GA4 vs Matomo 2026: Free at Scale vs Data You Control

The short answer: GA4 is free at effectively any traffic volume and connects to Google Ads in a way nothing else can, because it is the same company on both ends of the link. That is a large, real advantage and it is why most sites should stay. Matomo wins on the things GA4 caps: no sampling, no data thresholding, report data kept forever against GA4's 14-month standard retention limit, and the raw database when self-hosted. Matomo Cloud starts at USD 26 per month for 50,000 hits; On-Premise Community is free forever. Choose GA4 if Google Ads bidding is load-bearing. Choose Matomo if you keep hitting a wall where the number you need is sampled, thresholded, expired or not exportable.

The standard version of this comparison opens by calling GA4 confusing and works backwards from there. That is a weak argument. "Confusing" is a training problem, not a capability problem, and it dodges the two facts that decide the question: GA4 is free at a scale nobody else offers for free, and it feeds Google Ads bidding directly.

Every limit cited below comes from Google's or Matomo's own documentation, read in August 2026, with the source next to it.

At a glance

Sources: GA4 360 feature limits, GA4 data sampling, GA4 data thresholds, matomo.org/pricing. All read 17 August 2026.

Quick verdict

• Cost at scale: GA4, decisively. Nothing else is free at 50 million events.
• Google Ads performance: GA4, and no third-party tool can close this gap.
• Getting a complete, unsampled number: Matomo.
• Seeing demographic and search-query rows at all: Matomo.
• Reporting beyond 14 months: Matomo.
• Behaviour tooling in the box, and experimentation: Matomo.
• Warehouse-scale raw analysis with a data team: GA4, via BigQuery.

Cost: GA4 is free at a scale that has no competitor

Standard GA4 properties are free. No traffic tier, no pageview meter, no upgrade prompt at volume. A site doing 50 million events per month pays exactly what a site doing 50,000 pays.

Matomo Cloud's Business plan starts at USD 26 per month excluding tax for 50,000 hits per month, or USD 260 per year, with additional hits at USD 2.60 per 5,000. The traffic selector runs to 10 million hits, above which pricing is by quote. Matomo defines a hit as "a tracked page view, an event tracking, a download, an outlink, an onsite search or a content tracking request," so an event-heavy implementation consumes the meter far faster than its pageview count implies (matomo.org/pricing, verified August 2026).

Matomo On-Premise Community is free forever with unlimited users and unlimited hits. A genuine free option, but the premium capabilities most teams evaluate Matomo for are sold as plugin bundles: Team at 275 EUR per month (up to 4 users, up to 5M hits), Business at 1,450 EUR per month (up to 20 users, up to 30M hits), Enterprise at 3,400 EUR per month (up to 50 users, up to 100M hits).

Google Analytics 360 exists for organisations needing higher limits, but Google does not publish a price for it on its help pages; it is contracted through Google Marketing Platform or a Sales Partner. We could not verify a 360 figure from a Google-owned source, so we are not quoting one.

The fair counter-argument to "GA4 is free" is that analytics forcing a data team to rebuild reports in BigQuery is not free. True, but that cost only lands on organisations large enough to hit GA4's ceilings.

Winner: GA4. Self-hosting Matomo to match it means paying in servers, upgrades and ops time instead.

Google Ads integration: the advantage nobody can copy

This is the dimension most comparisons skip because it does not flatter the alternative.

Linking a GA4 property to Google Ads lets you turn Analytics key events into Google Ads conversions used for bidding, and push Analytics audiences into Ads for remarketing. You can create up to 400 links per property, and Google Ads data appears in Analytics reports within 48 hours. Google states that "customers who link their Google Ads or Google Marketing Platform accounts to a Google Analytics property are correlated with a 23% increase in conversions and a 10% reduction in cost per conversion" (Connect Google Ads to Google Analytics, read August 2026). That figure is Google's own and correlational, so treat it as directional. The mechanism underneath it is not in dispute.

Matomo's answer is Advertising Conversion Export, described by Matomo as an export of attributed goal conversions for use in ad networks like Google Ads, "so you no longer need a conversion pixel," alongside Multi-Channel Conversion Attribution. Both are included on Matomo Cloud plans (matomo.org/pricing).

That is a real feature. But an export is a different object from a live bidding signal. GA4 and Google Ads are the same company's systems talking continuously; Matomo is handing a file over a fence.

One related GA4 capability cuts both ways. Under consent mode, when a visitor denies consent GA4 sends cookieless pings and fills the gap with behavioural and conversion modeling, which recovers reporting other tools lose outright (Consent mode). It also means part of what GA4 shows is estimated rather than observed and is not labelled as such, and behavioural modeling requires the advanced consent mode implementation.

Winner: GA4, by a margin no competitor can close. If paid search is how your business makes money, this section alone may end the evaluation.

Sampling: where GA4 stops giving you the whole number

Google publishes the thresholds plainly, and they are tighter than most people assume.

