Plausible vs Fathom vs PostHog vs Umami: Best Privacy Analytics for Developer-Built Sites in 2026

Running seven content sites simultaneously means you're constantly staring at dashboards, trying to understand which articles are pulling traffic and which are dead weight. When Google Analytics 4 rolled out with its consent-wall requirements across EU traffic, I finally had a real reason to evaluate alternatives properly — not just bookmark them for later.

Over the past 18 months at Warung Digital Teknologi, I've integrated analytics into projects ranging from simple content aggregators to full SaaS products like our ContentForge AI Studio and SmartExam AI Generator. The requirements are completely different between a Laravel-powered content site on Hostinger VPS and a multi-tenant SaaS with funnel tracking, session replays, and feature flag experiments. After deploying all four of these tools across different projects, here's what actually matters.

Why Developers Are Leaving Google Analytics in 2026

GA4 is technically free, but the hidden costs add up. You need consent banners for EU visitors (killing 15-30% of trackable traffic), data sampling kicks in above certain thresholds, and the interface was clearly built for enterprise marketing teams — not developers who need clean API access to their data.

The real issue for privacy-conscious ops: GA4 sends user data to Google's servers in the US, which creates GDPR exposure. With the April 2026 Core Update putting renewed emphasis on site quality signals, having a slower site with a 135KB tracking script dragging down your LCP scores is no longer a trade-off worth making.

So: Plausible, Fathom, PostHog, or Umami. Here's the breakdown.

Quick Comparison Table

Plausible Analytics

Plausible is the tool I reach for first when spinning up a new content site. The 2.5KB script is genuinely imperceptible in Lighthouse audits — no consent banner required for GDPR compliance since no personal data is collected by default. The dashboard gives you everything a content operator needs in a single screen: pageviews, unique visitors, bounce rate, time on page, top sources, and top pages.

When I integrated Plausible into our seven aggregator sites on Hostinger VPS, setup took under 10 minutes per site. Drop in the script, set up a goal for newsletter signups, done. The API is clean and well-documented — I've used it to pull data into a Laravel dashboard for client reporting without any friction.

What Plausible does well:

• Simplest dashboard in this list — single page, no cognitive overhead
• Self-hosted version (AGPL) performs well on 1–2 vCPU; ClickHouse handles millions of events without breaking a sweat
• Revenue goals and funnels added in recent versions close the gap with PostHog for basic SaaS metrics
• EU-hosted cloud option (Frankfurt) makes GDPR compliance trivial to demonstrate

Where it falls short:

• No session replay or heatmaps — you'll need a separate tool if you're debugging UX issues
• Self-hosting requires ClickHouse, which needs at least 2GB RAM to run reliably. Fine on a VPS, painful on shared hosting
• Cloud plan counts per-site pageviews toward your limit — seven sites with combined 300K monthly pageviews pushes you into the $29–$49/month tier fast

Best for: Content sites, blogs, aggregators. Teams that want GDPR compliance without a lawyer.

Fathom Analytics

Fathom was the original "just track pageviews without the privacy headache" tool, and it still does that job well. The hosted version has excellent uptime, a genuinely polished UI, and EU data isolation built in. Their script size (~2.1KB) is actually smaller than Plausible's, and they have a custom domain proxy feature that bypasses most ad blockers.

The problem in 2026 is the value proposition has eroded. Starting at $14/month for 100K pageviews, Fathom is more expensive than Plausible for equivalent usage. Their self-hosted option ("Fathom Lite") hasn't had a Docker image update in over five years — effectively abandoned.

What Fathom does well:

• Cleanest dashboard UI in this comparison — it's genuinely beautiful
• EU data isolation is a first-class feature, not an add-on
• Ad-blocker bypass via custom domain is a real differentiator for ad-supported sites
• Excellent uptime SLA and fast CDN for the tracking script

Where it falls short:

• No self-hosting path that actually works in 2026
• No goals/funnels with the granularity Plausible now offers
• More expensive per-pageview than Plausible at every tier
• Feature development has slowed noticeably compared to Plausible and PostHog

Best for: Teams willing to pay a premium for a managed cloud-only tool with zero ops burden.

PostHog

PostHog is a different animal. This isn't a lightweight analytics script — it's a full product analytics platform: events, funnels, session replay, feature flags, A/B experiments, surveys, and a data warehouse. The free tier is genuinely generous: 1 million events per month, 5,000 session replays, and 1 million feature flag requests, all forever.

I use PostHog on our SaaS products — specifically on SmartExam AI Generator and BizChat Revenue Assistant — where I need funnel analysis (signup → activation → first exam generated), session replay to debug onboarding drop-offs, and feature flags to roll out new AI models to 10% of users before full release. PostHog handles all of that from a single SDK integration.

The tradeoff is real. PostHog's script is ~68KB compared to Plausible's 2.5KB. That's not a problem on a SaaS dashboard (users expect a heavier app), but it's a non-starter on content sites where Core Web Vitals matter. On our aggregator sites hitting ~200 daily record imports, every millisecond of LCP counts.

