The short answer
Without 360 you have three free routes: send every site's tag to a second, combined property; blend the properties in Data Studio (still called Looker Studio in many tutorials); or have an AI agent read each property's stored history and build the table. Whichever you choose, check time zones, currencies, channel rules and key events first.
- Roll-ups need a 360 order and 360 source properties, and they start with no history
- Rank on rates, not totals, and only once every property counts the same action as a conversion
- An agent over DuckDive starts from history you already have, with no retagging
Four ways to compare properties, side by side
Google's built-in answer to cross-property reporting is the roll-up property. If your sites run on standard GA4, that door is shut, and the alternatives trade money for effort in different ways.
| Route | Who can use it | History on day one | The catch |
|---|---|---|---|
| 360 roll-up property | 360 only, with an active 360 order | None, starts when each source joins | Extra cost, and every source must be 360 too |
| Dual tagging into a combined property | Anyone who can edit every site's tags | None, starts when the second tag fires | Retag every site; one time zone for all |
| Data Studio blend | Anyone with access to each property | Yes, whatever each property holds | Five sources per blend; GA4 API quotas apply |
| An agent over DuckDive's stored data | Anyone whose Google login sees each property | Yes, pulled in as each property is linked; agent ranges stop at a trailing year | Fixed windows; no event names, users or revenue |
Sources: Google's roll-up help page, Adswerve on roll-up properties, Google's blend documentation, checked November 2026.
None of the four fixes bad inputs. If one property counts sales and another counts newsletter sign-ups, every route hands you a tidy table that means nothing.
Roll-up properties and who can use them
A roll-up property collects events from several source properties into one combined data set. Google's help page is blunt about access: roll-ups are "available only to Google Analytics 360 accounts" tied to an active 360 order, and "only Google Analytics 360 properties can be sources." A standard property can't feed one, even when the roll-up itself sits in a 360 account.
If you are pricing 360 just for this, know what you get:
- Room for many sources. Google allows up to 200 source properties per roll-up (as of November 2026). Adswerve's 2022 write-up and the Optimize Smart tutorial still say 50, so check dates.
- No backfill. Google says a roll-up starts accumulating data on the date each source is added. Adswerve puts it plainly: "there is no backfill of historical data from the source properties."
- A second bill. Both Adswerve and Optimize Smart describe roll-up data as an additional charge, at half the rate of the same data in each source property.
- Little carries over. Custom dimensions, users and permissions are not inherited, per Google.
- Its own key events. You can mark events as key events in the roll-up without marking them in the sources.
That last point is the roll-up's quiet strength: one person decides, in one place, what counts as a conversion for every brand. If you already pay for 360, use it. The rest of this guide is for everyone else.
Dual tagging into a combined property
This is the free imitation of a roll-up. You create one extra standard property, then make every site send its data there as well as to its own property. Optimize Smart calls it "standard roll-up reporting" and describes firing "a second GA4 configuration tag on all the pages of the website."
With gtag.js, Google's guide to grouping and routing data shows the mechanics: each config command adds a target, so a second measurement ID sends events to both properties. What it costs you:
- A release on every site. Four sites means four deployments, four test rounds, four chances to lose the tag in a redesign.
- A blank past. The combined property knows nothing from before the second tag went live.
- One reporting day for all. Time zone and currency are property settings in Google's property definition, so every site gets one day boundary.
- Users that do not merge. Optimize Smart notes that de-duplication of unique users is not possible this way. A shopper who visits two of your sites still counts twice.
- Event names as sent. If one site sends
purchaseand anothergenerate_lead, starring both in the combined property puts a sale and an enquiry in one conversion count.
Dual tagging earns its keep when the sites share a developer, you want combined funnels or audiences from now on, and nobody needs the past.
Blending in Data Studio, formerly Looker Studio
Google's current documentation calls the tool Data Studio (as of November 2026); older tutorials use its previous name, Looker Studio. Same free tool, so they still apply.
A blend pulls fields from several data sources into one chart. Google's blend documentation sets the limits: up to five distinct data sources, five join operators, and a union operator. A union stacks each property's rows; a join on date lines up their sessions day by day. Three things catch people out:
- Five is the ceiling. A sixth property means a second blend.
- Rows are aggregated before they merge. Google says each table is grouped and aggregated by its dimensions first. Build any combined rate from summed sessions and conversions, never by averaging each property's rate.