The quota limit for event-level queries is 10 million events per query for standard GA4 properties, and up to 1 billion for Analytics 360 properties, which default to 100 million. Google also warns that "filtering large datasets by country may result in data sampling in standard Google Analytics reports and Explore, even when using expanded data requests," because filtering activates different processing above roughly 100 million events (About data sampling, read August 2026).

Two follow-on facts matter more than the headline number.

Unsampled explorations are not available on standard properties at all. 360 properties get 20,000 daily unsampled data tokens, maximum 5,000 per query. On a standard property, once you are over the line, no purchase or setting returns an exact figure in the interface (360 feature limits).

Even "unsampled" GA4 reports are approximations. Google uses HyperLogLog++ to estimate distinct counts for metrics like users and sessions, and states discrepancy rates are usually under 1% but "can be higher, especially when combining more than one HLL++ metric." Matomo's Cloud pricing table, by contrast, carries a flat "No data sampling" line on every plan.

Winner: Matomo. If your analysis routinely crosses 10 million events per query, GA4 standard will hand you an estimate and Matomo will hand you a count.

Data thresholding: the part the old blog posts get wrong

This is worth reading carefully, because the widely repeated advice is out of date.

The folklore says GA4 thresholding is caused by Google Signals and can be switched off by changing your reporting identity. Google's current documentation says something narrower: "Data thresholds are system defined. You can't adjust them." They apply to reports, explorations and API calls that include demographic data or audiences defined using demographic data, and to rows containing search query information where there are not enough total users (About data thresholds, read August 2026).

The only mitigations Google lists are widening the date range so more users fall inside it, and exporting to BigQuery, with the caveat that Analytics does not export Google Signals data to BigQuery, so event counts per user can differ between the two.

The practical effect is that low-traffic properties can find whole rows withheld with no way to force them open. That is a permanent constraint, not a configuration mistake, and Matomo has no equivalent suppression layer.

Winner: Matomo.

Data retention: 14 months against forever

GA4 standard properties retain event data for up to 14 months, with only two options, 2 or 14 months. Large and XL properties are limited to 2 months. Analytics 360 goes up to 50 months, with options at 2, 14, 26, 38 and 50 (360 feature limits).

Fourteen months covers a year-over-year comparison and not much more, and the "large properties get 2 months" clause catches high-traffic sites precisely when history matters most. Matomo Cloud retains raw data for 24 months and report data forever.

The standard workaround is BigQuery export, which is genuinely good and free to configure. Note the standard-property limit though: daily export is capped at 1 million events, while streaming export is unlimited and 360 daily export runs to billions.

Winner: Matomo.

Reporting depth and what ships in the box

On a standard GA4 property you get 30 key events, 100 audiences, 50 event-scoped custom dimensions, 25 event parameters per event, 5 calculated metrics and an API quota of 200,000 tokens per day. Generous for most sites, cramped for large ones; 360 roughly quadruples most of them.

Matomo Cloud's entry plan gives you 150 goals, 100 segments, 30 custom dimensions and 100 custom reports, plus things GA4 does not have at any price: heatmaps and session recording, form analytics, media analytics, cohorts, users flow, roll-up reporting, a tag manager, crash analytics, SAML and LDAP.

The honest split is by team shape, not feature count. GA4 plus BigQuery, in the hands of someone who writes SQL, out-analyses Matomo. Matomo without a data team out-reports GA4, because the reports a marketer wants are already built and nothing expires at 14 months.

Winner: GA4 if you have a data team and BigQuery. Matomo if you do not.

Experimentation: GA4 has none

Google Optimize and Optimize 360 were shut down on 30 September 2023, and any experiments still running that day ended. Google's stated reason is that Optimize "did not have many of the features and services that our customers request," and that it would instead invest in third-party integrations, naming AB Tasty, Optimizely and VWO (Google Optimize sunset). So GA4 in 2026 measures experiments run elsewhere. It does not run them.

Matomo does. A/B Testing is included on the Cloud-hosted Business plan, or sold for self-hosted installs at 269 USD per year for up to 4 users with unlimited websites, with a 30-day trial (Matomo A/B testing FAQ, plugins.matomo.org/AbTesting). Success metrics include goal conversions, goal revenue, ecommerce orders, ecommerce order revenue, total conversions, total revenue, pageviews, bounces and visit length, and it supports unlimited experiments, per-variation traffic allocation, a configurable confidence threshold and redirect-based tests for entirely different layouts.

Winner: Matomo, uncontested.

Migration cost

Staying on GA4 costs nothing, which is the baseline any migration has to beat.

Matomo publishes a native Google Analytics importer with a Quick Connect setup for Cloud. Read the two pages Matomo publishes next to it before committing: "Unsupported GA Dimensions" and "Limitations when importing Google Analytics data" (Import Google Analytics). The import works and it is not lossless.

The larger cost is not the data, it is the Google Ads link. Unlinking stops key events flowing to Ads as conversions, stops audiences accumulating for remarketing, and turns new Google Ads dimensions into "(not set)" in Analytics, though historical dimensions collected before the unlink remain visible (Connect Google Ads to Google Analytics).

