Google Analytics is slow, privacy-hostile, and reports fewer visitors than reality due to ad-blockers
First-party analytics in 2026 means lightweight scripts, no cookies, and server-side fallbacks
My stack is Plausible for traffic, PostHog for product analytics, and Shopify Analytics for sales
No cookie banner needed if you stick to aggregated first-party data with no personal identifiers
Total cost is 9 EUR a month and the dashboards load in under a second
I killed Google Analytics on raxxo.shop six weeks ago. Six minutes of migration, one afternoon of dashboard rebuilding, and suddenly my site loaded 240 milliseconds faster on mobile. No cookie banner. No consent flow. No 17-tab Analytics 4 dashboard where every report is labeled "Exploration" and nothing is where you left it.
Most solo builders and small teams are still running GA4 out of habit. It is not the best tool, and it has not been the best tool since around 2022. The 2026 first-party analytics stack is cheaper, faster, and respects visitors. It also gives you numbers you can trust, which is more than GA4 ever did.
Here is exactly what I run now, how I set it up, and what I gave up to get there.
Why I Killed Google Analytics on raxxo.shop
The final straw was not privacy. It was accuracy. GA4 was reporting about 63 percent of the traffic that Vercel Analytics saw on the same pages. The gap is real and it is not a bug. It is the cumulative effect of ad-blockers, Safari's ITP, Firefox's tracking protection, and the growing slice of visitors who just click "Reject All" on every cookie banner.
If you optimize against GA4 numbers, you are optimizing against a sample that excludes your privacy-conscious users. Those are often your best users. Developers, designers, people who pay for software. The demographic RAXXO actually sells to. Missing them in analytics means I was tuning my copy and layout for the wrong crowd.
The privacy side is the secondary win. I am based in Berlin. GDPR enforcement in Germany is not a theoretical risk. The Datenschutzbehörde in Bavaria fined a website 100,000 EUR in 2025 for GA implementation that wasn't even the worst version. I do not want to be that case study.
The performance hit is the tertiary win, but for an e-commerce site it matters. Google Tag Manager plus GA4 plus Google Ads pixel plus Facebook pixel used to push about 180 kilobytes of JavaScript into the critical path on every page. First-party analytics runs at 1 to 2 kilobytes. Mobile conversion improved by about 6 percent in the month after removal. I cannot prove causation, but the correlation is strong and my site feels faster when I test it on my own phone.
The Three-Tier First-Party Analytics Stack
I used to think analytics was one tool. It is not. It is three very different jobs, and trying to do them all in GA4 is why GA4 feels terrible.
Tier one is traffic analytics. Who is visiting, from where, on what device, to what pages. This is the dashboard you look at every morning. It needs to be fast, clean, and glanceable. Plausible, Fathom, and Simple Analytics all do this job well.
Tier two is product analytics. What are users doing once they arrive. Which buttons they click, which flows they abandon, where they get stuck. This is the dashboard you look at when you are deciding what to build next. PostHog, Amplitude, and Mixpanel do this job. PostHog is my pick because the free tier is generous and the self-hosted option exists as an escape hatch.
Tier three is money analytics. What is actually paying the bills. For me this lives entirely in Shopify Analytics because Shopify already knows about every order and subscription event. Your sales tool is whatever platform processes your payments. Do not try to rebuild it in a separate analytics tool.
The mistake with GA4 is pretending it can do all three. It cannot do any of them well. Three dedicated tools, each good at one job, costs less and works better than one tool trying to do everything.
Tool Comparison: Plausible vs Fathom vs Umami vs PostHog
I tried most of the first-party analytics options in 2025 before settling. Here is the shortlist with honest notes.
Plausible. 9 EUR a month for 10,000 page views. Open source, EU-hosted, no cookies by default, script is 1 kilobyte. Dashboard is a single page. Event tracking is clean. Goals and funnels are basic but present. This is my pick for traffic analytics.
Fathom. Similar to Plausible, slightly pricier, slightly prettier UI, also good. If Plausible did not exist I would use Fathom. The dashboards are almost interchangeable.
Umami. Free if you self-host. 20 EUR a month if you use the hosted version. Self-hosting is real work for a solo builder, and my time is worth more than the saving. I would only recommend Umami for teams who already run their own infra.
PostHog. Free up to 1 million events a month. Event tracking, session replay, feature flags, A/B tests, product analytics, all in one. This is the tool for tier two, not tier one. It can do traffic analytics but it is overkill and the dashboards are busier than Plausible.
