Audiences

How do I set up an A/B test of Advantage+ audience vs a manual audience in Meta Ads Manager?

By The Tofu team ·

Use Meta's A/B test tool: in Ads Manager, click A/B Test in the toolbar, copy your ad set and change only the audience. One version keeps Advantage+ audience; the other uses Further limit the reach of your ads with Use as a suggestion unchecked. Give both the same budget and run 7 to 30 days (Meta's minimum advice and its maximum). Meta picks the winner on cost per result, so also compare cost per qualified lead in your CRM before you switch.

As of November 2026.

Before you start

  • A campaign that already gets leads, so you know what a lead costs (Step 5).
  • Lead source in your CRM: each lead must carry its ad set or campaign (Step 8).
  • A campaign Meta can test. In the Experiments tool, campaigns already in another test, and campaigns using the reservation buying type, can't be used.
  • Your manual audience, such as job titles, a customer list of target accounts, or both.

Step 1: Write down the hypothesis and the metric

Meta's best practices ask for "a testable hypothesis that enables you to improve future campaigns." For a B2B audience test, that might be: "Advantage+ audience will bring a lower cost per qualified lead than our job-title audience."

Decide two metrics before you launch:

  • Meta's winner metric. You choose it from the options on screen. Meta says it measures each version "on a cost per result basis or cost per conversion lift basis." Pick the result your campaign optimizes for, such as leads.
  • Your CRM metric. Cost per qualified lead or cost per opportunity. Meta won't see this one, so you'll judge it yourself (Step 8).

Step 2: Build the manual audience version

In the copy's ad set, open Audience, select Further limit your audience or Further limit the reach of your ads, and add your age, gender, custom audiences or detailed targeting. Meta says "You may have to uncheck the Use as a suggestion boxes." This turns Advantage+ audience off for that ad set. In a Leads or Sales campaign it can also switch the campaign to Advantage+ off, which is expected (see how to turn off Advantage+ on a Meta Leads or Sales campaign).

Two traps make a "manual" audience less manual than it looks:

  • Forced expansion. For goals such as website conversions, link clicks and landing page views, Meta applies Advantage+ detailed targeting with no opt-out, so your job titles still only guide delivery. Jon Loomer, who writes about Meta ads, reported in February 2026 that leads goals are forced too. See what "Advantage+ detailed targeting has been applied" means. A customer list included as a hard limit is the cleanest manual version.
  • Different controls. Keep locations, minimum age, languages and exclusions the same in both versions. Otherwise you're testing those too.

For the Advantage+ version, decide on audience suggestions up front. Jon Loomer suggests testing Advantage+ audience with no suggestions against your interest and lookalike targeting.

Step 3: Keep everything else the same

Meta says to "Test only one variable" and keep ad sets "identical except for the variable that you're testing." So both versions need the same:

  • creative and ad copy
  • placements
  • performance goal, bid strategy and conversion event
  • Instant Form or landing page
  • budget and schedule

Meta recommends "using the same budget for both versions in a test to ensure a fair comparison." Don't put both versions under one Advantage+ campaign budget. Meta says that budget "might spend most of your budget on one ad set," which would skew the test. Use two campaigns with one ad set each, or ad set budgets.

Step 4: Create the test

From the Ads Manager toolbar (Meta's recommended way):

  1. In Ads Manager, click A/B Test in the toolbar.
  2. Select Make a copy of this ad, choose the ad from the dropdown and click Next.
  3. Select an available variable. If audience isn't offered, select Custom, which Meta says duplicates the campaign or ad set so you can edit variables.
  4. Choose the ad set to copy. Name the test, choose how to determine the winner, and set the start and end dates. Meta says these dates are separate from your campaigns' dates.
  5. Click Duplicate Ad Set, name the new ad set, change only the audience, then click Publish.

