GA4 Account Structure 101: Properties, Streams, Sub-properties & Roll-ups Explained

What you’ll learn

By the end of this 10-minute read you will:

• Know the 5 layers of GA4 architecture and why the event-based model killed Views
• See real-world diagrams pulled from Google’s official Q2’24 PDF (attached)
• Understand when to create a Sub-property vs. simply filter a report
• Avoid the #1 mistake that makes cross-domain tracking break
• Be able to estimate GA4 360 cost before your boss asks 😉

---

The GA4 hierarchy in one glance

Fig. 1: Accounts → Properties → Streams → Sub & Roll-ups (360)

Level	Purpose	Max number (Standard)	Max number (360)
Account	Billing & user management	1 → ∞	1 → ∞
Property	Data collection & processing	100 / acc	2 000 / acc
Data Stream	Source of events (Web, iOS, Android)	50 / prop	50 / prop
Sub-property	Filtered view + own settings	–	400 / acc
Roll-up	Combine >1 property	–	400 / acc

---

1. Account

Think of the Account as the “company folder” in Google Drive.

• One company = one Account (best-practice)
• Billing & admin permissions live here
• You can move full source properties between Accounts, but not Sub-properties or Roll-ups

> 🔍 Example: Parent company “Alphabet Inc.” has one Account. Inside it live Google Store, Nest, Fitbit, etc. as separate Properties.

---

2. Property

The Property is where data is stored & processed.

Key differences vs. Universal Analytics

UA	GA4
Session-based model	Event-based model
Hit types: page, event, e-commerce, social	Everything = event
Views allowed	Views removed

When to create a new Property

✅ DO create a new Property when:

• Legal requirement to isolate data (GDPR, HIPAA)
• Brands with zero overlapping users (e.g., B2B SaaS vs. consumer gaming)
• You need different currency or timezone settings

❌ DON’T create a new Property for:

• Region-specific reporting (use Sub-property or filtered report)
• Staging vs. production (use data filters instead)

---

3. Data Stream

A Data Stream is the pipe that sends raw events into a Property.

Best-practice: 3-stream rule

1. Web (all domains & sub-domains that share user journey)
2. Android
3. iOS

Fig. 2: One Property, three streams → unified user-ID

Multiple top-level domains?

Put them in the same web stream and enable cross-domain tracking in the Admin panel. Creating separate streams will:

• Split users into two stream_id dimensions
• Break sessionization
• Inflate user & session counts

---

4. Sub-property (GA4 360 only)

A Sub-property is a filtered copy of a source Property with its own settings, quotas and user-access rules.

Fig. 3: EMEA directors see only EMEA events

Use-cases:

• Data governance – regional teams must never see other regions
• Product-line isolation – t-shirt division vs. footwear
• Agency sandbox – give external agency access to only campaign data

Limits & pricing:

• Data starts from creation date – no backfill
• Events billed at 0.5× source-property rate
• Inherits full GA4 360 limits (50 custom dims, 1 B events/day, …)

---

5. Roll-up property (GA4 360 only)

A Roll-up combines ≥2 source Properties (or Sub-properties) into a single reporting view.

Fig. 4: Global executives see AMER + EMEA + APAC in one dashboard

Common patterns:

Pattern	Roll-up members	Typical user access
Brand roll-up	Nike, Converse, Jordan	Global CMO
Regional roll-up	AMER-North, AMER-South	AMER VP
Product roll-up	T-shirt sub-properties	Merchandising team

Pricing same as Sub-property: 0.5× per event.

> ⚠️ You cannot “nest” Roll-ups inside Roll-ups.

---

360 billing: what costs double, what costs half

Scenario	Billing multiplier
Source Property	1× (baseline)
Sub-property	+0.5× filtered events
Roll-up property	+0.5× combined events
Re-sending same event to >1 Property	1× each Property → 2× total

Pro-tip:
Estimate monthly cost with:
Source events × $0 + Sub (E×0.5) + Roll-up (E×0.5)
If your source Property fires 20 M events and you create one Sub + one Roll-up, budget for 20 M + 10 M + 10 M = 40 M events.

---

Checklist: how to choose the right structure

Answer the following before touching the Create button:

1. Do users ever overlap between sites/apps?
   → Yes = same Property
2. Is there a legal reason to isolate data?
   → Yes = separate Property
3. Do teams need different settings (currency, timezone)?
   → Yes = separate Property
4. Do teams need only filtered data & no sensitive access?
   → Use Sub-property (360) or custom report (Standard)
5. Do executives need aggregate view across brands/regions?
   → Create Roll-up (360) or Exploration with segments (Standard)

---

Next steps

1. Open your GA4 Admin panel and list every current Property & Stream
2. Map them to the decision tree above
3. If you’re on 360, prototype one Sub-property for your smallest region and measure loading speed & quota usage for 7 days
4. Share this article with your data governance team to align on cost vs. access trade-offs

Ready to migrate Firebase or link BigQuery?