Migrating to GA4 is not a flick of a switch, it is a project with a beginning, a middle, and a clear end. Get the order wrong and you end up with gaps in your reports, duplicated conversions, or historical data you can no longer open.
This guide gives you the checklist in the order you will actually use it: audit what you have, build the new property, run both tools side by side, validate the numbers, set up Consent Mode v2, archive your old data, and monitor after launch. Each section answers one question and tells you what to decide before moving on.
The path in 7 steps
The steps of this page, in order.
-
01 Audit what you have List every UA property, view, goal, and filter you rely on today.
-
02 Build the GA4 property Create the property and data streams, and install the new tag.
-
03 Run both in parallel Keep UA and GA4 collecting at the same time to compare and catch gaps.
-
04 Rebuild goals as events Translate each UA goal into a GA4 event and mark the key ones as conversions.
-
05 Validate the data Check numbers before, during, and after the switch against known benchmarks.
-
06 Set up Consent Mode v2 Wire consent signals so you do not lose data from users who decline.
-
07 Archive and monitor Export your UA history, then watch GA4 daily for the first weeks.
01 / 10
What is the difference between UA and GA4?
GA4 records events instead of sessions, so your old reports will not map one-to-one to the new ones.
In Universal Analytics (UA), the old Google Analytics that tracked sessions and pageviews and no longer collects new data, activity is grouped into sessions. GA4, the current version, records each action as an event, a single action like a click or a purchase.
Because of that, goals, funnels, and bounce rate are defined differently, so your old reports will not map one-to-one. Start with a pre-migration audit of your UA property to list every goal, filter, and custom dimension you rely on.
Core concepts: UA vs GA4
| Concept | Universal Analytics | GA4 |
|---|---|---|
| Unit of tracking | Session | Event |
| Pageview | A hit type | An event with parameter page_view |
| Goal or conversion | Goal (destination, duration, pages/session, event) | Conversion (any event marked as conversion) |
| Bounce rate | % of sessions with only one pageview | % of sessions with no engagement (not a true bounce by default) |
Comparison based on Google Analytics documentation (as of November 2026).
02 / 10
What should I check before I start the migration?
Before touching anything, write down every UA property, goal, and report your team actually uses.
A property is the container in Google Analytics that holds data for one website or app, and each one can have several views with different filters. Screenshot the admin list and note the account owner, editors, and read-only users, because accounts get locked when the owner leaves.
Export your goal list and the ten reports your team opens most. If Google Tag Manager is already installed, the upgrade is simpler: confirm the container ID, then identify which consent tool is running, because that setting must be reproduced in GA4.
List every UA property and view
Open Admin, then Account Settings and Property Settings, and record each ID, site, and view. Views hold the filters and goals you cannot recover after sunset.
Export your goal list
Go to Admin > Goals and download the table as CSV. Each goal name, type, and destination URL becomes your build spec for GA4 events.
Review account access
Open Admin > Account Access Management and list every user with their role. Remove departed contractors now, so the GA4 property is created with the right owners.
Confirm Google Tag Manager
Check whether the site loads from a GTM container and note the container ID. If yes, your GA4 tags will be added there rather than hard-coded on the site.
Identify your consent tool
Record the consent management platform (the banner that collects cookie choices) and where it fires. GA4 consent settings must mirror that behavior to stay compliant.
Save benchmark numbers
Export a 12-month report of sessions, users, and conversions per channel. These numbers are your only baseline for spotting gaps after the switch.
Going further
Show export steps
Method
How to export a UA report before you lose access
Download the reports you will need as CSV or PDF while the old interface still answers.
Open Universal Analytics, go to the report you want, and set the date range to cover the full period you need to keep. Use the Export button above the chart and choose CSV for data you may want to sort later, or PDF for a fixed snapshot your team can read.
For anything with more than a few thousand rows, switch to the Google Sheets export so you get every line instead of the preview limit. Repeat the download for each view, because a filtered view can reshape the numbers you see.
- Date range covers the full historySet the start date back far enough to capture at least one complete year of seasonality.
- File is saved outside Google AnalyticsStore the CSV or PDF on a shared drive, because access to UA ends when the property is deleted.
- Row count matches the interfaceCompare the exported total with the number shown on screen to catch sampling.
- File name includes view and dateName each export with the view, the date range, and the day you pulled it, so you can tell versions apart later.
03 / 10
How do I create the GA4 property and data stream?
Create the property, add a web data stream, and install the Google tag, ideally through Google Tag Manager.
Most small businesses only need one GA4 property per website. It keeps reporting simple and avoids splitting sessions across duplicates. Set it up in the admin panel, then add a web data stream, the connection between your site and the property.
Keep your old Universal Analytics tag running for now. Install GA4 through Google Tag Manager: add a GA4 configuration tag fired on all pages, use Preview to confirm it fires once per page, then publish.
-
Open Admin
In Google Analytics, go to Admin and select Create Property.
-
Name the property
Use a clear name such as 'Main website, GA4', plus your reporting time zone and currency.
-
Create a web data stream
Enter your website URL to generate the stream and its Measurement ID (a code like G-XXXXXXX).
