Web Analytics Services Compared: 10 Options for 2026

The right web analytics service depends on one question: do you need page-level traffic numbers, or user-level product behavior? For traffic, Google Analytics 4 and Cloudflare Web Analytics are free, and Plausible, Fathom and Umami are simple, cookie-free paid options. For funnels, retention and per-user events, look at PostHog or Mixpanel. Adobe Analytics is for enterprises with a sales contract.

Below is a web analytics comparison of ten services (plus Amplitude and Microsoft Clarity, which often end up on the same shortlist), with pricing taken from each vendor's own page in November 2026. Instead of repeating feature lists, it compares what actually differs between web analytics services once you're past the signup page: the data model, how long data is kept, sampling, privacy and where data is processed, self-hosting, APIs, raw data export, and performance monitoring. Prices change often, so check the linked page before you commit.

If you want a broader roundup of website traffic tools, including traffic checkers like Similarweb and Semrush that estimate other people's sites, read our guide to the 10 best website traffic analysis tools. This page goes deeper on choosing a web analytics service for your own site.

Quick comparison

"Cookies: No" means the vendor says its default script sets no cookies. Whether you need a consent banner also depends on your jurisdiction and what else your site collects, so check with whoever handles your privacy compliance. Matomo Cloud's price is the one shown to US visitors; it is shown in euros in Europe.

Google Analytics is the default by a wide margin: W3Techs finds it on 46.8% of all websites, or 81.7% of sites that use any traffic analysis tool it can detect, as of November 2026. Matomo is on 1.4%, and Plausible and PostHog on 0.2% each. Popularity tells you which tool agencies and new hires already know, not which one fits your site.

How to choose: four questions

1. Traffic or product behavior? "How many visitors, from where, to which pages" is traffic analytics. "Which users finished onboarding and came back on day 7" is product analytics. Product tools can do traffic too, but traffic tools can't do per-user funnels.
2. Do you need to avoid a cookie banner? Plausible, Fathom, Umami, Cloudflare and Vercel are built to work without cookies. GA4 sets cookies by default, so in the EU and UK it normally sits behind consent, and visitors who decline aren't fully measured.
3. Who owns the data? If data must stay on your servers, self-host Matomo, Plausible Community Edition, Umami or PostHog. If it only has to stay in the EU, several hosted services now offer EU processing (see privacy and data processing below).
4. How will the bill grow? Pageview-priced tools (Plausible, Fathom) are predictable. Event-priced tools (Mixpanel, PostHog, Vercel, Umami) grow with every tracked click, so decide what to track before you add track() calls everywhere.

Two follow-up questions settle most close calls. Will you need the raw data? If you plan to join analytics with your own database or a BI tool, only some services give you event-level export (the vendor comparison table has the details). Do you also want page speed data? Some services report Core Web Vitals from real visitors, which saves you a separate monitoring tool.

Web analytics vendors comparison: data model, retention, API and export

This is the table most website analytics comparison pages skip, and the one to use when you compare website analytics services for anything beyond a hobby site. Pricing is easy to change later; a service's data model, retention and export options decide what questions you can still answer a year from now. Values are from each vendor's pricing page or docs, as of November 2026. Adobe Analytics is left out because these terms are set in its sales contract.

Sources: GA4 360 vs standard, GA4 BigQuery export, Mixpanel pricing, Amplitude pricing, PostHog pricing and batch exports, Plausible pricing, Fathom pricing, Umami pricing, Matomo pricing, Cloudflare FAQ, Vercel limits, Clarity Data Export API.

Data models: pageviews, events and users

Web analytics services store one of three shapes of data, and the shape limits your reports more than any feature list:

• Aggregated pageviews and events (Plausible, Fathom, Cloudflare, Vercel). The service counts visits, pages, sources and goals, but never builds a profile of one person. You can't ask "what did this user do last week", which is the point for privacy, and a limit for a SaaS dashboard.
• Sessions with a visitor log (Matomo, Umami). You can see individual visits and their path through the site, and Matomo keeps the raw rows in its own database.
• Events tied to users (GA4, Mixpanel, Amplitude, PostHog). Every action is an event with a user ID, so funnels, cohorts and retention work across days and devices once you identify the user at login.

The model also drives pricing. Event-priced services bill every click you track, so a busy product can cost far more than its pageview count suggests. Amplitude's Plus plan, for example, is sized in both monthly tracked users and events ("up to 700K MTU / 70M EV" on its pricing page). For a high-traffic consumer app with many anonymous visitors, look at how each service prices anonymous versus identified events: PostHog charges $0.00005 per anonymous event and $0.000198 per identified event at its first paid tier.

