I tested 8 analytics tools. Here's what they all get wrong.

Published: (February 21, 2026 at 10:03 AM EST)
3 min read
Source: Dev.to

Source: Dev.to

I spent the last six months deep in web analytics while building my own platform. Along the way I tested GA4, Hotjar, Mixpanel, Amplitude, Plausible, Fathom, PostHog, and Matomo. Every single one forces you into a trade‑off that shouldn’t exist.

GA4

  • Free, but requires a certification to find basic data.
  • Filtering for a single page takes six clicks.
  • Reports are sampled above 10 M events with error rates up to 30 %.
  • Cookie‑consent banners hide 40‑60 % of EU traffic.
  • The “free” tool effectively requires BigQuery for anything useful.

Hotjar

  • Great heatmaps, but the script adds ~830 ms to page load and ~0.5 MB to page weight.
  • Free plan records only 35 sessions per day.
  • Full Business tier across all products costs $922/mo.
  • Session recordings break on Next.js and React apps when CSS filenames change between deploys.

Mixpanel

  • Powerful product analytics, but 10 M events/month costs ≈ $2,500/mo.
  • One Hacker News user reported being downgraded from 20 M free events to 10 K after switching to a paid plan.
  • Client‑side tracking loses 30‑50 % of users to ad blockers (per their own docs).

Amplitude

  • Similar power and similar problems.
  • Growth plan starts at ≈ $30 K/year.
  • UI has so many options that teams spend more time navigating the tool than gaining insights.
  • Requires 25 K+ monthly users minimum for accurate advanced analytics.

Plausible

  • Beautiful and simple.
  • Resets visitor identity daily, so one person visiting 5 days counts as 5 uniques.
  • No heatmaps, session replay, or funnels on the self‑hosted version.
  • Self‑hosted lacks the bot detection the cloud version uses—one developer saw stats jump from 200 to 5 000 visitors/day after switching.

Fathom

  • Clean and privacy‑first.
  • Custom events can’t send metadata.
  • No goal attribution with UTM parameters.
  • No funnels, entry/exit pages, or scroll depth.
  • Completely closed‑source with no self‑hosting option.
  • Shows you what happened, never why.

PostHog

  • Tries to do everything.
  • JS SDK is 52 KB+.
  • Self‑hosting requires ClickHouse, Kafka, Redis, PostgreSQL, Zookeeper, and MinIO running simultaneously.
  • One developer ran up a $10 K bill in half a day.
  • Another needed 64 GB RAM and still couldn’t get it working.

Matomo

  • The oldest open‑source option.
  • MySQL‑based architecture means reports can take hours to generate.
  • UI has barely evolved in 10+ years.
  • Heatmaps, session replay, funnels, and A/B testing are all premium plugins costing €1 200+/year even for self‑hosted installations.

The pattern nobody talks about

Every tool falls into one of two camps:

  1. Simple counters (e.g., Plausible, Fathom) that show what happened but not why.
  2. Enterprise platforms (e.g., GA4, Amplitude, PostHog) that can answer deep questions but require weeks of learning and dedicated engineering.

Nothing sits in the middle: simple to start, powerful when you need it, privacy‑first without sacrificing conversion tracking. That gap is exactly why I built Zenovay.

Why I built Zenovay

  • One dashboard with analytics, heatmaps, session replay, AI insights, and revenue attribution.
  • No cookies.
  • EU‑hosted.
  • $20/mo.

It’s not perfect yet, but the trade‑off between simple and powerful shouldn’t be a trade‑off at all.

What analytics tool are you currently using, and what’s the one thing that frustrates you most about it?

0 views
Back to Blog

Related posts

Read more »

Stop Using .any? the Wrong Way in Rails

Introduction A single block passed to .any? can silently load thousands of records into memory—no warnings, no errors, just unnecessary objects. Most Rails dev...