- Quotas. Google's troubleshooting guide says "Reports that connect to Google Analytics 4 (GA4) data are subject to GA4 Data API quotas." For a standard property the Data API limits are 200,000 tokens per property per day and 40,000 per hour (as of November 2026). Exceed them and you may wait up to 24 hours. The Google Analytics token usage menu shows which charts cost most.
Blending suits a dashboard someone opens every Monday, and is clumsy for one-off questions: each one is a new chart to build.
A fifth route
Teams that write SQL can send each property to BigQuery Export and union the tables. Standard properties can export to a free sandbox, with a daily limit of 1 million events; usage past sandbox limits is charged. Most flexible, most work.
Asking an agent over stored data
DuckDive changes nothing on your sites. You sign in with Google, add each site and link its GA4 property. DuckDive imports that property's history and keeps it, so your agent (Claude, ChatGPT, Cursor or any MCP client) reads stored numbers for each site and lays them side by side.
There is no single cross-property tool. The agent calls get_ga4_overview once per property and does the lining up itself. Nobody touches a tag, but the agent compares whatever each property calls a conversion, which is why the next section exists.
Ask your agentFor my four Hollowmere Outdoor properties, show sessions, conversions and conversion rate for the last 90 days in one table.
Tools it calls list_websites get_ga4_overview
Three GA4 tools cover daily totals, channels and landing pages, over fixed windows from 7 days to 12 months. The GA4 data source page defines each number, so this guide sticks to comparing them.
Ask your agentFirst Monday of the month: one table for all four Hollowmere sites with sessions, conversions, conversion rate and the share of conversions from organic search, last 30 days. Put the site that moved most since last month at the top.
Tools it calls list_websites get_ga4_overview get_ga4_channels
Time zones, currencies and key event names
Four property settings cause most wrong comparisons, and none of them shows up in a report. Here is how this guide's fictional retailer had its regional properties set up.
| Property | Time zone | Key event |
|---|---|---|
| Coast | Los Angeles | Online purchase |
| Highlands | Denver | Trip enquiry form |
| Lakes | Chicago | Online purchase |
| Desert | Phoenix | Newsletter sign-up |
Time zones move the day, not the week
Google defines a property's time zone as the "day boundary for reports, regardless of where the data originates," and notes that changing it "only affects data going forward, and is not applied retroactively." Properties an hour or two apart report the same 90 days almost identically, but a single launch day can split across two dates.
Compare multi-week windows across properties, and treat any single-day comparison with suspicion.
Ask your agentCoast and Lakes ran the same 24-hour flash sale last week. Show daily sessions and conversions for both over the last 14 days and tell me if the spike lands on different days.
Tools it calls list_websites get_ga4_overview
Currency behaves the opposite way
Here GA4 rewrites the past. Google's currency reference says "Changes to the global currency type will affect future as well as historical data." Local-currency transactions convert at the prior day's exchange rate. So Lakes' revenue, in Canadian dollars, never sits cleanly beside Coast's in a blended revenue chart. Through DuckDive this matters less: the GA4 tools return sessions and conversions, not revenue.
Channel rules can differ per property
"Organic Search" sounds universal. It is only universal in the default channel group. Google's custom channel groups page says a standard property gets two custom groups, any custom group "can be turned into your property's primary default," and custom groups apply to reports retroactively. If one site's analyst rebuilt the rules, its organic share uses a different ruler.
Before comparing organic shares, check each property's Admin for a custom primary group. Where there is one, ask the agent for traffic sources instead of channels.
Key event names are the big one
Google publishes recommended event names such as purchase, generate_lead and sign_up, and says "You should send recommended events with their prescribed parameters to get the most details in your reports." In practice each site stars something different, and a rate built on a sale, an enquiry and a sign-up is noise.
DuckDive's tools return conversion counts, not event names, but the agent can spot the pattern each event leaves on landing pages.
Ask your agentFor each Hollowmere property, list the 10 landing pages with the most conversions in the last 30 days, so I can see what each site is really counting as a conversion.
Tools it calls list_websites get_ga4_landing_pages
Worked example: four properties with different key events
The head of ecommerce wants a ranking for the quarterly review. The answer is only useful because the agent refuses to take the numbers at face value.