Simple Analytics. Good, pricier than Plausible, slightly nicer UI. Same job. Pick one, not both.
Vercel Analytics. Bundled with the Pro plan. Good real-user-monitoring data, decent traffic breakdown, but limited event tracking. I use it as a cross-check against Plausible. If the two numbers disagree by more than 5 percent, something is broken.
The comparison that matters is not feature-by-feature. It is which dashboard you will actually open every morning. The one you love looking at is the one that changes your behavior. I open Plausible every morning. I barely opened GA4.
The No-Cookie-Banner Setup
This is the part most people get wrong. You can run first-party analytics without a cookie banner, but only if you set it up correctly.
The rule that matters in the EU is this. If you store anything on the user's device that is not strictly necessary, you need consent. "Strictly necessary" means the thing required to deliver the service the user asked for. Analytics is not strictly necessary. So if your analytics tool sets cookies, you need a banner.
Plausible, Fathom, and Simple Analytics do not set cookies. They generate a daily rotating hash from visitor attributes like IP and user agent, use it to deduplicate visits in memory, and discard the raw data. No cookie, no stored identifier on the device, no banner required by most reasonable readings of ePrivacy and GDPR.
PostHog is where people trip up. Default PostHog sets cookies for session persistence. If you want to run PostHog without a banner, you have to configure it for anonymous, cookieless mode. This means no session replay, no cross-session user tracking, and no persistent identifiers. You get aggregate event counts, funnels on single-session data, and not much else.
My split is this. Plausible runs on every page. PostHog runs with cookies but only after consent. On raxxo.shop, the consent flow is a single button at the bottom of the shop pages that says "Enable product analytics to help improve this site." About 40 percent of visitors click it. That is enough data for product decisions, and the 60 percent who do not click get a faster page with nothing stored on their device.
One gotcha worth naming. IP addresses in server logs are technically personal data in the EU. Plausible and similar tools anonymize IPs at the edge, so they never log the raw address. If you are rolling your own analytics with raw server logs, you have work to do before it is GDPR compliant.
What I Measure Now That I Ignored Before
Switching tools made me rethink which numbers actually matter. Here is what I track now that I track.
Top ten pages by visits, daily. This is the front page of Plausible. If a blog post unexpectedly takes off, I see it within hours. GA4 buried this behind three clicks.
Referrer sources, daily. Hacker News, Dev.to, Twitter, direct. The sources that send me buyers are different from the sources that send me readers. I weight my content syndication effort accordingly.
Conversion events on key pages. PostHog tracks button clicks on the pricing page, the shop CTA, and the blog newsletter signup. I look at this weekly to spot dead buttons.
Sales per product. Shopify Analytics. I look at this on Monday mornings. Which products are moving, which are stale, which need a new blog post to boost search visibility.
Bounce rate on landing pages. Plausible's bounce rate calculation is simpler than GA4's and more useful in practice. Over 75 percent and I rewrite the hero.
What I stopped measuring. Time on site. Sessions. Engagement score. GA4-style "engagement rate." These were vanity metrics dressed up as quality signals. None of them predicted actual sales for me. I cut them and did not miss them.
What I started measuring instead. Ratio of returning to new visitors by week. This is the health metric for a content-driven site. If returning visitors are not growing, the content is not building a loyal audience, no matter how many new visitors the latest viral post brings in.
What This Actually Costs
Plausible Pro. 9 EUR a month.
PostHog. Free under 1 million events monthly. I am nowhere near the cap.
Shopify Analytics. Included in my Shopify plan.
Vercel Analytics. Included in my Vercel Pro plan.
Total. 9 EUR a month, plus the tools I was already paying for. I was paying Google nothing for GA4, but I was paying in load time, in broken reports, and in the hours I spent trying to get the dashboards to show what I wanted. I save more than 9 EUR of my time every week by not dealing with GA4.
Bottom Line
Google Analytics is not free. You pay in performance, in accuracy, in privacy exposure, and in the time you waste navigating a UI that was designed for ad agencies and not for you. First-party analytics in 2026 is mature, cheap, and fast. Plausible for traffic, PostHog for product, Shopify or your payments platform for sales. Three tools, each good at one job, 9 EUR a month, no cookie banner if you set it up right.
If you are still on GA4 out of habit, spend an afternoon migrating. You will not miss it. You will wonder why you waited.
Top comments (0)