-
Add the GA4 configuration tag
In Google Tag Manager, create a Google tag using that Measurement ID, triggered on All Pages.
-
Preview, then publish
Use Tag Assistant Preview to confirm the tag fires, then publish the container.
04 / 10
Should I run UA and GA4 at the same time?
Yes. Parallel tracking is the only way to compare old and new numbers while both sources are still reliable.
Parallel tracking means running your old Universal Analytics tag and your new GA4 tag on the same pages. Keep both live for at least four to eight weeks, so you can spot gaps while UA still reports. This overlap is your only reference point before UA data goes away.
Compare key metrics side by side each week, like purchases and form submissions, and note the natural gap between the two models. Use those comparisons to fix tagging problems while you still have a baseline. Once you turn UA off, that safety net is gone.
Why run both tags
Parallel tracking keeps your old and new numbers side by side, so you can catch tracking gaps before UA disappears. It is the only reliable way to validate GA4 against a known reference.
- Purpose
- Compare old vs new
- Risk if skipped
- Hidden data gaps
Key pointLeave the UA tag active while you validate GA4.
How long to run it
Four to eight weeks gives you enough data to see weekly patterns and seasonal shifts. Shorter windows can miss rare events like large purchases or form completions.
- Minimum
- 4 weeks
- Recommended
- 4–8 weeks
Key pointSet a calendar reminder before UA data is removed.
What to compare weekly
Check purchases, form submissions, and other key conversions in both tools. Differences are normal, but a growing gap means a tagging issue you can still fix.
- Metrics
- Purchases, forms, conversions
- Cadence
- Weekly review
Key pointLog every discrepancy and fix tags while UA is live.
05 / 10
How do I move my goals to GA4 events?
Each UA goal becomes a GA4 event, and only the events that matter to the business should be marked as conversions.
An event is a single action a user takes, a click, a form submit, a purchase, that GA4 records. Start by listing every UA goal and giving it one event name you will keep stable for years, so reports and audiences do not break when marketing asks for a change.
Mark only events tied to revenue or a lead as conversions; leave the rest unmarked to keep reports clean. Send each event through GTM's preview mode and check it in GA4's DebugView before you trust it.
UA goal to GA4 event mapping
| UA goal | GA4 event and conversion status |
|---|---|
| Newsletter signup | generate_lead, conversion |
| Contact form submit | form_submit, conversion |
| Purchase | purchase, conversion |
| PDF download | file_download, no conversion |
| Outbound link click | outbound_click, no conversion |
Example mapping for a small business. Event names are illustrative; keep your own naming convention consistent.
06 / 10
How do I know the data is correct before I switch off UA?
Compare a handful of known numbers, traffic, sign-ups, revenue, against a fixed reference before you retire the old tag.
Pick one full week that both tools tracked, then pull the same three numbers from GA4, UA, and your own back-end records: sessions, sign-ups, and revenue. A gap of a few percent is normal, because the two tools count sessions and attribute credit differently. A gap of 30 percent or more points to a tagging bug, not a reporting quirk.
Write the result in a validation log: the date, the numbers, the difference, and what you changed. That log shows what changed and when, which matters before you turn the old tag off. Note also your data retention setting, how long Google keeps event-level data before it is removed.
Before the switch: baseline week
Export one week of sessions, conversions, and revenue from UA and your back-end into a single sheet. This is the reference every later number is compared against.
Before the switch: GA4 totals match
Pull the same week in GA4 and compare totals. A few percent of difference is expected; 30 percent means a tagging bug to fix before anything else.
During: debug view on key events
Watch the events fire in real time for one sign-up and one purchase. Confirm each fires once, on the right page, with the value you expect.
During: currency and value check
Open a recent transaction and confirm the currency code and value match your store. A wrong currency inflates or deflates revenue without any visible error.
After: 48-hour reconciliation
Re-run the comparison two days after the switch and log the result. If conversions dropped to zero on one channel, a trigger was lost in the move.
Going further
See the five checks
Method
The five checks that catch a corrupted tracking setup
Five checks that expose a broken tracking setup before it costs you a quarter of clean data.
A corrupted setup rarely throws an error. It quietly doubles your purchases, drops your form submissions, or reports revenue in the wrong currency, and you only notice weeks later when the numbers stop making sense. Each check below targets one of those silent failures.
Run them in order: duplicate events first, then missing conversions, then currency and value, then traffic sources, then retention. Duplicates skew every average downstream, so they come first. If a check fails, fix it and re-run the same comparison rather than moving to the next one.
Assume a five-step purchase path and a test transaction of $120. If GA4 logs two purchase events for it, your reported revenue for the period reads $240. That single duplicate is enough to make a good month look like a record one.
- Duplicate eventsFire one test conversion and confirm it appears exactly once in the real-time report. Two entries mean a tag and a GTM trigger are both sending the same event.
- Missing conversionsCompare your main conversion count in GA4 with the same week in your back end. A conversion that exists in your records but not in GA4 lost its trigger during the move.
- Wrong currencyCheck the currency code on a recent transaction against your store settings. A mismatch changes the revenue figure without raising any error.
