Open source Google Analytics alternatives: Plausible, Umami, Matomo and PostHog

Why replace Google Analytics?

Common reasons are privacy rules, cookie banners, data ownership and an interface many site owners find hard to use. Self-hosted analytics keeps visitor data on your server and lets you decide what is collected.

Before choosing, separate two needs. Website analytics answers where visitors come from and which pages they read. Product analytics answers what signed-in users do inside an app. Tools are usually built for one of these.

Mixing them up is the most common reason teams switch tools twice.

Also consider who reads the reports. A founder who checks traffic weekly needs something different from a growth team building attribution models.

Which open source analytics tools should you compare?

The licence column matters mostly for agencies and SaaS builders. Running AGPL or GPL tools internally is straightforward; modifying and offering them as a service to others carries obligations worth reading about.

Plausible and Umami: the lightweight options

Plausible shows the essentials on one dashboard: visitors, pages, sources, countries and devices, with goals and custom events. Its script is small, and it is designed to work without cookies, which can simplify consent requirements depending on your jurisdiction.

Umami offers similar core metrics with a very simple setup: one Node.js app and a Postgres or MySQL database. It supports multiple sites, custom events and shareable dashboards.

Both are ideal for marketing sites, blogs and documentation. Neither tries to be a full product analytics suite.

Both install with a single script tag and can be self-hosted with Docker. Plausible also runs a paid hosted service, which funds development, and Umami offers a hosted option as well.

Their dashboards are simple enough that non-technical colleagues can read them without training, which is a real advantage over classic analytics suites.

Matomo and PostHog: the deep options

Matomo is the most complete like-for-like replacement for classic Google Analytics: goals, ecommerce tracking, segments, custom reports, tag management and an option to import historical data. It runs on a standard PHP and MySQL stack, and large sites need careful database tuning and archiving jobs.

PostHog focuses on products. It captures events from web and backend SDKs and adds funnels, retention, session replay, feature flags and experiments. Self-hosting it is heavier than the others, so many teams use its cloud and keep self-hosting for strict data needs.

If you need both marketing stats and product analytics, it is common to run a light tool for the website and PostHog for the app.

Matomo also has a hosted cloud option, and many plugins extend it with features such as heatmaps or custom dimensions. Check whether the plugins you need are free or paid.

PostHog’s strength is combining data you would otherwise spread across tools: events, replays and experiment results in one place. That makes debugging product questions much faster.

How to choose step by step

Keep the evaluation practical. Install two candidates on the same site for a week and ask the people who read reports which one they would keep.

Write down the decision and the reason. When someone later asks why a feature is missing, the answer is already documented.

• List the five reports you actually open each month in Google Analytics.
• Mark whether each is website traffic, marketing attribution or in-app behaviour.
• Traffic only: Plausible or Umami. Marketing depth and ecommerce: Matomo. In-app behaviour: PostHog.
• Check your consent obligations with someone who knows your jurisdiction; cookieless does not remove every obligation.
• Run the new tool in parallel with the old one for a few weeks and compare trends, not exact numbers.
• Decide what to do with historical data before you switch off the old tag.

How much does self-hosting cost to run?

Umami and GoatCounter run on very small servers. Plausible needs ClickHouse, which prefers more memory. Matomo scales with database size and needs scheduled archiving. PostHog’s self-hosted stack needs the most resources of the group.

Storage grows with traffic and retention, so set a retention policy early. Keeping raw events forever is rarely necessary and makes backups slow.

Put the analytics database on its own disk or volume if traffic is significant. Analytics writes are constant, and competing with your main app’s database slows both.

Common mistakes and where numbers disagree

Trust trends over absolute numbers. If two tools agree on direction and relative ranking of pages, both are doing their job.

Another common gap is forgetting to exclude internal traffic. Filter your own team, staging domains and uptime monitors, or they will distort small-site numbers.

• Expecting identical visitor counts across tools; each defines sessions and visitors differently.
• Forgetting that ad blockers block many analytics scripts, including self-hosted ones on obvious paths.
• Serving the script from a domain that privacy lists already block.
• Tracking personal data in custom events, which undoes the privacy benefit.
• Not setting up goals or events, then deciding the tool is too basic.
• Running Matomo without its archiving cron, which makes reports slow.

What can you build with self-hosted analytics data?

All four tools expose APIs, so you can pull numbers into reports, dashboards or an automation that flags unusual traffic drops. Plausible and Umami keep the API simple, while PostHog offers a query interface over events.

A useful pattern is a weekly summary: a workflow fetches key metrics and an LLM writes a short plain-language digest. RepoLoot’s catalog includes data and automation projects you can combine this way.

Owning the database opens options closed platforms make awkward. You can join analytics with your own sales or product data in SQL, build custom dashboards and keep data for exactly as long as you choose.

Agencies often run one instance for many client sites and share read-only dashboards. Plausible and Umami make multi-site setups simple, and Matomo supports detailed permissions per site.

Keep the raw data model in mind when joining. Cookieless tools deliberately avoid persistent visitor identifiers, so per-person journeys are not available by design.

Documentation sites benefit too: tracking search terms and page exits shows exactly which docs confuse readers, and that feedback is hard to get any other way.

Webhooks and scheduled exports also let you push goal completions into a CRM or ad platform, though you should confirm that doing so fits your privacy commitments.

Keep automated reports short. A digest that names the three biggest changes of the week is read; a full dashboard screenshot rarely is.

How do they compare on features people ask about?

These are the questions marketing and product teams most often raise during a switch. Verify details against current documentation, as each project adds features regularly.

Frequently asked questions

What is the best open source Google Analytics alternative?

It depends on the need. Matomo is the closest full replacement with ecommerce and detailed reports. Plausible and Umami are best for simple, privacy-friendly website stats. PostHog is the choice for product analytics inside web or mobile apps.

Do these tools need a cookie banner?

Plausible, Umami and GoatCounter are designed to work without cookies, and Matomo can be configured that way. Whether you still need consent depends on your jurisdiction and exactly what you collect, so check with someone who knows the rules that apply to you.

Can I import Google Analytics history?

Matomo offers an importer for Google Analytics data, and Plausible has provided import options as well. Imported history usually differs in structure from native data. Many teams simply keep an export of old reports and start fresh with the new tool.

Is self-hosted PostHog hard to run?

It is the heaviest of these tools to self-host, because it relies on ClickHouse and several supporting services. It is manageable for teams used to operating data infrastructure. Smaller teams often use the hosted version and reserve self-hosting for strict data requirements.