Google Analytics alternatives: how to choose the right analytics for your website

You added Google Analytics to your website because it's what everyone does. It's run quietly in the background ever since. Maybe you check it. Or maybe your dashboard is gathering dust because it's complicated, confusing, and wasn't set up for you from the start.

So you started looking at Google Analytics alternatives, and found long lists of tools, some of them written by the companies selling those tools.

The thing about choosing analytics is that it's not a contest between dashboards. You need to make three decisions:

1. what you need to know about your visitors,
2. who holds that data, and
3. what you're asking those visitors to agree to.

When you answer those three, the shortlist gets very short.

What should you use instead of Google Analytics?

If you want to know how many people visit your site, where they came from, which pages they read, and whether they got in touch, a lightweight cookieless analytics tool covers it. Tools like Plausible, Umami, Fathom or Simple Analytics, or the analytics built into your host if you're on Vercel or Cloudflare. If you run Google Ads, rely on audiences or build explorations, stay on Google Analytics 4 (GA4): none of the GA4 alternatives here replaces those. If you need every raw row of data and someone will run a server, you can self-host Matomo, Umami or Plausible's Community Edition.

That's the short version. The rest is the reasoning, so you can check it against your own site and choose what works best for you.

What Google Analytics gives you, and what it costs you

GA4 is free, and it's the most complete tool here: events, conversions, explorations, audiences, and a direct line into Google's advertising products. If you use those, nothing below replaces them. But there is a cost that shows up in four other places.

It sets cookies. GA4's _ga cookie lasts two years by default and is used "to distinguish users". In plain terms, it's a small file that lets Google recognise the same browser when it comes back. In the EU, that means you need to ask for consent before it runs. That's why cookie banners exist. Anyone who clicks "reject" isn't counted the usual way, so your numbers can come out lower than your real traffic.

Detailed data expires. On a standard property, the event data behind explorations and funnel reports is kept for 2 or 14 months, depending on a setting, and then deleted. The headline reports, like visits and top pages, aren't affected. If you're like me, you assumed that once the GA4 tag was on your site, everything was being saved until you got round to looking at it. It isn't. If you want to see how people moved through your site this time last year, to compare a busy season, that detail may already be gone, and you can't get it back. So check the setting now: in Admin, go to Data Settings, then Data Retention, and choose 14 months. The change only saves data that hasn't already been deleted.

Some reports are trimmed. Google hides rows where a report could reveal an individual, which shows up as missing rows in reports on age, gender and interests. If you're trying to understand who your visitors are, those are the reports you'd turn to, and they're the ones with holes in them. Sampling, where GA4 works from a portion of your data and estimates the rest, starts above 10 million events per query on a standard property. That's far beyond a small business site, so you can set it aside.

Consent changes what it sees. Consent Mode tells Google whether each visitor agreed to cookies. In its advanced setting, visitors who decline still send cookieless pings, and Google uses them to estimate what it couldn't measure, which it calls modelling. So part of what you see is Google's estimate, not a count. That's fine for spotting trends. It's less useful when you need an exact number, such as how many people filled in your contact form.

If your site was set up before 2023, check which version it has. Look at the ID pasted into your website builder's settings or your site's code. A GA4 ID starts with G-. One that starts with UA- is Universal Analytics, the old version, which stopped processing data on 1 July 2023. Google then closed access to it from the week of 1 July 2024 and said its data would be deleted. So if UA- is the only ID on your site, nothing has been recorded since mid-2023, and unless you exported it, the history before that is gone too. To find your GA4 ID, go to Admin, then Data streams, and click your website's stream. GA4 is what Google Analytics means now.

The Google Analytics alternatives, compared

All seven of these Google Analytics alternatives count visits, sources and pages. They differ on price, cookies, where the data lives and whether you can take it with you. Prices and limits are each vendor's own, checked on 4 November 2026.

Plausible is a single-page dashboard. Funnels start on its Business plan, and the free Community Edition leaves out funnels, ecommerce revenue goals, SSO and the sites API.

Umami has a free cloud plan up to 100,000 events a month and is MIT-licensed, so self-hosting costs nothing but the server: Node.js and a PostgreSQL database. Its plans include funnels, goals and journeys.

Fathom is hosted only, covers up to 50 sites on its entry plan, and keeps your full history while you subscribe.

Simple Analytics drops the IP address from every request. Its free plan deletes anything older than 30 days.

Matomo comes closest to GA4 in depth, and it's the only tool here that sets cookies by default. Self-hosted, every row of data sits on your own server.

Vercel Web Analytics is already there if your site runs on Vercel. On the free Hobby plan, reports cover one month and collection pauses once you pass the limit.

Cloudflare Web Analytics is free and uses no cookies or local storage. It keeps unsampled data for 7 days, then aggregates it to around 10%.

Microsoft Clarity is Microsoft's free tool in this space, but it does a different job: free session recordings and heatmaps, using first-party cookies, with recordings kept for 30 days. It sits beside an analytics tool rather than replacing one.

Do cookieless analytics tools need a cookie banner?

