Late last year, a friend who runs an indie blog vented to me about something. His site pulls maybe twenty or thirty thousand visits a month, and he had Google Analytics 4 installed. One afternoon he opened the dashboard just to see how much traffic last week’s post had actually generated. Twenty minutes later he was still lost somewhere inside GA4, unable to find a single clean number for page views. The reports nested inside reports, dimensions crossed other dimensions, and the exploration feature wanted him to drag and drop his own configuration. He said something that stuck with me: “I just wanted to know how many people read my article. GA4 made me feel like I was studying for a CPA exam.”
His frustration is not unusual. Ever since Google forced everyone onto GA4 in 2023, the same complaint keeps surfacing in indie developer and small-publisher circles. GA4 was built for large product teams and ad-spend optimization. Its event-driven data model is flexible, sure, but for someone who just wants to know “how many people showed up today, where they came from, and what they read,” it feels like using a scalpel to peel an apple.
The bigger headache is the compliance wall.
That Cookie Banner You Can’t Avoid
If your site gets European visitors, even one, GDPR concerns you. GA4 uses cookies to identify and track users, which means you have to plant a cookie consent banner on your site. Every visitor gets hit with a popup in their first second on the page: “This site uses cookies. Do you agree?”
The chain reaction from that popup is worse than most people assume. Based on industry experience, somewhere between 30% and 50% of visitors will flat-out decline cookies or just close the popup. For an analytics tool, that means a big chunk of your data is missing by design. The people who declined are simply invisible to you. You think last week brought 500 visitors, when the real number might have been 800.
The subtler cost is the hit to user experience. Someone visiting your site for the first time, before they’ve seen a shred of content, gets blocked by a popup that reads like a legal notice. That is not a warm first impression.
So more and more site owners have started hunting for a different option. Is there an analytics tool that doesn’t use cookies at all, needs no banner, is privacy-friendly by design, and still tells me the numbers that actually matter?
There is. And more than one.
Three Different Roads
The three privacy-first analytics tools drawing the most attention right now are Plausible, Umami, and Fathom Analytics. They share a lot of DNA: none use cookies, none collect personally identifying information, none require a consent banner, and all of them are dramatically lighter than GA4. But they travel three very different roads, with different philosophies and trade-offs underneath.
Understanding those differences is what lets you pick the one that actually fits you.
Let’s start with an open-source project out of Europe.
The Lightweight Philosophy from Estonia
Plausible Analytics was born in Estonia, an open-source project written in Elixir under the AGPLv3 license. From the start, the founding team wanted to build the opposite of GA: no tracking of individuals, no cookies, just the core metrics.
What impresses people most about Plausible is how light it is. Its tracking script is under 1KB. For comparison, GA4’s script weighs about 135KB gzipped, dozens of times heavier. Why does that gap matter? If you care about page load speed, and you should, because Core Web Vitals feed directly into SEO ranking, Plausible adds almost nothing to your pages.
Open the Plausible dashboard and you get a single-page view: visitors, page views, bounce rate, average visit duration, traffic sources, top pages, device types, geographic breakdown. No submenus, no exploration reports to configure yourself, all the key numbers in plain sight. For someone like my friend, this is exactly what he’d been after.
Plausible offers both cloud hosting and self-hosting. The cloud plan bills monthly, with the Starter tier around $9 a month, priced in tiers by page views, and annual billing knocks off roughly 17%. Data is stored in the EU, a plus for anyone who has to satisfy GDPR. If you have the technical chops, you can deploy it on your own server for free, since the source is right there on GitHub.
Who does Plausible fit best? People who want an analytics tool they can install and forget. You don’t configure a complex event model, you don’t dig through a pile of dimensions, and you don’t sweat compliance. It just sits there quietly, handing you the numbers you asked for.
A Free Garden for Technical People
Umami takes a different shape. It’s a fully open-source project under the MIT license, one of the most permissive open-source licenses there is, which means you can do nearly anything with it. Built on Next.js with PostgreSQL or MySQL on the backend, its stack feels right at home for front-end developers.