Sampling and data freshness

Sampling means a report is computed from a subset of your data and scaled up. For most small sites it never shows up. In GA4, explorations are sampled once a query touches more than 10 million events, and unsampled explorations are an Analytics 360 feature (Google). Cloudflare keeps unsampled beacon data for 7 days, then aggregates it to about 10% of the original volume, and the GraphQL API exposes a sampleInterval field so you can see how much sampling was applied (Cloudflare FAQ). Microsoft Clarity says it does not sample. If you need exact counts for a low-traffic page months later, pick a service that keeps raw rows (Matomo, PostHog, GA4 via BigQuery).

APIs and raw data export

An API lets you pull numbers into your own dashboard; a raw export gives you every event so you can run any query. They're not the same thing:

• GA4 has the Data API for reports and a BigQuery export for raw events. Standard properties can export up to 1 million events a day in the daily export; streaming export is unlimited but costs $0.05 per GB (roughly 600,000 events per GB), plus BigQuery storage and query costs (Google). You can start on the free BigQuery sandbox.
• PostHog has batch exports to seven destinations, including S3, BigQuery and Snowflake, and a SQL editor inside the app.
• Mixpanel gives every plan the ingestion and export API, but the Query API and warehouse connectors start on Growth.
• Plausible puts its Stats API on the Business plan and raw event exports on Enterprise. Fathom includes its Analytics API and exports on every plan. Umami Cloud includes the API from Pro up.
• Microsoft Clarity's Data Export API is tight: 10 requests per project per day, only the last 1 to 3 days, up to three dimensions and 1,000 rows per response.

Here's what a raw-data query looks like once GA4's BigQuery export is on. It counts pageviews per day for the last 28 days (replace the project and dataset ID with yours):

SELECT
  event_date,
  COUNT(*) AS pageviews,
  COUNT(DISTINCT user_pseudo_id) AS users
FROM `my-project.analytics_123456789.events_*`
WHERE event_name = 'page_view'
  AND _TABLE_SUFFIX BETWEEN FORMAT_DATE('%Y%m%d', DATE_SUB(CURRENT_DATE(), INTERVAL 28 DAY))
                        AND FORMAT_DATE('%Y%m%d', CURRENT_DATE())
GROUP BY event_date
ORDER BY event_date;

And the same idea with the Plausible Stats API v2, which takes a JSON query (docs):

curl -X POST https://plausible.io/api/v2/query \
  -H "Authorization: Bearer $PLAUSIBLE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"site_id": "example.com", "metrics": ["visitors", "pageviews"], "date_range": "28d", "dimensions": ["time:day"]}'

Web analytics with performance monitoring (Core Web Vitals)

Traffic numbers tell you what people did; performance data tells you whether slow pages stopped them. A few web analytics services collect both from the same script, which is the cheapest way to get real-user web performance analytics. Google's thresholds for a "good" page, measured at the 75th percentile of page loads, are an LCP of 2.5 seconds or less, an INP of 200 milliseconds or less and a CLS of 0.1 or less (web.dev).

Sources: Cloudflare Core Web Vitals, Vercel Speed Insights pricing, PostHog web vitals, Matomo page performance.

Cloudflare's Debug View is the standout for a free tool: it lists the top five elements hurting each metric, and for LCP it splits load time into time to first byte, resource load delay, resource load duration and element render delay. That's usually enough to know whether to fix the server, the image or the render path.

To get Core Web Vitals into GA4, install Google's web-vitals library (about 3 KB brotli-compressed) and send each metric as an event:

npm install web-vitals

import { onCLS, onINP, onLCP } from 'web-vitals';

function sendToGoogleAnalytics({ name, delta, value, id }) {
  // assumes the gtag() snippet is already on the page
  gtag('event', name, {
    value: delta, // use delta so values can be summed
    metric_id: id, // groups the deltas of one page load
    metric_value: value,
    metric_delta: delta,
  });
}

onCLS(sendToGoogleAnalytics);
onINP(sendToGoogleAnalytics);
onLCP(sendToGoogleAnalytics);

Register metric_value as a custom metric in GA4 if you want to chart it; standard properties allow 50 event-scoped custom metrics.