Rank the four properties by conversion rate, and warn me if any of them seems to count a different kind of conversion.
list_websitesfinds the four Hollowmere websites and confirms GA4 is linked on eachget_ga4_overviewonce per property, last 90 days: sessions, conversions and rateget_ga4_landing_pagesDesert and Highlands, last 90 days, to see where their conversions happen
On paper Hollowmere Desert wins by a wide margin. I would not put this ranking in front of anyone yet.
| Property | Conversions | Rate |
|---|---|---|
| Desert | 1,215 | 9.80% |
| Lakes | 412 | 1.80% |
| Highlands | 537 | 1.70% |
| Coast | 723 | 1.50% |
Desert's conversions cluster on content, not the shop: /newsletter logged 377 from 610 sessions and /journal/desert-water-planning 402 from 2,980, while the tents page converted 9 of 1,140. That fits a newsletter sign-up, not a sale.
Highlands' conversions sit on /trips/ pages, which fits an enquiry form. I can't read key event names through DuckDive, so confirm both under Key events in each property's Admin. Until then only Coast and Lakes look comparable, and Lakes leads, 1.80% to 1.50%.
Once a mismatch is confirmed, you have three honest options. Compare only properties that count the same act. Star a shared event everywhere and compare windows that start after that day (the GA4 integration guide explains why). Or compare each site with its own past, which survives any naming mess.
Find the property that counts something else
Link your GA4 properties once, then ask which one has the suspicious rate before the quarterly deck goes out.
Free for 2 sites · No credit card
Worked example: comparing on rates, not totals
"Which site is best?" sounds like one question. When one site gets most of the traffic, it is two: who produces the most, and who does the most with each visit. Same retailer, a later window, all four sites now counting purchases only.
Which Hollowmere site is best at organic search? Last 30 days.
list_websitesfinds the four Hollowmere websitesget_ga4_channelsonce per property, last 30 days, keeping only the Organic Search row: sessions and purchases
"Best" has two answers here, and they point at different sites. Coast brings in the most organic sales, 118. Lakes turns organic visits into sales best, at 2.30%, close to double Coast's 1.20%.
| Property | Organic sales | Organic rate |
|---|---|---|
| Coast | 118 | 1.20% |
| Highlands | 82 | 2.00% |
| Lakes | 61 | 2.30% |
| Desert | 21 | 1.10% |
Sorted by sales, Coast is first and Lakes third. Sorted by rate, Lakes is first and Coast third. Coast gets 3.7 times as many organic visits as Lakes (9,800 to 2,650), enough to lead on orders while converting a smaller slice of them.
Pick the column by the decision. Deciding where one more SEO article earns most per visitor? Lakes. Forecasting organic orders or stock for next quarter? Coast. Highlands is second on both lists, so it is the safest answer if you need just one name. Tell me which decision this is for and I will drop the other column.
Four rules keep rate comparisons honest on any route:
- Same window everywhere. Ask for one fixed window across all properties, and in a dashboard, drive every chart from one date control.
- Sum, then divide. A group rate is total conversions over total sessions. The average of four rates is a different, usually wrong, number.
- Set a floor. Set a minimum session count before a site is ranked (the benchmarking guide covers how to pick one).
- Sessions add up; users don't. Someone who visits two of your sites is a user in both properties, so summing users overstates reach.
Ask your agentAcross the four Hollowmere sites, compute one combined conversion rate for the last 6 months from total conversions over total sessions, then show which site pulls it up and which drags it down.
Tools it calls list_websites get_ga4_overview
Is a combined property worth it?
A combined property is the only free route that gives you cross-site funnels, audiences and explorations, and the only one that needs a code change on every site. Run through this first.
Build a combined property only if most of these are true
- You need cross-site funnels or audiences. A table alone doesn't justify retagging.
- You control every site's tags. Sites built by outside vendors or franchisees add a coordination step to every change.
- One time zone suits all of them. The combined property has exactly one day boundary.
- Nobody needs the past. The combined property starts empty.
- Every site sends the same event name for the same act. Otherwise you inherit the mismatch.
- Someone owns the tag after the next redesign. A second tag is easy to lose when a site is rebuilt.
If two or fewer apply, compare without combining: a Data Studio blend for a recurring dashboard of up to five properties, an agent over stored history for questions that change weekly.
Groups with separately run sites usually land in the second camp: a holding company comparing its brands' websites, a private equity team baselining acquired companies (which needs the months before the deal), an auto dealer group with one site per rooftop, or a manager with a website per apartment community. The wider case is on the portfolios page.
Setting up the comparison in DuckDive
Plan on one DuckDive website per GA4 property: four for a group like Hollowmere.
