Hunor.digital
Tool comparison

GA4 vs PostHog: web analytics vs product analytics

One tracks how people find you, the other tracks what they do once inside. The honest comparison, plus the case where running both is the right answer.

Lázár Hunor · Digital Fixer↑ Part of Choosing an analytics tool: GA4, its alternatives, and when to switchLast reviewed 05 Jul 2026
The short answer

GA4 and PostHog answer different questions. GA4 is web analytics: which channels and campaigns bring visitors who convert. PostHog is product analytics: what users actually do inside your product, with session replay, funnels, and feature flags. Its first million events per month are free. Many SaaS teams run both, deliberately.

The short answer

This is the comparison people ask for and the category error people make. GA4 and PostHog overlap maybe thirty percent, and each is bad at the other's core job.

GA4 exists to answer marketing questions: which channels, campaigns, and keywords bring visitors who convert, and how to feed that back into Google Ads. PostHog exists to answer product questions: where users get stuck, which features retain them, what happened in the session right before they churned.

If your revenue lives on a website (e-commerce, lead gen), GA4 or one of its alternatives is your center of gravity. If your revenue lives inside a product (SaaS, app), PostHog-shaped tooling is, and GA4 alone will leave your product team guessing.

What GA4 actually is

Google's free web analytics: acquisition reporting, conversion tracking, audiences, and a tight loop with Google Ads, plus raw export to BigQuery. Free at any traffic volume.

Its real strengths are the ad-platform integration and the price. Its real weaknesses for product work: it is event-based in name but campaign-shaped in soul. Reconstructing "what did this cohort of users do across their first two weeks" means custom dimensions, BigQuery, and patience. Nobody's product team lives in GA4 by choice.

It also carries the EU baggage: consent required before load, and data processed by Google under a transfer framework that is currently valid and currently being challenged in court. The full story is in the GA4 vs Matomo comparison.

What PostHog actually is

Product analytics built for engineers: event capture (including autocapture), funnels, retention and cohort analysis, session replay, feature flags, and A/B testing in one platform. You can self-host the open-source version or use their cloud, which offers an EU region for data residency.

The pricing model is usage-based with a generous floor: the first 1,000,000 events and the first 5,000 session recordings each month are free, then you pay per unit at rates that fall with volume. A small SaaS can genuinely run on the free tier for a long time; a high-traffic marketing site pumping every pageview in as an event cannot, so instrument deliberately.

Its honest weakness is the mirror of GA4's: marketing attribution. Channel reporting exists but is not the point, there is no ad-platform bidding loop, and your agency will not know what to do with it.

Where people pick the wrong one

The two failure modes I keep finding in audits:

Both are the same mistake: choosing the tool before naming the questions the business actually needs answered. That naming step is where I start every Tracking Foundation build, because it decides everything downstream.

Can you run both?

Yes, and for SaaS it is often the right answer rather than a compromise: GA4 (or a lightweight, cookieless alternative) on the marketing site for acquisition, PostHog behind the login for product behavior. The free tiers make the combination close to costless; the real cost is discipline.

Two rules keep it sane. First, one naming convention across both tools, written down, so "signup" means one thing everywhere; I install exactly this in Tracking Foundation builds. Second, one owner for the seam: the handoff event (signup, purchase) must be verified in both tools, at the destination, or the two datasets quietly drift apart and you are back to dashboards that argue.

Which should you pick?