How to migrate from Google Analytics to MetricMind

What you gain

• No cookies, and an option to store nothing on the visitor's device.
• Web Vitals, JavaScript errors, rage and dead clicks, contacts and journeys, from the same script.
• Numbers that are measured, not sampled or modeled, with the query behind every answer.
• An AI analyst you can ask anything, watching your data every morning for real changes. Investigations come with Business and Scale.
• Event history kept for as long as your account is active.

What you give up

• Google Ads audiences, remarketing and key events imported as conversions.
• Search Console data in your reports, which we don't have yet.
• BigQuery export and the Looker Studio connector. You get CSV export, a query API and MCP instead.
• City-level location and demographics. MetricMind stops at country for now.
• Your GA4 history inside the new tool, because we can't import it yet.

Step 1: Get access and add your site

MetricMind is in private beta, so join the waitlist first. Once you're in, add your domain: the analyst reads your public site to learn what your business does, and you get your site ID and install snippet. Your GA4 setup doesn't change.

Step 2: Add the script next to your GA4 tag

Paste one line into the <head> of every page and leave Google's tag in place. The setup page shows live when the first event arrives. For Next.js there's a snippet under Settings → Tracking, or you can give your coding agent the one-line install prompt.

Tag Manager: the script reads its settings from its own tag, so a Custom HTML tag should work, but we haven't tested it. Check that the setup page sees your first event, or add the script to your page template instead.

<script src="https://cdn.metricmind.si/mm.js" data-site="YOUR_SITE_ID" defer></script>

Step 3: Keep your history

There's no GA4 import yet. Keep the GA property so you can go back to it, and download any reports you want to keep. Google's retention setting of 2 or 14 months applies to event-level data in explorations and funnel reports, while standard reports aren't affected. If you've moved from Plausible, PostHog or Simple Analytics before, those histories can be imported under Settings → Data import.

Step 4: Look at what's already there

Before recreating anything, open the dashboard. Pageviews, sources, UTM campaigns, ad clicks by network, countries, devices, scroll depth, outbound clicks, downloads, site search, forms, Web Vitals, errors and contact clicks are collected automatically, so much of your GA4 event setup may no longer be needed.

Step 5: Recreate your events and goals

In GA4 you collect an event and mark it as a key event. In MetricMind you track an event, or use a page, and make it a goal. A goal is either a page path such as /thank-you, which needs no code, or an event name. You can keep your existing event names, so reports stay familiar.

Without writing JavaScript, a data-track attribute on a button does the same job. Use mm.identify() where you set a GA4 user ID, and report revenue with the event itself or by connecting Stripe or Paddle.

// GA4
gtag("event", "sign_up", { method: "google" });

// MetricMind
window.mm?.track("sign_up", { method: "google" });

// Or with no JavaScript
<button data-track="sign_up">Create account</button>

Step 6: Rebuild what your team looks at

• Explorations: a funnel of up to 8 steps in the funnel builder, or simply ask the analyst.
• Saved reports: pin panels and answers to your board, which re-runs them for any date range.
• Audiences you only use for analysis: saved segments, one click from any report.
• Internal traffic: exclude your office address under Settings → Security.
• Weekly reports: your own alerts, the morning briefing and, on Business and Scale, scheduled questions.

Step 7: Compare, then remove GA

Run both for a few weeks, long enough to cover a normal week and a campaign. Compare trends and proportions rather than exact totals. When you're happy, remove the Google tag from your pages or from Tag Manager.

If GA4 was the only reason for your consent banner, ask your counsel whether you still need one. The free AI visibility check lists the trackers left in your HTML, though not those loaded through Tag Manager.

Why your numbers will look different

The two tools will never match exactly, and we won't promise a typical gap, because it depends on your audience and your consent setup. Sessions aren't the usual cause: both end a session after 30 idle minutes. These are:

• Consent. If your banner blocks GA4 until visitors accept, MetricMind counts visitors GA4 never sees. GA4 may then estimate some of them through behavioral modeling.
• Identity. GA4 uses cookies, MetricMind a random ID in the site's storage. In storage-free mode a returning visitor counts as new each day, so unique visitors run higher over long ranges.
• Blockers and bots. Each tool has its own bot filter, and browser extensions block each tool differently.
• Bounces. GA4 counts a session as bounced if it lasted 10 seconds or less, with one page and no key event. MetricMind counts one page, under 10 seconds of engaged time and nothing done. Similar, not identical.
• Scroll. GA4's scroll event fires when someone reaches the bottom of a page. MetricMind reports depth from 0 to 100%.

GA4 to MetricMind: feature reference

Sources

• MetricMind vs Google Analytics 4: the full comparison
• MetricMind docs: install, events, users and options
• Google: key events
• Google: engaged sessions and bounce rate
• Google: sessions and the 30-minute timeout
• Google: automatically collected events
• Google: behavioral modeling for consent mode
• Google: data retention