GA4 cross-channel budgeting: projection plans and scenario plans

GA4 cross-channel budgeting is a feature, still in beta, that uses your property's cost and conversion history for two kinds of plans: a projection plan, which tells you whether an in-flight campaign will hit its spend, conversion or revenue target, and a scenario plan, which models budget levels for a future period and suggests how to split spend across channels. You'll find it under Advertising → Budgeting. Google itself says the outputs are model estimates, not guarantees. So treat a plan as a starting point: before you move serious money, validate it with an incrementality test.

What it is and where to find it

The feature showed up in Google Analytics' release notes in January 2026, in beta. To find it: in the left menu, go to Advertising, then under Budgeting pick Projection or Scenarios. The overview page lists recent plans, linked products, cost imports and manual integrations.

Three caveats from Google's help center before you start:

• It may not be available in your property. Access is being expanded gradually.
• Desktop only. The Advertising section doesn't appear in the app.
• The models are at different stages. The data-driven attribution model is in beta. The Meridian-based model is in alpha and limited to select Google Analytics 360 customers.

Projection plan vs scenario plan

Google recommends using both: scenarios to decide the split up front, projections to track it afterwards.

Two scenario-plan limits catch a lot of people: it handles one conversion at a time, and it doesn't refresh. If the market moved, you build a new plan.

What data GA4 uses

This is what decides whether the plan is any good. The model draws on:

• Linked Google products, such as Google Ads and Display and Video 360.
• Imported cost from non-Google channels like Meta and TikTok, through campaign data import.
• Manual third-party integrations.
• Offline data, if you've set up offline imports.

On the conversion side, it uses click-through conversions and YouTube engaged-view conversions, along with all cost data. Google recommends including impressions where available. On September 29, 2026, the help center noted that cross-channel budgeting currently supports web conversions only.

The requirements: roughly a year of web conversion data (key events can backfill it), campaign data from linked or imported channels, and, for scenario plans, data-driven attribution.

The practical upshot: if Meta's cost isn't imported, Meta doesn't exist as far as the plan is concerned. And if your TikTok UTMs are messy, the cost arrives but doesn't match conversions, so the channel looks worse than it is. Before opening the planner, check that every paid channel has both cost and conversions in GA4.

What the plan doesn't know

The model looks at your property's past. It doesn't know:

• What happens in your CRM. If leads turn into sales days later, through a sales team, and you don't send those sales to GA4, the plan optimizes for leads. The channel bringing cheap leads and zero sales will look like your best one. That trap is covered in low CPL, no sales.
• What's cause and what's coincidence. History shows correlation. If remarketing always looks great because it talks to people who were going to buy anyway, the plan will ask for more remarketing.
• Real-world limits. Inventory, sales team capacity, creative fatigue, the competitor who came in hot this month.

That's why Google spells it out: these tools are for planning only, won't touch budgets or spend, and deliver estimates, not guarantees.

A worked example

Say the scenario plan suggests moving 20% of your Meta budget to Google Ads next month, promising more conversions for the same total. Before you accept, ask three questions:

1. Was Meta's cost imported for the whole period the model used? If a month is missing, Meta looks more expensive than it was.
2. Is the conversion you picked the sale or the form fill? If it's the form, the plan is comparing leads, not customers.
3. Can you test the shift in one region, or for two weeks, before applying it everywhere?

If all three answers hold up, the plan has become a testable hypothesis. If not, it's an opinion with nice formatting.

Why you should validate with an incrementality test

The plan says "put more into channel X". A test tells you whether channel X causes sales. Different questions, and only the second protects the budget. The method is laid out in incrementality testing.

Google's most recent options for this:

• Meridian GeoX. Google's open-source library for geo experiments, which came out of beta in September 2026. It works with any ad platform, not just Google. It needs daily data, enough geographic variation, and someone to design and read the experiment.
• Search Lift via the API. Since October 14, 2026, version 26 of the Google Ads API includes Search Lift metrics, which measure the increase in searches caused by video and Demand Gen campaigns. Useful for upper-funnel channels that last-click can't see.

A simple flow: run a scenario plan, pick the biggest change it suggests, test that change in one region or one period before rolling it out, and only then move the budget. How to shift money in steps is covered in AI ad budget allocation.

Where Vazante fits

Next step

Before your first plan, get the basics right: costs imported for every paid channel, consistent UTMs and a year of web conversions. Then run a scenario, compare it with your current split and test the biggest difference before going further. If budget splits are still a gut call, start with splitting budget across channels. And to track the plan against real spend across channels, see paid ads reporting.