The services in detail

Google Analytics 4

GA4 is free and the default choice for marketing teams, mainly because it connects to Google Ads and Search Console. Everything is an event. Pageviews, scrolls and outbound clicks are collected automatically with Enhanced Measurement.

The limits catch people out. In the standard (free) version, event data retention is 2 or 14 months, explorations are sampled above 10 million events per query, and the BigQuery daily export is capped at 1 million events (Google's 360 vs standard table). The default retention is 2 months, so change it to 14 months in Admin > Data collection and modification > Data retention on day one. Standard properties also cap you at 50 event-scoped custom dimensions, 50 custom metrics and 30 key events. Analytics 360 raises the limits (up to 50 months of retention, 1 billion events per query before sampling, 125 custom dimensions) but is sold through a sales contract.

Good for: marketing sites, ad-funded traffic, teams already in Google's ecosystem. Weak at: simple dashboards, privacy-sensitive audiences. If you're a developer working with the GA4 APIs, Measurement Protocol or DebugView, see our guide to web analytics tools for developers.

Adobe Analytics

Adobe Analytics is an enterprise product sold with Adobe Experience Cloud, alongside Customer Journey Analytics. Pricing isn't published, and you book a demo to get a quote. It makes sense when you already run Adobe's marketing stack and have analysts to configure it. It's rarely a fit for startups or developer tools.

Mixpanel

Mixpanel is event-based product analytics: funnels, retention, flows and cohorts. As of November 2026 the free plan covers up to 1M events per month, unlimited seats and 5 saved reports per seat (Mixpanel pricing). The Growth plan keeps the first 1M events free and then steps up: 1.5M events cost $140/month billed monthly ($100 billed yearly), 5M cost $613/month ($438 yearly) and 20M cost $2,289/month ($1,635 yearly). Data is kept for 2 years by default, and US or EU data residency is chosen when you create a project and is off by default, so decide before you send the first event.

Good for: SaaS and apps that care about activation and retention. You need to plan an event schema up front. Random event names quickly make reports useless.

PostHog

PostHog bundles product analytics, web analytics, session replay, feature flags, A/B tests and more. As of November 2026, its pricing page lists 1M analytics events, 5,000 session recordings, 1M feature flag requests and 1M data warehouse rows free each month. After that, analytics costs $0.00005 per anonymous event and $0.000198 per identified event at the first paid tier. Free projects keep event data for 1 year; paid plans keep it for 7 years. You choose US (Virginia) or EU (Frankfurt) hosting.

The self-hosted version is a free Docker Compose deployment under the MIT license, but PostHog calls it officially unsupported and suggests a machine of about 4 vCPU and 16 GB RAM (PostHog self-host docs). Its own guidance is that if you expect to ingest more than 300k events or 1k recordings into the instance, Cloud is the better choice.

Good for: engineering-led teams who want one tool instead of five. It's a lot of product, so start with web analytics and replay before turning on everything.

Plausible

Plausible is a lightweight, cookie-free traffic dashboard: visitors, pages, sources, countries, devices, goals. Cloud plans are Starter $9/month, Growth $14/month and Business $19/month at the 10k-pageview tier, with a 30-day free trial and no card required. Starter covers one site with no retention limit; Growth adds up to 3 sites and team members; Business adds up to 10 sites, the Stats API, funnels and ecommerce revenue attribution (Plausible pricing). Data is processed and stored in the EU on infrastructure owned by European companies, and unique visitors are counted with a hash whose salt is deleted every 24 hours (data policy).

Plausible Community Edition is free to self-host under AGPLv3, but it's released twice a year and leaves out cloud-only features like funnels, ecommerce revenue goals and SSO (details). Self-hosting is community supported only.

Fathom

Fathom is similar in spirit to Plausible: a one-page dashboard and no cookies. It starts at $15/month for up to 100,000 monthly pageviews, then $25 for 200,000, $45 for 500,000 and $60 for 1,000,000, with 50 sites included on every plan. There's a 7-day trial (card required) and no free plan. Hosted only. The Analytics API and exports are on every plan.

Fathom's EU isolation is on by default for everyone: EU visitors' pageviews go to Fathom's servers inside the EU, where the IP address is stripped before anything reaches its US infrastructure; anonymized data is then stored in the US (Fathom EU isolation).

Umami

Umami is open source and free to self-host. Umami Cloud's Hobby plan is free for up to 100K events a month on 1 website with 6 months of retention. Pro is $20/month for 1M events, up to 20 websites and 2 years of retention, then $0.00003 per extra event; Business is $200/month for 10M events and 5 years (Umami pricing). It's a good fit for developers who want Plausible-style simplicity and are happy to run a small Node.js app and a database. DevHunt has a Plausible vs Umami comparison if you're deciding between the two.

