How to Migrate From Google Analytics to Plausible: A Safe Playbook

Google Analytics 4 is built for ad attribution and user profiling, and most business websites need neither. Plausible is a lighter, cookieless alternative that covers what most teams actually check: traffic, sources, pages, and conversions. The migration takes an afternoon for a simple site if you follow the right order: run both trackers in parallel, import your GA4 history before removing anything, rebuild your goals, and only then cut over. This guide covers that sequence, the real costs of cloud versus self-hosted, and the cases where you should not migrate at all.

What you keep and what you leave behind

Plausible is not a smaller GA4. It makes a different trade: less data, deliberately, in exchange for simplicity and no consent banner. Plausible's own migration guide is blunt about both sides.

What you gain:

• One dashboard. Traffic, sources, pages, locations, devices, and goals on a single page. No custom reports for the basics.
• No cookie consent banner. Plausible uses no cookies and no persistent identifiers, so in most jurisdictions you can drop the analytics consent prompt. Fewer banners means fewer visitors lost to declined consent.
• Real numbers instead of modeled ones. GA4 fills consent-related gaps with estimates. Plausible counts what it sees and models nothing.
• A script that survives ad blockers better. The Plausible script is under 1 KB (GA4's tag is 45 KB or more) and is blocked far less often, so more of your actual traffic shows up.

What you leave behind:

• User-level reports and identity-stitched journeys. Plausible does not track individuals at all.
• Cohort and retention analysis. Not available.
• Retargeting audiences. Plausible cannot export audiences to Google Ads.
• Multi-touch attribution. Plausible reports last touch only.

If you are still weighing the trade-off, our Plausible vs Google Analytics cost and privacy breakdown goes deeper on the comparison. The rest of this post assumes you have decided to move.

The migration sequence

The order matters more than any single step. Import before you delete, overlap before you cut over, and never judge the new tool by day-one numbers.

Step 1: Run both trackers in parallel

Add the Plausible script to your site next to the existing GA4 tag and leave both running for two to four weeks. Plausible's script adds no meaningful page-speed cost, and the overlap period does three things: it lets you verify your goals fire correctly, it gives you a baseline for comparing the two tools' numbers, and it gives stakeholders time to get used to the new dashboard before the old one disappears.

You can start on Plausible's 30-day trial, so this step costs nothing.

Step 2: Import your GA4 history before you remove anything

Plausible has a built-in GA4 importer, documented in the official import guide. You link your Google account in site settings, pick a property, and confirm. Most imports finish in minutes; large properties can take a couple of hours because of Google's API limits. More than 15,000 sites have used it.

Know the limits before you rely on it:

• Aggregates, not events. The import brings in per-date totals: pageviews, visitors, sources, countries, devices, pages. It does not bring events, user demographics, or behaviour flows.
• Whole properties only. If one GA4 property tracked several sites, all of that data lands in the single Plausible site you import into. You cannot split it by hostname.
• Five properties maximum per Plausible dashboard. If you import several, do it oldest first: each later import takes the longest date range that does not overlap what is already there, so the order decides which property's data wins any overlap.
• No gap filling. The import covers the period from your first GA visitor to your first Plausible visitor, with the range adjusted to avoid double-counting. Traffic after Plausible started collecting is not backfilled.
• GA4 retention limits apply to detail, not totals. Google's data retention documentation sets user-level and event-level retention at 2 or 14 months on standard properties. The importer pulls aggregated reports through the Google Analytics Data API, so older totals often survive past that window, but if you want the event-level history for your own records, export it from GA4 before you migrate.

Do the import while GA4 is still live. Once you remove the tag, whatever you did not import or export is gone the day your retention window closes.

Step 3: Recreate goals, funnels and events

GA4 conversions do not transfer. In Plausible you rebuild them as either pageview goals (a visit to a thank-you page) or custom events (form submit, signup, download) fired from your tag manager or a small snippet.

A practical approach: list every GA4 key event you actually look at in reports, not the ones that exist. Most sites end up with five to ten goals that matter. Recreate those, verify them against the parallel GA4 data during the overlap weeks, and ignore the rest.

One constraint to know before you choose where to host: marketing funnels and ecommerce revenue goals are not in the self-hosted Community Edition at all, and on Plausible Cloud they sit behind the Business plan ($19 a month at 10,000 pageviews, $39 at 100,000). If funnel analysis matters to you, budget for Business or accept simpler goal tracking.

Step 4: Connect Search Console and rebuild reporting habits

Connect Google Search Console in Plausible's site settings to bring search keyword data into the dashboard. Then recreate whatever reporting rhythm your team had: the weekly email, the Slack summary, the CSV export for the board deck. Plausible supports scheduled email reports, a stats API, and CSV export, so most reporting workflows map across. The people who consumed GA4 reports care about the numbers arriving, not which tool produced them.

Why the numbers will not match

During the parallel period, Plausible and GA4 will disagree by 10 to 20 percent. This is normal and expected, not a sign one tool is broken. The reasons, per Plausible's documentation:

• Consent modeling. GA4 estimates data for visitors who decline cookies. Plausible has no cookies to decline, so it counts everyone the script reaches.
• Bot traffic. GA4 counts a share of bot and spam traffic that Plausible filters out.
• Counting method. GA4 identifies visitors with cookies. Plausible generates a daily identifier from the IP address and User-Agent, specific to one site and one device, and stores neither the raw IP nor the User-Agent, per its data policy. The same person on Monday and Tuesday counts as two unique visitors in Plausible.
• Ad blockers. GA4's tag is on every blocklist. Plausible's is blocked far less, so Plausible frequently shows more traffic on tech-heavy audiences.