Umami has more than 38,000 stars on GitHub and an active community. The crowd it draws is the “I’d rather do it myself” technical user. If you’re the kind of person who sees an open-source project and immediately wants to fork it and poke around, Umami will feel like your natural habitat.
Self-hosting Umami is completely free, with no feature caps, no data sampling, no retention limits. Your data lives in your own database and you hold absolute control. That’s Umami’s hardest-edged selling point: the data is entirely yours and never passes through any third-party server.
If you’d rather not deal with server maintenance, Umami also offers a managed service, Umami Cloud. It has a free Hobby plan: 100,000 events a month, one website, six months of data retention. For a personal blog or a small project, that free tier is basically enough. Need more capacity and the Pro plan runs $20 a month, covering a million events, up to 20 sites, and two years of retention.
One thing to note: some of Umami Cloud’s advanced features, like session replay and heatmaps, live only in the higher Business plan at $200 a month, and the self-hosted version doesn’t include them either. If all you need is basic traffic stats and event tracking, that’s a non-issue.
Who does Umami fit best? Technical indie developers and small teams. You have a VPS, you know Docker, and you enjoy keeping everything under your own roof. Umami gives you the most freedom at the lowest cost.
The Canadian Managed Philosophy
Fathom Analytics walks a third road. It isn’t open source. It’s a pure commercial SaaS product from a Canadian company. That makes it something of a minority in the privacy-analytics world. Plausible and Umami both treat open source as a core part of their identity, while Fathom chose a closed-source but highly polished commercial path.
Fathom’s central pitch is that you shouldn’t spend any energy operating your analytics tool. No servers to manage, no databases to back up, no versions to upgrade. You add one line of script to your site and go do the work that actually matters.
On that “zero ops” promise, Fathom delivers. The dashboard is clean, it loads fast, and the feature set is enough without being bloated. Event tracking, UTM parsing, email reports, public shareable dashboards, all the essentials are there. One standout is permanent data retention: from your very first day, all your historical data stays put and never expires.
On pricing, Fathom’s entry plan is $15 a month for up to 100,000 page views. That’s higher than Plausible’s $9, but every Fathom plan includes unlimited sites and unlimited team members. If you run several sites, that pricing structure is actually a good deal. Annual billing brings it to about $12.5 a month. Higher-traffic sites bill in tiers, so 500,000 views a month runs $45, and a million runs $60.
Fathom hosts data in Canada and the EU and offers an EU data isolation option, which satisfies users with strict data-residency requirements.
Worth mentioning: Fathom used to offer a custom-domain feature to help bypass ad blockers, but later removed it. In their official explanation, they said ad blockers can shut down a custom-domain approach with a single line of code, while Fathom would have to burn heavy development time responding again and again. The math didn’t add up. They’re exploring new alternatives, but for now the feature is gone.
Who does Fathom fit best? People who don’t want to touch technical details and just want to pay for peace of mind. Founders, content creators, marketers, non-technical site owners, anyone who feels “install and forget” isn’t enough and would rather have it installed for them too.
One Table to See the Core Differences
Now that you’ve met all three tools, you probably have a rough leaning. The comparison below lays the key decision factors side by side to help you make the final call. These particular factors made the cut because they directly shape your experience, your long-term cost, and your grip on your own data, the three things that really matter when picking an analytics tool.
| Factor | Plausible | Umami | Fathom |
|---|---|---|---|
| Starting price | ~$9/mo (cloud) | Free (self-host) / $0 (Cloud Hobby) / $20/mo (Cloud Pro) | $15/mo |
| Deployment | Cloud + self-host | Cloud + self-host | Cloud only |
| Open-source license | AGPLv3 | MIT | Closed |
| Data ownership | Cloud: Plausible-hosted (EU); self-host: fully yours | Cloud: Umami-hosted (US/EU); self-host: fully yours | Fathom-hosted (Canada/EU) |
| Cookie dependency | None | None | None |
| Self-host difficulty | Moderate (Elixir + ClickHouse/PostgreSQL, Docker support) | Low (Next.js + PostgreSQL/MySQL, one-command Docker) | Not supported |
| Tracking script size | Under 1KB | About 2KB | Lightweight (undisclosed, far smaller than GA4) |
| Data retention | Cloud: 3 years+ | Cloud Hobby: 6 months; Cloud Pro: 2 years; self-host: unlimited | Permanent |
| Site limits | Starter: 1; Growth: 3; Business: 10 | Cloud Hobby: 1; Cloud Pro: 20; self-host: unlimited | All plans: unlimited |
| Ecosystem | WordPress plugin, Carrd, Ghost, Webflow; API available | WordPress plugin, framework integrations; API available | WordPress, Ghost, Squarespace; API available |
A few points in that table deserve a closer look.
