Campaign budget or ad set budget for creative tests on Meta

What is the difference between campaign budget and ad set budget on Meta?

Meta’s budget page offers three options: Advantage+ campaign budget, ad set budgets and ad set budget sharing. With Advantage+ campaign budget, one budget “continuously distributes in real time to ad sets with the best opportunities”. Ad set budgets give each ad set its own amount, listed for when “You want to control the amount spent on each ad set.”

Ad set budget sharing sits between them. Each ad set keeps its budget, and “up to 20%” of the daily budget moves to other ad sets in real time (budget page).

Meta’s Advantage+ campaign budget page adds the catch for testers. A campaign budget “may not spend your budget equally for each ad set”, and with 2 active ad sets “we might spend most of your budget on one ad set”. Meta’s advice follows: “analyze results at the campaign level, rather than at the ad set level.”

Which budget level suits a creative test?

Ad set budgets suit a creative test that must read every variant, because each variant then gets the spend you set. AI tools make variants cheaply. Each one still needs enough conversions to be read, and an Advantage+ campaign budget can put most of the spend on one of them.

Advantage+ campaign budget suits the next step: scaling the variants that already won. Meta lists it for when “You value all ad sets the same”, which a test of unproven creatives does not.

A middle route exists. Meta’s budget page says you can set “minimum and maximum spending limits for each ad set” under Advantage+ campaign budget. A minimum spend limit on each ad set keeps every variant spending while Meta moves the rest.

How much each variant needs, at your CPA, is the maths in the creative testing framework. The creative testing budget calculator runs it with your own figures.

Meta’s three ways to test creatives: creative test, A/B test or manual

Only Meta’s A/B test reports a confidence level.

The creative test compares “up to 7 creative variants”. Meta suggests “no more than 20%” of the existing budget for test ads. After the test, the test ads keep running “until you turn them off”, and the test schedule “only determines when results are measured”.

The A/B test shows each version to a separate part of the audience. Meta recommends the same budget for both versions, and it picks a winner on cost per result with a confidence percentage. It advises against testing informally by switching ad sets on and off, which risks overlapping audiences.

A manual test is a campaign of ad sets you compare yourself. Meta’s pages give it no confidence level and no audience split.

Which edits restart the learning phase?

Six edits always restart the learning phase, according to Meta’s significant edits page:

• “Any change to targeting”
• “Any change to ad creative”
• “Any change to optimization event”
• “Adding a new ad to your ad set”
• “Pausing your ad set for 7 days or longer”
• “Changing bid strategy”

Three more count only when the change is large: the ad set spending limit, the bid control or cost goal, and the budget. Meta’s example: a budget moved from $100 to $101 “isn’t likely” to restart learning. Moving it to $1,000 means “one or more ad sets may reenter the learning phase”.

Under Advantage+ campaign budget, Meta answers three common worries on the same page. Spend moving between ad sets does not restart learning. An edit at one ad set’s level leaves the others alone. And “adding an ad set to a campaign with Advantage+ campaign budget won’t cause other ad sets within the same campaign to reenter the learning phase.”

An ad set leaves learning “after about 50 results in the week after the ad set’s last significant edit” (learning phase page). Ads Manager shows a “Last significant edit” column for checking it.

How to add new AI creatives without resetting learning

Every new ad restarts its ad set’s learning, so the aim is fewer restarts. Meta’s own pages support four ways.

1. Add a batch of ads in one edit. The 50 results count from “the ad set’s last significant edit”, so five ads added on five days restart the week five times, and five added together restart it once.
2. Put new creatives in a new ad set. The new ad set starts its own learning. Under Advantage+ campaign budget it leaves the other ad sets’ learning alone, per the significant edits page.
3. Use one ad with several assets. Meta’s ad volume page says “One ad can contain multiple (up to 10) creative assets” and recommends dynamic creative “If testing multiple creative variants”.
4. Run Meta’s creative test inside the campaign. Meta says the creative test keeps “delivery system learnings retained”. Adding a new ad is still on the significant edits list.

How to test new creatives in MetaSam Piliero

Editing a live ad’s creative in place restarts learning as well: “Any change to ad creative” is on the list. Pausing a weak ad is not on it; pausing the whole ad set for 7 days or longer is.

Meta also warns against volume. When an advertiser runs too many ads at once, “each ad delivers less often” and “fewer ads exit the learning phase”. The page asks advertisers to “Decrease ads per ad set” and names no number of ads. A dated log of each batch, with its edit date, fits in the creative testing template.

For the generators that make the variants, see AI Facebook ad generators.