Self-hosting needs Node.js 20.19 or newer and PostgreSQL 12.14 or newer, or you can run the official Docker Compose file (Umami install docs):

git clone https://github.com/umami-software/umami.git
cd umami
docker compose up -d

Matomo

Matomo is the closest open-source match to the old Universal Analytics: goals, ecommerce, heatmaps (paid plugin), tag manager. Matomo On-Premise is free with unlimited users and hits. Matomo Cloud starts at $26/month (excluding tax) for 50,000 hits as shown to US visitors, with euro pricing in Europe. Cloud keeps raw data for 24 months and processed reports forever, and stores data in Frankfurt, Germany. You can run it without cookies, but some features work better with them.

For On-Premise, Matomo 6 needs PHP 8.1 or newer and MySQL 8.0 or newer (or MariaDB 10.6 or newer). Matomo's own sizing guide suggests one server with 2 CPUs, 2 GB RAM and 50 GB SSD for up to 100,000 monthly pageviews, and 4 CPUs, 8 GB RAM and 250 GB SSD for up to 1 million (Matomo requirements). Above 10 million pageviews it recommends a separate database server.

Cloudflare Web Analytics

Free, and it works without moving your DNS to Cloudflare: you add a JavaScript snippet (sites proxied through Cloudflare can enable it automatically). Cloudflare says it uses no cookies or localStorage and doesn't fingerprint visitors by IP address or user agent. You get pageviews, referrers, countries and Core Web Vitals, but no custom events or funnels. There's a soft limit of 10 sites per account and six months of history (Cloudflare FAQ). It's a solid zero-cost baseline for a docs site or blog.

Vercel Web Analytics

If you deploy on Vercel, Web Analytics is a toggle plus a package. Hobby includes 50,000 events a month with a 1-month reporting window; after a three-day grace period, collection pauses. Pro has no included events and charges $0.03 per 1,000 events, with a 12-month reporting window, custom events with 2 properties each, and the Web Analytics Plus add-on ($10/month per team) for 24 months, 8 properties and UTM parameters (Vercel limits and pricing). Events are shared across all projects in the team. It's convenient, but it only covers sites hosted on Vercel.

More services on the shortlist: Amplitude and Microsoft Clarity

Amplitude competes with Mixpanel for product analytics. Its free plan covers 2M events a month, 10,000 session replays and 1 year of data, with unlimited team members and no card. Plus starts at $0 with the first 2M events free and keeps 2 years of data; Growth and Enterprise are custom quotes. An EU data center is an option on every plan.

Microsoft Clarity isn't a traffic counter first. It's free session recording and heatmaps, with no traffic limits and no sampling, and it uses first-party cookies, so Microsoft requires a consent signal for visitors in the EEA, UK and Switzerland. Pair it with a counter like GA4 or Plausible.

Vercel Analytics vs Google Analytics vs PostHog

These three come up together because many Next.js sites start on Vercel and then wonder whether to add GA4 or PostHog. They solve different problems:

The short version: Vercel Analytics vs Google Analytics is simplicity and privacy against depth and ad integrations. If you run ads or need conversion attribution, GA4 wins. If you only want clean traffic numbers on a Vercel-hosted site, Vercel's dashboard is enough and needs no consent banner in most setups. Vercel Analytics vs PostHog is a different question: PostHog is the pick when the site is also a product and you need per-user funnels, replays or feature flags, and it works wherever you host. Plenty of teams run Vercel Analytics for the marketing pages and PostHog inside the app.

Privacy, consent and where your web analytics data is processed

Under the GDPR, any hosted web analytics service that handles personal data (an IP address counts) on your behalf is a processor, and you are the controller. Article 28 requires a contract that sets out the subject matter, duration, nature and purpose of the processing and the types of personal data involved. In practice that means each vendor's data processing agreement (DPA), which most publish on their site. Treat your web analytics data processing service like any other subprocessor: know where the data goes and what is stored.