First, the license difference. Plausible uses AGPLv3, which requires that if you modify the code and offer it as a network service, you also open-source your modifications. For ordinary self-hosters this changes nothing about daily use, but if you want to build on Plausible and resell it, you’re bound by that clause. Umami’s MIT license carries almost no such strings. You can do anything with it: commercial use, modification, closed-source redistribution.
Second, data retention. Fathom’s permanent-retention policy is a real differentiator. The cloud versions of Plausible and Umami both have retention windows. You can sidestep those limits by self-hosting, but then you’re on the hook for the long-term storage and upkeep of the database yourself.
Finally, site count. If you’re a freelancer or a small studio juggling five or six different sites at once, Fathom’s “unlimited sites on every plan” is a compelling condition. The low-priced tiers from Plausible and Umami cap site count, and you’d need to move to a higher plan to unlock more.
Three People, Three Ways to Choose
At this point you can probably decide. But to sharpen it, here’s how I’d map three typical scenarios.
If you’re a budget-conscious indie developer with a project or two, maybe an open-source site, and you’re comfortable managing a server, go with self-hosted Umami. Cost is close to zero, only the server itself, the data stays entirely in your hands, and the MIT license gives you maximum flexibility. If you’d rather not manage a server but money is tight, Umami Cloud’s free Hobby plan will get you going too.
If your attitude is “I paid, now don’t make me touch any technical problem,” and you run a handful of sites, and you don’t need self-hosting or care about open source, go with Fathom. Fifteen dollars a month buys zero ops, permanent data retention, and unlimited sites. It’s probably the least mentally taxing of the three to live with.
If you want a balance point, open-source auditability and transparency on one side, a clean and pleasant managed experience on the other, plus EU data storage, go with Plausible. Its $9 starting price is the lowest of the three cloud options at the same traffic tier, and both the simplicity of its dashboard and the thoroughness of its privacy compliance are top-tier.
These scenarios aren’t strictly either-or, of course. Some people cut their teeth on self-hosted Umami, decide server maintenance is too much of a chore, and migrate to Plausible’s cloud. Others start on Fathom, then as the company grows and gains an ops team, switch to self-hosted Umami for deeper control of their data. Tools are means, not faith.
Choosing a Tool Is Choosing a Stance
Back to my friend’s story. He later tried Plausible’s 30-day free trial and deleted the GA4 script within a week. He told me, “I finally know how many people actually read a post, instead of being trapped in a pile of dimensions I can’t parse.”
The shared value of these three tools isn’t just “no cookies” or “smaller script.” They represent a rethinking of what web analytics should be: the data you actually need is far less than you assumed, and getting it should never come at the cost of your visitors’ privacy.
With GDPR, CCPA, and the ePrivacy directive tightening steadily, choosing a privacy-first analytics tool isn’t just a technical preference. It’s a forward-looking compliance strategy. While your competitors are still agonizing over whether the cookie-banner button should be green or blue, you don’t even need the banner anymore.
Choosing an analytics tool is, in the end, choosing the relationship you want with your users’ data. Do you treat it as an ore body to extract without limit, or do you take only what you need and respect where it came from? Plausible, Umami, and Fathom all give the same answer. They just get there by different roads.
Find the one that fits your road, and that’s enough.
Related reading
- 5 Postman Alternatives That Respect Your Data (2026)
- AI Meeting Notes Tools: Otter.ai vs Fireflies vs Fathom vs Granola in 2026
- MCP vs Zapier vs n8n: Which Layer Should You Build First for Your AI Agent in 2026?
- AI Agents Wont Replace Your Job Title. Theyll Eat the Flowchart Inside It.
Stay updated with our latest AI insights