Winner: GA4, because the incumbent is free and the switching cost is concentrated in the one integration you cannot replace.

GA4 advantages

• Free at any traffic volume. No pageview meter, no event tier, no upgrade wall.
• Native Google Ads integration: key events become bidding conversions, audiences become remarketing lists, up to 400 links per property, data flowing within 48 hours.
• Behavioural and conversion modeling under advanced consent mode, recovering reporting from visitors who decline consent.
• Free BigQuery export, with unlimited streaming export even on standard properties.
• Integrations across Search Console, Looker Studio and Merchant Center that no independent vendor can match.
• Zero infrastructure, and every agency and contractor already knows the tool.

Matomo advantages

• No data sampling on any Cloud plan, against GA4's 10 million events per query.
• No thresholding. GA4 withholds demographic and search-query rows by a system rule you cannot change.
• Report data retained forever, against GA4 standard's 14-month cap and 2 months for large properties.
• Heatmaps, session recording and form analytics in the same product, which GA4 does not offer at any price.
• A/B testing included on Cloud Business, with revenue-based success metrics.
• Direct MySQL access when self-hosted, and a Community edition that is free at unlimited users and hits.
• 30 sites and 30 team members on the entry Cloud plan, plus data in Frankfurt and ISO 27001:2022 certification.

Choose GA4 if you...

• Run Google Ads and use Analytics key events for bidding or audiences for remarketing.
• Have high traffic and no budget line for analytics.
• Have a data team, or will have one, and want raw events in BigQuery.
• Need Search Console, Looker Studio and Merchant Center to hang together without integration work.
• Stay comfortably under 10 million events per query and 14 months of history.

Choose Matomo if you...

• Keep hitting sampled numbers in Explore and need exact counts.
• Have watched demographic or search-query rows get withheld and cannot get them back.
• Need more than 14 months of history without standing up a warehouse.
• Want heatmaps, session recording or form analytics in the same tool as your analytics.
• Want to run A/B tests without adding a separate experimentation vendor.
• Have a regulatory requirement to control where the database lives, or run several sites you want on one plan.

Conclusion

Do not switch off GA4 because it is confusing. Switch, or add Matomo alongside it, when you can name the specific number GA4 will not give you.

That number is usually one of four: a figure that came back sampled because the query crossed 10 million events, a demographic or search-query row withheld by a threshold you cannot adjust, a comparison you cannot run because standard retention stopped at 14 months, or a report that does not exist because GA4 has no heatmaps and no experimentation. Matomo answers all four. If none of them describe your week, GA4 is free, it is the best-integrated tool in paid search by a distance no competitor can close, and staying is right.

For teams that need both, the strongest setup is not a migration at all: keep GA4 linked to Ads for bidding, and run Matomo as the reporting source of truth.

There is a fifth kind of missing number, and neither tool is built for it. Disclosure: we build Humblytics, which pairs analytics with A/B testing in one script and resolves the test verdict against Stripe revenue rather than a modeled key event, so it is worth a look only if your job is changing the page and proving the change earned money.

Frequently asked questions

Is Matomo really free?

Matomo On-Premise Community is free forever with unlimited users and unlimited hits. The premium capabilities most teams evaluate Matomo for, including funnels, heatmaps, session recording and A/B testing, are sold separately as plugins or as bundles starting at 275 EUR per month for On-Premise. Matomo Cloud, which includes them hosted, starts at USD 26 per month for 50,000 hits, verified August 2026 at matomo.org/pricing.

At what point does GA4 start sampling my data?

The quota limit for event-level queries is 10 million events per query on standard GA4 properties, and up to 1 billion on Analytics 360 properties, which default to 100 million. Google also warns that filtering large datasets by country can trigger sampling in standard reports and Explore above roughly 100 million events. Unsampled explorations are not available on standard properties at any price. Source: About data sampling.

Can I turn off GA4 data thresholding?

No. Google's documentation states that data thresholds are system defined and cannot be adjusted. They apply to reports, explorations and API calls involving demographic data or audiences built from it, and to search-query rows with too few users. The only mitigations listed are widening the date range and exporting to BigQuery, and Google Signals data is not included in that export. Source: About data thresholds.

Does GA4 have built-in A/B testing?

No. Google Optimize and Optimize 360 were shut down on 30 September 2023 and GA4 has no native replacement. Google now points users to third-party testing tools and has published APIs so any of them can integrate. Matomo does include A/B testing on its Cloud Business plan, or as a 269 USD per year plugin for self-hosted installs.

Can Matomo replace the Google Ads integration?

Not fully, and this is the most important thing to understand before migrating. Matomo offers Advertising Conversion Export, which exports attributed goal conversions for use in ad networks like Google Ads without a conversion pixel, plus multi-channel conversion attribution. What it cannot replicate is a live, continuous link into Google's own bidding system. If Google Ads bidding drives your revenue, run both tools rather than switching.