From the Experiments tool (when you built both versions first):

  1. Go to Experiments and click Create test.
  2. Find A/B Testing and click Get started.
  3. Choose Test another campaign and select your campaigns (up to 5).
  4. Set dates, a name and how the winner is determined, click Review test details, then Create test.

Jon Loomer prefers this route: build both versions, check them, then create the test. Through the Marketing API, Meta's split testing can test audiences set at the ad set level, "one variable per test."

Step 5: Set the length and budget

  • Length. Meta recommends "a minimum of 7-day tests" and says tests "can only be run for a maximum of 30 days." In Ads Manager, the schedule must be 1 to 30 days. If your buyers take more than 7 days to convert after seeing an ad, Meta suggests running longer, "such as 10 days."
  • Budget. Meta's best practices give no minimum amount. They say your budget should "produce enough results to confidently determine a winning strategy." Meta says some tests include an estimated power before launch, and "We typically suggest tests have an estimated power of 80 percent or higher." It adds that this "doesn't guarantee a causal result."

To size the budget yourself, work backward from qualified leads. Multiply the leads you need per version by your cost per lead, then by two. For example, at $80 a lead, 60 leads per version means $9,600 in total. If 20% qualify, that's about 12 qualified leads per version, and a gap of 12 to 10 tells you little, so spend more or run longer if you can.

Step 6: Leave the test alone while it runs

  • Avoid overlap. Meta says your test audience shouldn't be used "for any other campaign" running at the same time, because "Overlapping audiences may result in delivery problems and contaminate test results." Advantage+ audience is broad, so pause other prospecting campaigns in the same countries if you can.
  • Don't edit one side. Meta counts any change to targeting, creative or the optimization event as a significant edit that can send an ad set back into learning.
  • Don't switch versions on and off. Meta advises against "turning ad sets or campaigns on and off manually."

Step 7: Read Meta's result

Go to Experiments, select View results, then View results on your test. Meta marks the version with the lowest cost per result on your key metric with "a green trophy icon." With no clear winner, it shows top performers on other metrics. If the test lacked data, Meta suggests running it longer. You also get an email when it ends.

Meta also shows a confidence figure: the likelihood that the outcome "would be consistent if repeated again under the same conditions." Meta says that "For A/B tests, a 65 percent or higher confidence percentage represents a winning result."

Step 8: Judge it on qualified pipeline

Meta's winner is the cheaper version per result. For B2B, check which brings more pipeline per dollar.

  1. Tag every lead with its version. Meta's lead download includes adset_id, adset_name, campaign_id and campaign_name. For website forms, pass the ad set in your UTMs. See how to add UTM parameters to Meta ads and Instant Forms.
  2. Wait for leads to be worked. Sales may need weeks after the test ends to qualify them.
  3. Compare per version: spend, leads, qualified leads, cost per qualified lead and, later, pipeline.
  4. Decide. If Advantage+ audience wins on cost per qualified lead, keep it and add the controls in whether to turn off Advantage+ audience for B2B. If your manual audience wins, keep it and retest later.

To let Meta optimize toward qualified leads itself, send your CRM lead stages back to Meta.

Common problems

ProblemWhy it happensFix
No audience option when you pick a variableMeta doesn't list which variables it offersSelect Custom and change only the audience
One version spent far moreBoth versions share one campaign budget, or budgets differUse separate campaigns or equal ad set budgets
No winner when the test endsNot enough resultsRun longer or raise the budget, up to 30 days
The manual version still reaches people outside your job titlesForced Advantage+ detailed targeting for your goalUse a customer list as the hard limit
Meta's winner brings worse leadsMeta judges cost per result, not lead qualityDecide on cost per qualified lead in your CRM

Sources

Ask Tofu next

Pick a question and Tofu Ads Agent answers it in chat.

Related answers

More on

Let an agent build your next campaign.

Tofu Ads Agent reads your CRM, plans the funnel, builds the audience and creative, and stages full campaigns across LinkedIn, Google and Meta for your approval.

Try it free Get a demo