Sometimes not, but removing the cookie doesn't settle it on its own. It depends on where your visitors are, what the tool does on their device, and what happens to the data afterwards.

In the EU, the law behind cookie banners is the ePrivacy Directive. It says you need a visitor's consent before you save anything on their phone or computer, or read anything already saved there: "the storing of information, or the gaining of access to information already stored". There's no exception for analytics. And it isn't only about cookies. Europe's data protection regulators say it also covers tracking pixels, local storage (another way a site saves data in your browser), tracking by IP address, and any code that singles out one visitor from another. So a tool can drop the cookie and still need consent, if it does one of those things instead.

France has its own exception. Its regulator, the CNIL, lets you count visits without consent if the statistics are only for you, they're anonymous, and the data isn't combined with anything else or passed on. That only applies in France, not across the EU.

In the UK, the rules changed on 5 February 2026. You can now run analytics without asking first, as long as all four of these are true:

• you use it only to see how your site is used, so you can improve it
• you share the data only with whoever helps you make those improvements
• you tell visitors clearly that it's running
• you give them a simple, free way to say no

That last point is the catch. Visitors don't have to opt in, but they must be able to opt out, so the notice and the way to object still have to be there. What changes is that you don't have to ask before the analytics runs.

In the US, there's no national law requiring cookie consent. California's privacy law works the other way round: visitors don't opt in, they can opt out. It only applies to businesses above its thresholds, and earning more than $25 million a year is one of them. The opt-out covers selling or sharing people's data for advertising that follows them across other sites and apps. So for a US site, the question isn't whether your analytics uses a cookie. It's whether the data is sold or passed on for that kind of advertising.

The vendors don't agree on how far their own tools go. Plausible says you "do not need cookie banners for analytics". Fathom says "Cookie-free does not automatically mean outside device-access rules". Umami says whether you need a notice "depends on the other services you use and applicable requirements". Matomo recommends treating its JavaScript tracking as needing consent. If the vendors can't agree, it's no surprise the rest of us find this confusing.

So with a cookieless tool used only to understand your own site, you may not need a banner for analytics. Whether you do depends on your visitors' countries, the tool's settings and everything else on the page. A booking widget or an ad pixel can bring the banner straight back.

Self-hosted or cloud: who owns the data

Every vendor here says the data is yours. What that means in practice is whether you can take it with you.

Export is the test. Plausible's cloud export is a set of CSVs holding the top 300 rows of each report; raw events are on its Enterprise plan. Fathom exports aggregated CSVs. Umami Cloud exports everything, and Simple Analytics offers raw data points through an API. Self-hosted Matomo, Umami or Plausible keep the whole database on your server.

Self-hosting moves the work to you. Each of the three is an open source Google Analytics alternative: Umami is MIT-licensed, Matomo is GPLv3 and Plausible's Community Edition is AGPLv3, so all of them are free to run. What you pay instead is a server, upgrades, backups and security patches, for as long as the site exists. If you already run a server for something else, such as automation workflows or a database, adding one of these may be a small step. If nobody on your team will own that work, use the hosted version of the same tool and keep the export.

Why I'm running Umami alongside Google Analytics on this site

This site has run Umami Cloud alongside GA4 since 5 October 2026, while GA4 is phased out. I'm testing both together to make a comparison before I fully choose to move over to Umami. There's nothing to lose by doing so. Running both for a while is the safest way to switch: you see whether the new numbers make sense against the old ones before you rely on them, and you keep the GA4 history until you've decided what to export.

Expect the two to disagree. GA4 depends on a cookie some visitors decline or delete, and a cookieless tool counts visitors a different way. Any script can be blocked: Plausible estimates that strict blockers miss between 5% and 25% of visitors to its own script, depending on the audience. Compare trends, not exact counts.

Both scripts on this site load after the page has finished loading, so neither delays what a visitor sees first.

Whatever you choose, keep Search Console

None of these tools shows what happens on Google before someone clicks. Google Search Console does: which searches showed your pages, how often, and whether Google could index them. It's free, it's separate from Google Analytics, and it keeps working if you remove GA4 entirely.

Leaving Google Analytics doesn't mean losing Google's search data. Keep Search Console connected, and let your analytics tool handle what happens after the click. My article, What Search Console and Analytics each show, covers the split in more detail.

How to choose

Stay on GA4 if you run Google Ads and use Analytics for conversions or audiences (or if you think you'll use them in the near future), if you build explorations, or if someone on your team reads it every month. Set up consent properly and keep it.

Switch to a hosted cookieless tool if what you need is visits, sources, top pages and enquiries. Choose on price and export: Umami for a free plan with room to grow, Plausible or Fathom for the plainest paid dashboards, Simple Analytics if dropping the IP address matters most to you. If you're already on Vercel or Cloudflare, try the built-in option first.

Self-host Matomo, Umami or Plausible only if you need every raw row and someone will maintain the server.

Whichever you pick, run it beside the old tool for a few weeks, and keep Search Console.

The question was never which analytics tool is best. It's what you need to know, and who should hold it.