-
Check that one Google login sees every property
DuckDive lists only properties your Google account can open. If one is missing, an admin on that property adds you as a user.
-
Sign in with Google
Use that same login on the Continue with Google button. Your DuckDive account is made during this step, and Google asks you once to let DuckDive see the Analytics properties and Search Console sites the login can reach. Approve it and all four properties become available to link in step 4.
-
Add a website for each property
Add each regional site by its address. If a property is already linked to another site in your account, DuckDive checks with you before moving it.
-
Link each GA4 property and let history import
list_available_sourcesshows the properties your login reaches, with the likely match for each address first, andconnect_website_sourceslinks them and starts importing history. Usually a few minutes; longer for a long history.get_sync_statusshows progress. -
Connect your agent, or let it do steps 3 and 4
Add the MCP server
https://api.duckdive.ai/mcpand continue with Google when the DuckDive page opens. The menu is usually called "Connectors", "Apps" or "MCP". For the agent to add sites itself, keep "Also allow actions" ticked on an admin account.
Give your agent this promptSet up DuckDive for my website. Follow https://analytics.duckdive.ai/
agents.md -
Ask the comparison you have been putting off
Start with the 90-day table, then check landing pages wherever a rate looks too good.
Where an agent comparison falls short
- It suspects, it can't prove. The tools return conversion counts, not event names, so a mismatch is inferred from landing pages. You confirm it in GA4 Admin.
- No custom dates. The agent chooses from six preset spans. A window that opens on the day a shared key event went live, or this month beside the same month last year, can't be requested yet.
- No users, devices, countries or revenue. Sessions, conversions, channels and landing pages only.
- Not a shared dashboard. For the same chart every Monday, a Data Studio blend serves better.
- Deep single-property work. For event-level questions, see Google's GA MCP server compared with a hosted connector.
- One site is free; four are not. The free plan covers one site, so a four-property comparison needs a paid plan, priced below.
Questions about comparing GA4 properties
Can I create a roll-up property from standard GA4 properties?
No. Google's help page says roll-ups are available only to Analytics 360 accounts with an active 360 order, and only 360 properties can be sources. A standard property can't feed one.
Does a roll-up or a combined property include data from before I set it up?
Neither does. A roll-up starts on the date each source is added, with no backfill; a combined property starts when the second tag fires. Blends and an agent over DuckDive both use history each property already holds.
How many GA4 properties can I blend in Data Studio?
Google's documentation allows up to five distinct data sources in one blend. Past five, you need a second blend or another route. Busy blended reports can also hit GA4 Data API quota errors.
Why does one of my properties show a much higher conversion rate than the others?
Most often it counts a different action. A property that stars a newsletter sign-up will beat one that stars purchases every time. Look at which landing pages carry its conversions, then check its key events in Admin before you trust the ranking.
How does DuckDive compare several properties if it has no roll-up?
Each property is linked to its own website in DuckDive and its history is stored. Your agent reads each website in turn with the GA4 tools and builds the table itself. Nothing on your sites changes, and the free plan covers one site.
Can the agent set each property's November beside the same month a year earlier?
Not through the agent today. Its longest GA4 range is the trailing 12 months, which is enough to rank the four properties over a full year but not to split that year from the one before. For a same-month check, turn on Compare in each property's GA4 date range picker, or build that single chart as a Data Studio blend with one date control.
Sources
- [GA4] About roll-up properties, Analytics Help, checked November 16, 2026.
- All About Google Analytics 4: Aggregating Data with GA4 360 Roll-Up Properties, Adswerve, checked November 16, 2026.
- How blends work in Data Studio, Google Cloud Documentation, checked November 16, 2026.
- Roll up Property in Google Analytics 4: Tutorial, Optimize Smart, checked November 16, 2026.
- Group and route data, Tag Platform, Google for Developers, checked November 16, 2026.
- REST Resource: properties, Google Analytics Admin API, Google for Developers, checked November 16, 2026.
- Troubleshoot errors in Data Studio, Google Cloud Documentation, checked November 16, 2026.
- Data API limits and quotas, Google Analytics, Google for Developers, checked November 16, 2026.
- BigQuery Export, Analytics Help, checked November 16, 2026.
- Currency reference, Analytics Help, checked November 16, 2026.
- Custom channel groups, Analytics Help, checked November 16, 2026.
- [GA4] Recommended events, Analytics Help, checked November 16, 2026.
Last reviewed . Numbers in EXAMPLE DATA blocks are fictional.