Self-hosting PostHog is possible but non-trivial: it requires Docker Compose at minimum, with PostgreSQL, Redis, ClickHouse, and Kafka in the stack. Minimum viable self-hosted setup needs 4 vCPU and 16GB RAM. That's fine for a dedicated VPS but adds serious ops overhead.

What PostHog does well:

• The only tool here with session replay — essential for SaaS UX debugging
• Feature flags with percentage rollouts and multivariate experiments built in
• SQL access to your event data directly from the PostHog UI
• 1M events/month free is enough for most early-stage SaaS products indefinitely
• Active open-source development — new features ship constantly

Where it falls short:

• 68KB script is too heavy for content/SEO sites
• Self-hosting requires serious infrastructure — not a weekend project
• Dashboard complexity can overwhelm teams that just need pageviews
• Privacy posture is more complex — session replay captures user behavior, which requires careful GDPR handling

Best for: SaaS products that need product analytics, not just traffic numbers. Teams building features and need funnel data.

Umami

Umami is the self-hoster's favorite. MIT-licensed (no AGPL copyleft concerns), it runs on Node.js with MySQL or PostgreSQL — the exact stack most Laravel/Node projects already have running. Deploying Umami on an existing Hostinger VPS alongside a MySQL database that's already provisioned takes under 20 minutes.

The script footprint (~2KB) matches Plausible, it's cookieless by default, and the dashboard covers the essentials: pageviews, visitors, bounce rate, referrers, and custom events. Recent versions added team/member management and a cloud tier with 1M events/month free.

Where Umami wins on raw economics: if you're already paying for VPS hosting and have spare capacity, Umami is zero incremental cost. Across 7 sites with shared MySQL infra, adding Umami to an existing server that has headroom is genuinely free. I measured the additional query load as negligible on a 2-vCPU Hostinger VPS running three moderately-trafficked sites simultaneously.

What Umami does well:

• MIT license — no copyleft, can embed in commercial projects freely
• Runs on MySQL/PostgreSQL you likely already have — minimal infrastructure delta
• Free cloud tier (1M events/mo for 3 sites) genuinely competes with Plausible's self-hosted option
• Custom events API is clean and well-documented for Laravel/Node integrations

Where it falls short:

• UI is functional but behind Plausible and Fathom in polish
• No revenue tracking, funnels, or goals beyond basic custom events
• Smaller community means slower ecosystem (integrations, plugins)
• Self-hosted version requires manual updates — no auto-update mechanism

Best for: Developers who self-host everything and want zero licensing friction. Sites with existing MySQL/PostgreSQL infrastructure.

My Actual Setup: What I Use Where

From 11+ years building and running web projects, the "one tool fits all" approach to analytics fails every time. Here's the split I've settled on after running all four in production:

Content aggregator sites (all 7): Plausible self-hosted on a 2-vCPU VPS. The ClickHouse requirement was initially annoying, but the query performance on 100K+ monthly pageviews is instant. I keep all 7 sites under one Plausible instance, which is supported out of the box.

SaaS products (SmartExam, BizChat, ContentForge): PostHog cloud on the free tier. 1M events/month comfortably covers early-stage product usage. Session replay alone is worth the integration cost — I've caught three critical onboarding bugs in the past six months purely from watching session replays of users who dropped off at the activation step.

Client projects where self-hosting is required: Umami. MIT license means I can hand off the codebase to clients without AGPL compliance questions, and it runs on their existing MySQL setup without provisioning new infrastructure.

I haven't actively used Fathom since mid-2025 — the price-to-feature ratio no longer holds up against Plausible for content sites, and for SaaS you need PostHog anyway.

Which Should You Pick?

Here's the honest decision tree:

• Content site / blog / aggregator + want simplest possible setup: Start with Umami cloud (free tier) or Plausible cloud ($9/mo). Both are GDPR-compliant out of the box.
• Self-host everything on existing VPS with MySQL: Umami. Zero new infrastructure, MIT license, done.
• SaaS product that needs funnel analysis + session replay: PostHog free tier. Non-negotiable — nothing else in this list does what PostHog does.
• Managed cloud, beautiful UI, willing to pay a small premium: Fathom. It's the Stripe of analytics — just works, no ops overhead.
• Running many sites and want unified dashboard + self-host: Plausible self-hosted. Multi-site support is first-class, ClickHouse handles the scale.

The mistake I see developers make is reaching for PostHog on content sites because it's "more capable." You don't need session replay on a blog. What you need is fast page load and clean traffic numbers — and a 68KB analytics script fights directly against both. Match the tool to the use case, not the feature list.

Bottom Line

In 2026, there is no excuse to still be on Google Analytics 4 for privacy-sensitive projects. All four of these tools offer better GDPR compliance, lighter scripts, and cleaner data than GA4 — the question is just which tradeoffs matter for your specific stack.

I'd recommend Plausible as the default for most developer-built content sites. Start with Umami if you're already managing a VPS and want zero additional cost. Graduate to PostHog when your SaaS product needs real product analytics. And skip Fathom unless you specifically need the managed EU-isolation guarantee and don't want to think about ops at all.

All four have free plans or self-hosted options — there's no reason not to test them on a real project before committing.