What the vendors document, as of November 2026:

• Plausible: processed and stored in the EU on European-owned infrastructure; no cookies, browser cache or local storage; no raw IP addresses or user-agent strings stored.
• Fathom: EU visitors' data is processed in the EU and the IP is stripped there; anonymized data is stored in the US.
• PostHog: choose US (Virginia) or EU (Frankfurt) cloud at signup.
• Mixpanel: US or EU data residency, chosen when you create the project.
• Amplitude: EU data center option on every plan.
• Matomo Cloud: stored in Frankfurt, Germany. Matomo On-Premise keeps everything on your own server, so there's no analytics processor at all.
• Cloudflare: no cookies or localStorage, and no fingerprinting by IP or user agent.
• GA4: first-party _ga cookies with a 2-year default lifetime; in the EU and UK it normally needs consent.

Cost at scale: what 5 million events a month costs

Entry prices look similar; the gap opens as traffic grows. Here is the monthly price at 5 million pageviews or events, calculated from each vendor's published rates as of November 2026:

Self-hosting changes the math: Matomo On-Premise, Umami and Plausible CE cost only the server, which for a few million pageviews is one machine in Matomo's sizing guide. The real cost is the hours you spend on upgrades, backups and security patches, so price your time in.

Web server analytics and other ways to measure

If you run a web service, analytics doesn't have to come from a JavaScript tag. Web server analytics read your access logs instead, so they see visitors who block scripts, API clients and bots. That's a strength for an API or a developer audience and a weakness for marketing reports, because logs count every asset request and crawler. Matomo can import Apache, Nginx and IIS logs (Matomo log analytics), and the free GoAccess log analyzer turns an Nginx or Apache log into a report in one command; the traffic tools guide has the commands. For definitions of sessions, bounce rate and unique visitors across these tools, see understanding web traffic stats.

Website analysis tools are not web analytics software

A lot of searches for "website analysis tools" or "the best website analysis tool" are really after page checks, not traffic. Web page analysis tools such as PageSpeed Insights and Lighthouse grade one URL for speed, accessibility and SEO, and website benchmarking tools compare those scores across pages or competitors. Web analytics software, by contrast, records what real visitors do on your own site over time. You usually want both: an analysis tool before you ship, and one of the web analytics platforms above after. Website analytics software, website statistics tools and web tracking software are all names for the second kind.

Web analytics services vs agencies: implementation services

"Web analytics services" also means agencies and freelancers who set up tracking for you. They're worth it when you have ad spend, ecommerce or a complex GA4 and Google Tag Manager setup, and not worth it for a blog that needs a Plausible snippet. If you hire for web analytics implementation services, ask for:

1. A written tracking plan: every event name, its properties and why you need it.
2. Consent handling that matches your regions (for GA4, how consent is passed to tags).
3. Key events and conversions marked in the tool, not just tracked.
4. Admin access in your name, not the agency's, so you keep the property if you part ways.
5. A short handover document and a test that every event fires.

Whichever web analytics service providers you work with, the account should belong to you. Most small teams don't need an agency at all: the website analytics services compared above install in under an hour.

Which web analytics service should you pick?

Setting one up: a worked example with Plausible

Most privacy-first services install the same way. Using Plausible as the example:

1. Create a site in the dashboard with your domain.
2. Add the script tag it gives you to the <head> of every page. In Next.js, put it in the root layout.
3. Load your site and confirm the visit appears in the real-time view. In DevTools, the Network tab should show a request to the Plausible event endpoint with a 2xx status.
4. Add goals for the actions that matter, such as signups or "Copy install command" clicks.
5. Exclude your own visits (each service documents a way, usually a localStorage flag or a browser extension).

Run two services side by side for a few weeks if you're migrating. Numbers will differ noticeably, because each tool handles ad blockers, bots and sessions differently. Compare trends, not absolute totals. Plausible's Starter plan can import your Google Analytics history, which helps if you switch.

Common mistakes

• Tracking everything. Event-based bills grow with every listener. Track the 5-10 actions you'll actually make decisions on. PostHog's own docs estimate that web vitals alone add about 30% to pageview event volume.
• Leaving GA4 retention at 2 months. Year-over-year explorations become impossible.
• Not exporting raw data early. Free GA4 deletes event data after 14 months at most. A BigQuery export (or a service with longer retention) is how you keep it.
• Picking the wrong data residency. Mixpanel sets US or EU residency at project creation. Decide before you send the first event.
• Comparing raw totals across tools. Ad blockers block many analytics scripts. Developer audiences block more than most, so expect lower numbers from client-side scripts on developer sites.
• No naming convention. Pick object_action (for example signup_completed) before the first event ships.
• Self-hosting with default credentials. Change Umami's admin / umami login and keep Matomo and PostHog patched; an analytics server holds visitor data.

