Matomo vs Google Analytics

Notice the first two rows. Matomo is the only one of the three you can either pay someone to host or run on your own server for free.

Most comparisons treat Matomo as one product. It's really sold two ways, and they differ in price, in features and in how much work they leave to you.

Notice the A/B testing row. Cloud includes it; on your own server it's a paid plugin, or part of the top bundle.

That's the pattern. Cloud includes what On-Premise sells as plugins, within allowances: funnels, heatmaps, session recordings, A/B tests, custom reports, Search Console keywords and the GA importer. Heatmaps and recordings are capped by tier, from 1,500 heatmap page views and 150 recordings a month at 50,000 hits to 10,000 and 1,000 at 1 million (pricing). The Business plan covers 30 sites and 30 users, which suits an agency with a handful of clients.

On-Premise is the same open-source code, free for any number of sites, users and hits. Since 2026 Matomo sells the extras in bundles as well as one plugin at a time, and existing customers can keep their old prices. You aren't locked in either way: Matomo imports On-Premise data into Cloud, or exports Cloud data for you to host (pricing FAQ).

Our read: if you don't already run servers, Cloud is the simpler way to get all of Matomo. On-Premise saves the fee and hands you the upkeep, which we cover below.

We installed each tool alone on a test page and opened it in Chrome four ways: as a person who moves the mouse and scrolls, and as three kinds of bot. One announces it's automated. One hides that. One clicks without moving the mouse, the way AI agents click. No data reached Matomo or Google: apart from fetching each script, our lab answered every request itself.

Notice the weight row. Matomo builds its tracker per install. The one served for Matomo's own Cloud demo bundles the heatmap, media, form, crash and A/B trackers and weighs 66 KB; it also fetched the heatmap plugin's settings. The core matomo.js weighs 22 KB. Matomo itself says about 19 KB by default, and up to about 50 KB with many plugins (Matomo).

Then the cookie row. Matomo left _pk_id and _pk_ses. Matomo documents 13 months and 30 minutes for them, plus a third, _pk_ref, kept 6 months (cookies). GA4 left _ga and _ga_<property>, set for two years and cut to 400 days by Chrome (Google). Nora left nothing, in cookies or in storage.

Now the bot rows. Matomo's script sent a page view for all four visitors, just as GA4's did. Matomo's server then checks each request against a list of over 800 bot user agents and a few search-engine IP ranges (code). That list has no entry for HeadlessChrome (list). The Tracking Spam Prevention plugin blocks headless browsers and cloud IP ranges once you switch it on, though Matomo says it can't spot a headless browser whose user agent has been changed (Matomo).

Ad blockers block both. EasyPrivacy, the list most of them use, blocks matomo.cloud and GA4's tag outright. It also has rules for /matomo.js and /matomo.php on any domain, so a self-hosted Matomo on your own domain gets blocked by file name too (list). Nora isn't on it.

The chart is Matomo Cloud. On your own server, the software itself is free.

Matomo Cloud bills hits: every page view, event, download, outlink, site search and content impression (pricing). So a site with 10,000 page views still pays for the lowest tier, 50,000 hits. Prices follow your location; we were shown dollars at three tiers and euros at all of them. Paying for a year costs ten months.

The 21-day trial includes every feature and doesn't ask for a card. If you don't buy, the site is deleted a few days later (trial). Bots that Matomo recognises aren't recorded unless you ask, and Matomo warns that recording them raises your Cloud hit usage (Matomo).

Matomo Cloud per 1,000 hits

Notice the last column: each 1,000 hits costs less as you grow, from €0.44 at 50,000 hits to €0.16 at 10,000,000.

Matomo On-Premise bundles

Notice the step from Team to Business: €275 to €1,450 a month, for 20 users instead of 4.

That step is where agencies that self-host for clients have pushed back. In August 2026 an agency owner wrote on Matomo's forum that the €1,450 Business bundle prices small clients out. Matomo replied that plugins can still be bought one at a time, and that existing customers keep their old prices (forum).

GA4 is free with no monthly cap. Its paid edition, GA4 360, is sold through Google's sales team, and outside estimates start it near $50,000 a year (Cardinal Path). Nora is free up to 1 million visitors a month. Note that Nora counts visitors, not hits.

The software is free. The server, the upkeep and the time aren't. We haven't benchmarked a self-hosted install, so these are Matomo's own figures and what people who run it report.

Notice the step at 1 million page views. Below it, one server can be enough. Above it, Matomo wants the database on a server of its own.

