lightweight idea as Plausible. I focused on these three because each covers a different kind of need.