Use the overlap period to establish the ratio between the two tools for your site. Then stop comparing. The point of the migration is a consistent, complete series going forward, not a perfect reconciliation with the past.

Cloud or self-hosted: the cost and feature trade-off

Plausible runs as a managed cloud service or as the free, AGPL-licensed Community Edition (CE) you host yourself. The code is the same; the feature and operations split is not.

A worked example for a business site at about 100,000 pageviews a month:

• Plausible Cloud: roughly $19 a month at that volume on the entry plan, with plans starting at $9 a month for 10,000 pageviews. Cloud caps how many sites a plan covers (one on Starter, three on Growth, ten on Business), which matters for agencies.
• Self-hosted CE on a DigitalOcean droplet: the app runs Plausible plus PostgreSQL plus ClickHouse, and the official Docker Compose setup wants at least 2 GB of RAM, with 4 GB a comfortable margin. DigitalOcean's droplet pricing puts that at $12 to $24 a month (2 GB at $12, 4 GB at $24, 80 GB SSD on the latter). ClickHouse compresses well, roughly 1 GB of disk per 10 million pageviews, so 80 GB holds years of history. CPU must support SSE 4.2, which any modern VPS does.

So at 100,000 pageviews the bills are similar, and the decision is not about the hosting fee. It is about these differences:

• Feature parity. CE lacks marketing funnels, user journeys, ecommerce revenue goals, SSO, and the sites API. Cloud has all of them.
• Release cadence. CE is a long-term release published twice a year. Cloud ships improvements continuously.
• Bot filtering. Cloud excludes around 32,000 data-center IP ranges plus user-agent and referrer-spam filtering. CE does basic user-agent and referrer-spam filtering only, so self-hosted numbers run slightly dirtier.
• Operations. Self-hosting means you own upgrades, backups, uptime, and capacity. That is real work, though modest work for a single instance.
• Data location. Cloud processes visitor data on EU-owned infrastructure in Germany. Self-hosted, data lives wherever your server is, which is exactly why some organisations choose it: full control of analytics data under data protection rules like Kenya's Data Protection Act.

The rough crossover where self-hosting clearly wins on cost sits around 250,000 pageviews a month, or earlier if you run many sites, because CE places no cap on sites. If you want the server-side detail, our guide to self-hosting Plausible with real setup and cost numbers covers sizing, and the Plausible vs Matomo comparison is worth a read if you are also evaluating heavier self-hosted options.

When migrating to Plausible is the wrong choice

Be honest about fit before you start:

• You run significant paid acquisition with retargeting. GA4's audience export to Google Ads has no Plausible equivalent. Keep GA4 for the ads pipeline even if you add Plausible for everything else.
• You need multi-touch attribution. Plausible is last-touch only. If budget decisions depend on attribution modeling, look at a dedicated attribution tool instead.
• You are measuring a product, not a website. Cohorts, retention curves, and user-level journeys are product analytics. Pair Plausible with PostHog or similar rather than expecting it to replace them.
• Your organisation requires Google ecosystem integration. BigQuery export, Data Studio pipelines, and deep Google Ads reporting are GA4 territory.
• You cannot get sign-off on the number shift. Some boards and agencies treat the GA4 series as canonical. If a 10 to 20 percent discontinuity will trigger a political fight you cannot win, run both permanently or not at all.

Common migration mistakes

• Deleting the GA4 tag before importing. The importer only covers history up to your first Plausible visitor. Remove GA4 first and the gap is unrecoverable.
• Judging Plausible by week one. Numbers will differ from GA4 by design. Calibrate during the parallel period, then move on.
• Importing a multi-site property without noticing. The import takes the whole GA4 property. If it tracked your marketing site and your app, both land in one Plausible site.
• Expecting events and funnels in the import. Only aggregate daily totals come across. Historical custom events, demographics, and flows do not.
• Recreating every GA4 event. Most GA4 setups carry dozens of unused events. Rebuild the five to ten goals someone reads.
• Self-hosting for funnels. Funnels and revenue goals are Cloud-only, and on Cloud they need the Business plan. Teams regularly discover this after standing up a server.

Frequently asked questions

Can I import historical data from Google Analytics into Plausible? Yes, from GA4 properties via the built-in importer. You get per-date aggregates for pageviews, visitors, sources, countries, devices, and pages, up to five properties per dashboard. Custom events, demographics, and behaviour flows do not transfer.

Will my Plausible numbers match GA4? No, and they are not supposed to. Expect a 10 to 20 percent difference from consent modeling, bot filtering, cookie versus IP-based visitor counting, and ad-blocker rates. Run both in parallel for a few weeks to learn your site's ratio, then stop comparing.

Does Plausible need a cookie consent banner? No. Plausible sets no cookies and stores no persistent identifiers or raw IP addresses, so analytics-specific consent prompts are generally not required. Your site's other trackers may still need one.

How much does it cost to self-host Plausible? The software is free under AGPL. A DigitalOcean droplet with enough RAM for Plausible, PostgreSQL, and ClickHouse costs $12 to $24 a month, and disk usage is roughly 1 GB per 10 million pageviews. Budget your own time for upgrades, backups, and monitoring.

Can I run Plausible and Google Analytics together? Yes, and you should during migration. Both scripts coexist with no meaningful speed impact. Some teams keep GA4 permanently for Google Ads audiences while using Plausible as the reporting dashboard.

Where to go from here

The migration itself is a day of work for most sites; the discipline is in the sequence and in setting expectations about the numbers. If you would rather not run the infrastructure side, our managed open-source stack service deploys and operates Plausible alongside tools like Twenty, Chatwoot, and Listmonk, with upgrades, backups, and monitoring handled for you. To talk through whether cloud, self-hosted, or managed fits your situation, book a call with an engineer.