Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Marketing Analytics › Google Analytics 4

Moving to a New GA4 Property

How to move to a new GA4 property when the old one cannot be kept: run both side by side, move links, conversions and reports, then mark the break.
Tutorialintermediate13 min readUpdated

When a new property is the right answer

Replacing a property is the last of three routes, and the order matters because each earlier route keeps something the later ones lose. The first is a transfer. The GA4 property handover checklist secures administrator rights for a named person at the business and moves ownership, users, links and documentation from one owner to the next, so history, audiences and imported conversions stay where they are. The second is repair. When the business already owns the property but its key events, definitions and rules are a muddle, the clean-up tutorial for an inherited property fixes them in a safe order without discarding anything that still works.

A new property is right only when neither is possible. The situations that lead here are about ownership rather than quality: the property sits inside an agency's account beside other clients and the agency will not move it; the only administrator has left and cannot be reached; the business has been sold or split and the property stays with the other party; or the property was created under a personal login nobody can recover. The setup tutorial warns that property ownership is hard to move later, which is how these situations arise, and the tutorial for taking over someone else's marketing prefers a transfer for the same reason: a new account starts without the history the old one built.

In the terms of the framework for reversible and irreversible decisions, the move is a one-way door with a two-way door in front of it. Creating the new property is easy to reverse; switching off the old feed is not. So, as general practice for this move, the steps keep the old property collecting until the very end, and that last step is taken only on evidence.

What moves across and what starts again

Nothing in this tutorial copies an object from one property to the other. Every object is rebuilt, and the new property starts with whatever the site's guides say a fresh object starts with.

ObjectIn the new propertyWhat the site's guides say
Event history and reportsBegins on the day the new tags first sendHistory stays in the old property, where explorations reach back only as far as its retention setting
BigQuery exportBegins with the first full day after the new linkThe export holds nothing from before its link and nothing earlier is backfilled
Key eventsCreated again by name, with counting method and valueChanges act from the day they are made and do not recalculate the past
Custom definitionsRegistered againRegistration is not retroactive, so each starts empty
Interface rulesRepeated, or better replaced by a tag fixRules belong to the property they were saved in
AudiencesBuilt againNot retroactive; plan as though each new list starts empty
Data filters and internal trafficDefined again, tested, then activatedA filter applies only from the moment it is activated, and what it discards is gone
Google Ads conversionsNew conversion actions from a new importPrimary or secondary status decides what bidding uses
Looker Studio reportsNew data sources on the new propertyCharts inherit their source's fields, so each field is chosen again

The guide behind each row is linked from the step that handles it.

Before you start

  • A written decision, kept as general practice, that transfer and repair are not possible: what was asked, of whom, when, and the answer. A later owner will need it.
  • Access to the old property while it lasts, at least enough to read and export reports. If the other party is removing access, the copying step comes before everything else.
  • The tracking plan, or the nearest thing the business has to one, listing every event, parameter, key event and audience the business relies on. The tracking plan guide describes a complete one.
  • Publish rights on the container (GTM) and administrator rights on the business's own Analytics account, Google Ads account and BigQuery project. If the container also sits in the other party's account, export it, import it into a container in the business's own account, replace the old container's snippet on the site with the new one's, and check that the old property still receives every event before the parallel run starts.
  • A reader list: every Google Ads conversion action and audience, Looker Studio report, saved exploration and export query that reads the old property, and the people who receive reports from it.
  • The business's own daily counts of orders, bookings or leads, so the two properties can be compared with something neither of them produced.

The steps

  1. Confirm the old property can be neither transferred nor kept

    Ask whoever holds the account, in writing, either to move the property into an account the business owns or to give a named person at the business administrator rights. If the business turns out to own the property already, stop here and clean it up instead. Record each request and its answer in the decision note. When the refusal is part of a wider dispute with a former agency, the diagnosis for an agency that will not hand over accounts covers the contract and the other accounts caught up in it; the analytics property is one line on that list.

  2. Copy what only the old property holds

    History cannot be moved, so copy the parts the business will want to read later. Export the standard reports used in monthly and annual reviews at the grain they are read, the results of the explorations people rely on, and the configuration: key events with their counting methods, custom definitions with their scopes, audiences with their conditions and durations, filters, referral exclusions and the cross-domain list. The guide to keeping copies of marketing tool data treats event history as something the business keeps its own copy of, and this is the moment that copy is needed. If the old export runs into a project the other party pays for, take what you need from it now, because the handover checklist warns that the export, and the history it held, can go when that project closes.