For a broader list of options, browse analytics tools on DevHunt or the free alternatives to Google Analytics and to Mixpanel. New to the topic? Start with Web analytics 101.

FAQ

What is the best free web analytics service?

For traffic only, Google Analytics 4 and Cloudflare Web Analytics are both free. Cloudflare is simpler and cookie-free, while GA4 is more powerful. If you can self-host, Matomo On-Premise, Plausible Community Edition and Umami are free open-source options. For product analytics, Mixpanel and PostHog both have free tiers of 1M events a month and Amplitude's free plan covers 2M (as of November 2026).

Which web analytics services don't use cookies?

Plausible, Fathom, Umami, Cloudflare Web Analytics and Vercel Web Analytics all work without cookies by default. Matomo can be configured to run cookie-less, and PostHog has a cookieless mode. GA4 uses first-party cookies by default, and so does Microsoft Clarity.

What's the difference between web analytics and product analytics?

Web analytics measures traffic: visitors, pageviews, sources and pages. Product analytics tracks what identified users do over time (funnels, retention, cohorts) and is event-based. GA4 sits in between. Mixpanel, Amplitude and PostHog are product analytics tools that also cover basic traffic.

Is Google Analytics 4 really free?

Yes, the standard version is free. The paid tier, Analytics 360, raises limits such as data retention (up to 50 months instead of 14), sampling thresholds (1 billion events per query instead of 10 million) and BigQuery export volume. Google sells it through a sales contract.

Can I self-host web analytics?

Yes. Matomo On-Premise, Plausible Community Edition and Umami are free to self-host, and PostHog has a self-hosted Docker Compose option that it doesn't support and recommends only for small volumes (under about 300k events). You take on running the server, backups and upgrades yourself.

Who are the leading web analytics service providers?

By usage, Google Analytics leads by far: W3Techs finds it on 46.8% of all websites as of November 2026. Among the alternatives, Matomo is the most used open-source option (1.4% of websites), Adobe Analytics and Analytics 360 serve enterprises, Mixpanel, Amplitude and PostHog are the best-known product analytics services, and Plausible, Fathom and Umami are the best-known privacy-first options.

Which web analytics service includes performance monitoring?

Cloudflare Web Analytics reports LCP, INP and CLS for free from the same script that counts visits. Vercel offers Speed Insights alongside Web Analytics (full Core Web Vitals need the Plus tier at $10 per project per month on Pro), and PostHog captures web vitals as regular events. GA4 can record them if you send them with Google's web-vitals library.

What are my options for reliable web performance analytics?

For real-user data, use an analytics service that collects Core Web Vitals (Cloudflare, Vercel Speed Insights, PostHog) or send them to GA4 yourself. For lab testing, use PageSpeed Insights and Lighthouse. Use both: lab tests to catch problems before release, real-user data to see what visitors actually get.

Is Vercel Analytics better than Google Analytics?

It depends on the job. Vercel Web Analytics is simpler, cookie-free and built into Vercel deployments, but keeps only 1 month of data on Hobby and works only on Vercel. Google Analytics is free on any host, keeps up to 14 months, and adds conversions, audiences and Google Ads links. Pick Vercel for clean traffic numbers, GA4 for marketing attribution.

Should I use Vercel Analytics or PostHog?

Use Vercel Analytics if you want pageviews and referrers for a Vercel-hosted site with minimal setup. Use PostHog if you need per-user funnels, session replay, feature flags or A/B tests, or if you host elsewhere. PostHog's free tier (1M events a month) is larger than Vercel Hobby's 50,000 events.

How much does website analytics cost?

Anywhere from $0 to thousands a month. GA4, Cloudflare and Microsoft Clarity are free. Privacy-first services start at $9 (Plausible), $15 (Fathom) or $20 (Umami Pro) a month. Event-priced services are free up to 1M or 2M events, then scale with volume: at 5M events Mixpanel Growth costs $613 a month billed monthly. Enterprise suites like Adobe Analytics and Analytics 360 are quoted by sales.

Which web analytics services let me export raw data?

GA4 exports raw events to BigQuery (1M events a day on the daily export for standard properties, unlimited streaming at $0.05 per GB). PostHog has batch exports to S3, BigQuery, Snowflake, Postgres, Redshift, Databricks and Azure. Mixpanel has warehouse connectors on Growth, Plausible offers raw exports on Enterprise, and self-hosted Matomo, Umami and Plausible CE keep the raw data in your own database.