• Archiving. Matomo turns raw visits into reports in a step it calls archiving. Above a few hundred visits a day, it says to run that from cron every hour; otherwise reports are built when someone opens them, which gets slow. It calls checking the archiving messages, daily or weekly, an administrator's most important job (Matomo).
• Upgrades. Matomo shipped 13 stable releases in 2026 up to 4 October (GitHub). Matomo 6, now in beta, needs at least PHP 8.1 and MySQL 8.0 or MariaDB 10.6 (requirements). After moving from 5.4.0 to 5.8.0, one operator reported visits down about half across 25 sites; Matomo replied it knew of no wider problem (forum).
• Scale. On Hacker News in May 2026, one team said Matomo turned slow at a sustained 5 to 25 tracked logins a second (HN). Another said two self-hosted clients above 8 million visits a month ran well on the right infrastructure (HN). A third moved to Umami in 2025 after dashboards stayed slow despite MySQL tuning and cron archiving (HN).
• WordPress. Matomo for WordPress has more than 100,000 active installs and keeps all its data in your WordPress database. Each page view adds a request to your own server. It needs 128 MB of PHP memory, suggests 256 MB, and builds reports through WP-Cron. For busy sites or large multisites, Matomo recommends On-Premise, or Cloud with the Connect Matomo plugin (plugin page, archiving).

Matomo can run three ways. Here's what each maker documents. It isn't legal advice.

• With cookies, the default. Matomo sets first-party _pk_id, _pk_ses and _pk_ref, and shares none with third parties (Matomo).
• Without cookies. One setting turns them off. Matomo warns that returning visitors become less accurate, and that a few functional cookies can still appear: one for the opt-out form and two for consent (Matomo).
• CNIL mode, for France. Since July 2025 the CNIL asks providers to check their tools against its criteria with a self-assessment table, and says it may still check them itself (CNIL). Matomo says that, on its own assessment, Cloud and On-Premise can run without consent when set up its way, through a setting called Enforce compliance (Matomo).

The CNIL mode is narrow, though. It covers audience measurement only, and leaves out measuring campaigns, acquisition channels, advertising and conversions. It masks two bytes of the IP, keeps cookies to 13 months, drops User ID, and restricts the visits log, heatmaps, recordings, A/B tests and cohorts (Matomo). One person reported that the CNIL told them UTM tracking cancels the exemption (GitHub). So if you want campaign reports and the exemption, you'll have to choose.

GA4 sets _ga and _ga_<property>, and Google's policy requires consent from visitors in the EEA, the UK and Switzerland (Google). With consent mode, GA4 estimates the visitors who decline only once a site has 1,000 declined events a day for seven days and 1,000 consenting users a day (Google). Nora sets no cookies and stores nothing in the browser; we measured that.

Run both on one site and the totals won't match. Four things cause most of the gap.

• Consent. If Matomo runs without cookies, or in CNIL mode, it counts the visitors who decline a banner. GA4 leaves them out of its reports unless your site is big enough for Google to model them.
• Bots. Matomo removes bots that name themselves. GA4 removes known bots by a list it doesn't show, and never says how many (Google). Bots that pose as ordinary Chrome are hard for both. One Matomo owner watched daily visits rise 15-fold, mostly from such bots, and kept deleting them by hand after Tracking Spam Prevention failed to stop them (forum).
• Sampling and thresholds. Matomo says it never samples (Matomo). GA4 samples explorations above 10 million events and hides rows with few users (sampling, thresholds).
• Blocking. EasyPrivacy blocks both, Matomo even on your own domain, so ad blockers cut both counts.

For a sense of scale: on the first site running Nora, 3.3% of the page views its script saw over nine days in October 2026 were bots or AI agents.

Google Analytics 4 drops known bots using Google's research and the IAB spiders and bots list. It always does, and it never says how many (Google). Since 13 May 2026, an AI Assistant channel groups visits referred by ChatGPT, Gemini, Claude and others, going by the referrer only (Google). Nothing in GA4 labels AI agents.

Matomo has added three AI features in a year, and each one reaches a different slice of traffic.