  3. Create the new property in the business's own account

    Create the property under the account the business controls and work through the GA4 setup tutorial for every setting: a web Data Stream per site, each enhanced measurement option reviewed, the longer retention choice unless a policy forbids it, internal traffic rules, the cross-domain list and the unwanted referrals. Match the time zone and currency to the old property unless they were wrong, because every comparison between the two depends on where each day starts and which currency revenue is recorded in. Grant access outward to the people and suppliers who need it. Leave the internal traffic filter in its testing state for now; it is activated once the comparison shows exactly what it removes. When the property is built, tick the GA4 setup checklist against it, so nothing the old property had is missed by accident.

  4. Send every event to both properties from the same container

    Add the new property's Google tag beside the existing one rather than replacing it, with the new measurement ID held in its own constant variable. The tutorial on installing GA4 through Tag Manager keeps the configuration parameters in a configuration settings variable so that a second property can reuse them; point the new Google tag at the same variable, then give each event tag a twin that sends to the new ID, or add a second generic event tag where the container uses one. Apply the consent settings the old tags carry to the new ones, following the consent mode guide, or the two properties will see different visitors. Test in preview, and in a staging environment where the site has one, as the containers and workspaces guide describes, then publish with a version note. This is the start of a Parallel Run: both properties now receive the same events from the same pages.

  5. Link Google Ads and the BigQuery export

    Link Google Ads from the new property with personalised advertising and auto-tagging switched on, and link BigQuery to a project the business owns and pays for. Do this at the start of the run rather than at the switch. The BigQuery export guide explains that the export holds nothing from before its link, so every day the link waits is a day of event-level history the new property will never have. Leave the old property's links where they are; Google Ads keeps reading the old conversions until a later step moves them. Keeping the old Google Ads link while adding the new one is general practice for this move, not something the site's guides describe, so confirm in the Google Ads help centre that the account can hold a link to the second property before relying on the parallel import. Search Console stays with the old property for now and moves at the switch.

  6. Rebuild the key events, definitions and audiences that pass the clean-up tests

    Bring across only what earns its place. The clean-up tutorial's tests decide which events deserve the Key Event mark, which definitions a report or dashboard still reads and which audiences a live campaign uses; apply them to the configuration copied earlier rather than rebuilding the old property's clutter. Create each key event by name with its counting method and default value, as the key events guide sets out, and register each Custom Dimension at the scope the site actually sends. Rebuild an interface rule only where the tag cannot be corrected first: the events and parameters guide explains that such rules are patches that live in one property and that the tag should stay the source of truth. Build the audiences early, because the audiences guide advises planning as though a new list starts empty, and campaigns cannot use these until they have filled.

  7. Compare the old and new properties over closed days

    Once both properties hold complete days, set them side by side for the measures the business steers by: sessions by channel, each key event, purchases and revenue, and the same counts from the order system or CRM. The two are fed by the same pages, so they should move together. A step up in one points at a tag firing twice, a step down at a template or an event the new tags miss, and a gap on one channel at a missing referral exclusion or cross-domain entry. The method is the count comparison the tutorial on moving hard-coded tags runs after its cut-over. Activate the new internal traffic filter once its test label shows only the visits meant to be removed, and judge how close is close enough from the business's own daily volumes rather than from a borrowed figure.

  8. Move Google Ads goals in stages

    Import the new property's key events into Google Ads as secondary actions, so they are recorded while bidding ignores them, as the Google Ads integration tutorial describes; two primary actions for the same purchase would count every order twice. When the secondary actions agree with the old ones and with the business's records over closed days, switch which actions are primary with the conversion goal change checklist, which is the gate for that change. Move campaigns from the old property's audiences to the new ones only after the new lists have filled.

  9. Repoint dashboards and data sources

    For each Looker Studio report on the reader list, create a data source on the new property and point the charts at it, working on a copy and choosing each field again, because each chart inherits its source's fields and the new definitions are new objects. Choose credentials on purpose: the data sources guide explains that owner's credentials stop when the owner's account is suspended and that viewer's credentials need every viewer to hold access. Rewrite export queries that name the old dataset and rebuild the explorations the team uses. Show each repointed report beside the original for a reporting cycle before its viewers switch.

