Choosing between Google Analytics 4 and Matomo is less a feature-by-feature contest than a decision about where your visitor data lives and who controls it. GA4 is a hosted Google service that costs nothing to run and plugs natively into Google Ads and Search Console. Matomo is open-source software you can either self-host on your own infrastructure or buy as a managed European-hosted subscription. Both will tell you which pages got traffic last week. They diverge sharply on sampling, raw data access, retention and cost structure.
What follows is built on current vendor documentation and court records, because several figures repeated across comparison blogs, particularly on pricing and EU legality, are out of date.
Deployment and ownership: the fundamental split
GA4 collects into Google's infrastructure. You reach that data through the Analytics interface, the Data API or a BigQuery export, but the underlying event store belongs to Google. There is no database you can connect to directly.
Matomo inverts the arrangement. The self-hosted Community edition writes to a MySQL or MariaDB instance you administer, so raw hit-level data is queryable with SQL and governed only by your own retention policy. Matomo's own comparison page leans on this, claiming "100% data ownership", "full access to raw data via SQL" and no sampling. Treat those as vendor marketing, but the architecture behind them is real: self-hosted Matomo puts the event store on hardware you control.
That control carries operational weight. Someone must patch the server, tune MySQL, keep archiving cron jobs running and absorb traffic spikes. Matomo Cloud removes that burden, at the price of handing hosting back to a vendor, albeit an EU-based one.
Sampling, limits and data retention
This is where the free tier of GA4 shows its edges. Google's published limits for standard properties are specific and worth knowing before you commit:
- Data retention is set to either 2 or 14 months, with 14 months the maximum for user- and event-level data.
- Explorations sample at 10 million events per query, and unsampled explorations are not available outside Analytics 360.
- Custom dimensions cap at 50 event-scoped and 10 item-scoped per property, with 25 parameters per event.
- Key events are limited to 30, saved segments and comparisons to 50 each, and calculated metrics to 5.
- BigQuery daily export is capped at 1 million events per day for standard properties, though streaming export is uncapped.
The retention ceiling is the one that surprises people. Standard aggregated reports persist, but the granular data feeding explorations and custom segments ages out at 14 months, which makes multi-year cohort work impossible unless you push events to BigQuery early and keep them there. Analytics 360 lifts these limits substantially, moving daily BigQuery export into the billions of events, but it is enterprise-priced through a sales process.
Matomo's constraints are different in kind. Self-hosted, retention is a configuration setting and sampling is not applied by default, so the practical limit is your database size and query performance rather than a policy ceiling. On Matomo Cloud the limit is commercial: your plan's monthly hit allowance.
What the two actually cost
GA4 is free for standard properties, and for most small and mid-sized sites the free tier covers real analytical needs without hitting the caps above.
Matomo splits into two very different commercial models. According to Matomo's pricing page, Cloud starts at €29 per month excluding tax for 50,000 hits per month across 30 websites, with tiers scaling through 100,000, 300,000, 600,000, 1 million, 2 million and 5 million hits. Annual billing takes two months off. The On-Premise Community edition is free forever with unlimited users and hits, while packaged On-Premise bundles with support begin at €275 per month for the Team tier.
One detail trips up budgeting: Matomo Cloud meters hits, not pageviews. Every tracked event, download, outbound click, site search and ecommerce interaction counts. A site with heavy event tracking can burn through a tier far faster than its pageview count suggests.
The free self-hosted route also has an asterisk. Several features that Matomo Cloud plans include, among them heatmaps, session recording, A/B testing and form analytics, are sold as separate premium plugins for On-Premise users. "Free" describes the core platform, not the full feature set.
Privacy and GDPR: what actually changed
Plenty of comparison articles still assert that Google Analytics is illegal in Europe. That claim reflects a moment that has since passed, and repeating it will mislead your compliance decision.
Between 2021 and 2023, data protection authorities in Austria, France, Italy and elsewhere found that using Google Analytics breached the GDPR, because after the Schrems II judgment there was no valid mechanism for transferring EU visitors' data to Google in the United States. The Austrian DSB decision landed in January 2022 and the French CNIL followed in February 2022.
The European Commission then adopted its adequacy decision for the EU-US Data Privacy Framework on 10 July 2023, and Google LLC has been certified under it since August 2023. That restored a lawful transfer route. The framework survived its first judicial test when the EU General Court dismissed Latombe v Commission (T-553/23) on 3 September 2025, rejecting challenges to the Data Protection Review Court's independence and to US bulk-collection limits. An appeal to the Court of Justice is pending.
So the honest position is this: transatlantic transfers under the DPF are currently lawful, but the framework is the third of its kind after Safe Harbour and Privacy Shield, both of which were struck down. If your risk tolerance for a third invalidation is low, or you operate in a regulated sector with data residency obligations, self-hosted Matomo removes the transfer question rather than betting on the framework holding. Matomo also uses a config_id approach rather than third-party cookies, and a suitably configured deployment has been recognised by the CNIL as exempt from tracking consent, which materially improves data completeness where consent banners suppress collection.
Which one fits your situation
GA4 is the stronger choice when
- You run paid search or shopping campaigns and need native Google Ads and Search Console integration for attribution.
- You want a managed service with no server administration and no licence spend.
- Your analysis horizon sits comfortably inside 14 months, or you are willing to run a BigQuery export.
- Your team already knows the interface, and retraining cost matters.
Matomo is the stronger choice when
- Data residency or sector regulation requires visitor data to stay on infrastructure you control.
- You need unsampled, hit-level data over multi-year windows for cohort or attribution modelling.
- You want direct SQL access to raw data rather than working through an API or export pipeline.
- Consent-banner drop-off is materially degrading your measurement and a consent-exempt configuration would recover it.
For a marketing team optimising Google ad spend on a mid-sized site, GA4's free tier and ecosystem integration are difficult to justify replacing. For a healthcare provider, public-sector body or European firm treating US transfer risk as a live concern, self-hosted Matomo answers a question GA4 structurally cannot. Model the hit volume before you price Matomo Cloud, and check the premium plugin list before you assume self-hosting is free.