1. AI Assistants channel (5.5.0, 8 October 2025). It groups visits referred by ChatGPT, Copilot, Gemini, Claude, Perplexity and others (changelog), and only for data collected after the update (Matomo).
2. AI Agents (5.6.0, 26 November 2025). It tags visits from ChatGPT agent mode and Amazon Nova Act as AI agents, with ready-made "ai agent" and "human" segments. It's on for new installs. Existing On-Premise installs have to switch it on, which adds a column to the visits table (changelog). In the code, ChatGPT is recognised by the signature headers it sends, which name chatgpt.com, but the code doesn't verify the signature. Nova Act is recognised by its user agent (code).
3. AI Chatbots (5.8.0, 4 March 2026; page reports in 5.12.0). It counts pages that assistants fetch on someone's behalf, by user-agent text: ChatGPT-User, Claude-User, Perplexity-User, MistralAI-User, Gemini-Deep-Research and Google-NotebookLM. The page script can't see these fetches, so you send them from a Cloudflare Worker, CloudFront, WordPress or the HTTP API, and they create no visits (Matomo, 5.12.0). Crawlers that collect training data are left out of the AI reports (Matomo).

People on Matomo's forum are split on them. In a 35-post thread, one called the chatbot reports worthless by their standards, since only well-known crawlers name themselves (forum). In another, people found that referrer-based and agent-based AI segments give very different numbers, because one counts people sent by AI and the other counts machines (forum).

Nora keeps every page view and labels it a person, an AI agent or a bot, with the test that caught it: a signature check, the name it gives, how it behaves, or machine signs. In our lab it caught the headless browser and the agent-style click. The disguised browser that never touched the page passed as a person, as it did everywhere.

Our read: Matomo has done more about AI traffic in a year than GA4 has. All three of its features still go by what a visitor or its referrer says, so an agent in an ordinary browser that says nothing gets past them.

Of the three, only Matomo has a built-in way to take in assistant fetches that never run a script, through its server and edge integrations. Nora needs server or CDN logs for those, and they aren't connected yet.

Where GA4 is ahead

• Google Ads. Conversions for bidding, audiences and spend against results, built in. Matomo's conversion export to ad platforms is a premium feature, in Cloud or the Business bundle.
• Nothing to run. It's free at any traffic, with nothing to install, patch or archive.
• Raw data. A free daily export of raw events to BigQuery, up to 1 million events a day (Google). Matomo Cloud's raw export is a paid add-on; On-Premise gives you the database.
• Demographics from Google signals, though small reports are hidden by thresholds.

Where Matomo is ahead

• Hosting you choose. Frankfurt, or your own servers, with the code open under GPL-3.0.
• A visitor log and visitor profiles, and no sampling.
• More in one tool. Heatmaps, session recordings, form and media analytics and A/B tests, in Cloud or as plugins.
• Your data, for as long as you like, on your own server. GA4 keeps the event-level data behind its explorations 2 or 14 months (Google).
• Free in the core: server-log import and a mode for running without cookies.

Features from each maker's pages on 13 October 2026. "Premium plugin" means included in Matomo Cloud and paid on On-Premise.

We read reviews and threads from the past two years. The same few things come up again and again. Here they are, in our words, with links to where people said them.

1. Choose Cloud or On-Premise first. Everything else depends on it.
2. Run both for two to four weeks. Compare weekday with weekday; consent and bots will move the totals.
3. Import your history with the Google Analytics Importer (guide). It brings aggregated reports only: no visitor log, segments or custom reports on imported data, unique visitors only by day, and goals without their metrics (limitations).
4. Pick your cookie setting: default cookies, cookies off, or CNIL mode if you can give up campaign reports.
5. Turn on Tracking Spam Prevention. On your own server, set up the hourly archiving job too.
6. Keep Google's tag for Ads if you rely on GA4 conversions for bidding.
7. Remove GA4, then check what your cookie banner still has to ask.

Nora is for you if both Matomo and GA4 are more than you use, and you want to know how much of your traffic is people. It's free up to 1 million visitors a month, sets no cookies, weighs 2.7 KB, and labels every page view person, AI agent or bot, with the test that caught it. You also get live visitors, pages, sources (AI assistants among them), countries, goal pages, a weekly email and CSV export.

It gives up most of what makes Matomo Matomo. There are no funnels, heatmaps or recordings, no GA import, no self-hosting and no source code, and your data is in the US. It launches on 1 November 2026, so it has no track record yet. If you need EU or own-server hosting, open source, ISO 27001 certification or GA-level reports, Matomo is the better pick.

Nora Lab, the research team at Nora Analytics, ran every test on this page the same way for each tool. Here is exactly how, so you can judge the results yourself.

The full method, with what each visitor stands for, is on How we test. Every result from the lab run is public as JSON.