  10. Switch reporting, mark the break and stop the old feed

    Pick a switch date from which every report reads the new property, and on that date move the Search Console link to the new property and publish its collection there, as the setup tutorial describes. Any chart that crosses the date compares two properties, so treat the date as a break in the way the guide to comparing periods sets out: restate where the raw data allows, compare within one side, or read a measure the move did not touch, such as orders in the shop system, and keep the overlap from the run, because it shows how far the two properties differ. Only when every reader has moved and a reporting cycle has been read from the new property, remove the old tags in a container release. As general practice, keep read access to the old property for as long as it is offered, and the copies for as long as the business needs that history.

  11. Record the move in the change log and the tracking plan

    Write a dated Change Log entry for each milestone: the new property's creation, the start of the run, each filter activated or exclusion added during the comparison, the Google Ads switch, the reporting switch and the removal of the old tags, as the measurement change log tutorial describes. Update the tracking plan with the new measurement ID, key events, definitions and audiences, and note where the old history and the copies live and who can open them.

A stained-glass maker leaves its agency's account (illustrative)

Illustrative example — not a real result

Tarmivosk Stained Glass moves to a property it owns

Tarmivosk Stained Glass, an invented maker of leaded glass panels that sells online and takes commissions, ends its contract with an agency and asks for its analytics property. The property sits in the agency's account beside 31 other clients, and after 3 written requests the agency declines to move it. Before access ends, the owner exports 24 months of channel, landing page and e-commerce reports and the configuration: 9 key events, 14 custom definitions and 11 audiences.

DayStepWhat happened
1New property and twin tags publishedSame time zone and currency as before; internal traffic filter in testing
2Google Ads and BigQuery linkedThe export holds data from day 3 onwards
3Rebuild3 key events, 5 definitions and 4 audiences pass the clean-up tests
8ComparisonThe payment provider appears as a referral in the new property only; exclusion added
9Google Ads importNew purchase action added as secondary
21Comparison over days 10 to 20186 purchases in the old property, 187 in the new, 191 orders in the shop; internal traffic filter activated
22Goal switchNew purchase action made primary, old one set to secondary, date logged
29ReportsWeekly Looker Studio report repointed and shown beside the old one
36SwitchReporting reads the new property and the Search Console link moves across; old tags removed on day 43

The change log gains 7 dated entries, and every trend chart in the weekly report carries a note that its series crosses from one property to the other on the switch date, with the comparison from the overlap kept for reference.

The GA4 hub explains the event model each rebuilt setting configures and lists the guides that cover every object in more depth.

Common mistakes

  • Starting again when the business could have owned the old property. A transfer keeps history, audiences and conversions; ask in writing before building anything.
  • Replacing the old tag instead of adding one beside it. With nothing to compare against, a gap in the new property looks like a fall in demand.
  • Rebuilding everything the old property held. The clutter moves across with it; bring only what the clean-up tests keep.
  • Linking the export at the end. Every day without the link is event-level history the new property never gets.
  • Making the new purchase primary while the old one still is. Bidding counts each order twice until one is demoted.
  • Switching reports without marking the date. The next year-on-year comparison reads the move as growth or decline.
  • Removing the old tags first. It is the step that cannot be undone, so it comes last, after every reader has moved.

Frequently asked questions

Can the old property's history be copied into the new one?

Not by any route this site describes. History stays in the old property, and the new property's BigQuery export begins with the first full day after its own link. Export the reports the business reads year on year before access ends, keep read access to the old property for as long as it is offered, and treat the switch date as a break in every trend that crosses it, using the overlap from the parallel run to show how far the two properties differ.

How long should both properties run side by side?

Until the evidence says the new one can carry the business, not for a set number of weeks. That means closed days in both properties moving together and close to the order system or CRM, every Google Ads action, audience, dashboard and query moved across, and a full reporting cycle read from the new property. A business with few orders a day needs a longer run to tell a real gap from day-to-day noise than a business with many.

Is a new property better than cleaning up a messy one?

Only when the business cannot own the old one. A property the business controls keeps its history, audiences and imported conversions through a clean-up, and the clean-up's register lists every setting a rebuild would have to decide on. Start again when ownership cannot be secured, or when the property stays with someone else after the business changes hands, and transfer or repair it in every other case.

Terms used on this page

Frameworks

Sources

  1. Google Analytics Help — Google Analytics Help
  2. [GA4] Set up BigQuery Export — Google Analytics Help
  3. [GA4] Audiences — Google Analytics Help
  4. Google Ads Help — Google Ads Help

Related