- Traffic source mismatchCompare sessions by channel for the baseline week in both tools. Paid and organic should rank in the same order; a channel that collapsed points to a broken campaign tag.
07 / 10
Do I need Consent Mode v2 for GA4?
If you show ads or collect data from EEA users, Consent Mode v2 is expected and skipping it can cost you data and ad performance.
Consent Mode v2 (a Google feature that adjusts tag behavior based on each visitor's consent choices) doesn't block everything by default. Instead, your tags fire with a consent state attached, and Google applies modeling to fill gaps in measurement. That's why a broken banner or missing signal is one of the most common causes of data loss after a migration, not a tracking bug.
This is a compliance topic, not legal advice. Check the current official Google requirements and your own legal obligations for EEA and UK visitors, then apply them to your setup with your consent management platform or Google Tag Manager.
Banner connected
Confirm your consent banner or CMP is linked to Consent Mode v2 and can send updates when a visitor changes their choice.
Signals firing
In Google Tag Manager preview mode, verify that consent signals fire before your GA4 tags and that all four consent types are present.
Default state set
Set a default consent state before any tag loads, so tags know what to do for visitors who haven't interacted with the banner yet.
Test from a clean browser
Open your site in a private window with no cookies, accept or reject, and confirm the tags behave as expected in DebugView.
Going further
How it works
Concept
What Consent Mode v2 actually changes in your tags
Basic mode withholds all data when consent is denied; advanced mode sends cookieless pings so Google can model the gaps.
Basic consent mode is all-or-nothing: if a visitor rejects cookies, your Google tags don't fire at all. You see nothing from those users. It's simpler to set up but leaves a bigger hole in your reports.
Advanced consent mode lets tags send cookieless pings even when consent is denied. Google then uses conversion modeling to estimate what those users would have done. You get more complete numbers, but they are estimates, not raw data.
Both modes need the same thing from you: a banner that passes the visitor's choice as a consent signal. Advanced mode just adds the cookieless measurement layer. For a small team, advanced mode is usually worth the extra setup because it reduces the reporting gap without new tools.
- Default consent before tagsEnsure the default state is set in GTM before any GA4 tag fires.
- Four consent typesCheck that ad_storage, analytics_storage, ad_user_data, and ad_personalization are all passed.
- Update on choiceVerify the banner triggers a consent update when a visitor accepts or rejects.
- DebugView testTest accept, reject, and no-interaction paths in a private window.
08 / 10
What happens to my old Universal Analytics data?
UA stopped receiving data, and access to the old interface will not last forever, export what you need now.
Google stopped collecting data in Universal Analytics in 2023 and has since withdrawn access to UA properties. What survives is what you exported yourself: report files, spreadsheets, or a UA dataset in BigQuery if your property had that export switched on. Check your shared drive and your Google Cloud project before buying any migration tool.
GA4 cannot import UA history: the two data models record sessions and events differently, so treat old reports as reference documents, not live data. Store exports with a name like UA-2023-traffic-monthly.csv and note each file's date range and metric definitions. This matters most when a tool asks for last year's totals and you no longer have a dashboard to open.
-
Find what still exists
Search your shared drive, email attachments, and Google Cloud project for any UA exports made while the properties were still open.
-
Export or reconstruct the reports you need
For reports you can still open, download CSVs for traffic sources, top landing pages, conversions, and audience overview. Where access is gone, rebuild quarterly totals from a BigQuery export if your property had one, or from a third-party tool that imported UA data before the sunset.
-
Document where the files live
Create one index file listing each export, its date range, its metric definitions, and the folder path or BigQuery table name. Share it with anyone who reports on marketing performance, so nobody hunts through drives next quarter.
09 / 10
What should I check in the first weeks after the switch?
Treat the first month as a monitoring period, with one named owner checking the numbers each week.
Weekly, compare GA4 against your back-end records: orders, sign-ups, revenue. Pick one person as owner and log every fix, what broke, what you changed, when. Set the data retention period, meaning how long Google keeps event-level data before it is removed, to the longest option your account offers.
Expect a dip right after the switch: consent banners, redirects, and cross-domain tracking are the usual culprits. If validation still fails after you fix the obvious issues, book a few hours with an outside specialist.
Weekly owner tasks
One named person compares GA4 to the back end every week. They log each discrepancy and each fix in a shared document.
- Cadence
- Weekly, same day
- Evidence
- Back-end orders and sign-ups
Key pointBook 30 minutes every Monday and close the loop.
Warning signs
A sudden traffic drop, missing conversions, or a gap between GA4 and back-end numbers that keeps growing. These point to broken tags, consent, or filters.
- First check
- Tag firing and consent state
- Second check
- Internal traffic filters
Key pointFix the obvious issue, then re-measure for one week.
When to call a specialist
Validation still fails after you have fixed tags, consent, and filters, or the numbers are off by a margin you cannot explain. A few hours of expert time usually cost less than weeks of blind guessing.
- Trigger
- Repeat failure after fixes
- Scope
- Audit and repair, not rebuild
Key pointPrepare your log and back-end samples before the call.
10 / 10










