Are you hesitant to fully trust Meta's Advantage+ Audience? Meta frequently recommends using this AI-driven targeting feature, warning that reverting to original manual targeting options could hurt your ad performance. But how reliable is Meta's AI algorithm, and does providing "audience suggestions" actually make a tangible difference? We ran an A/B test to find out, and the results were highly unexpected.
The Promise of Advantage+ Audience
The core selling point of Advantage+ Audience is automation. Meta claims the system automatically prioritizes users who have already visited your website or interacted with your ads. According to Meta, even if you do not provide any audience suggestions, the algorithm will automatically prioritize conversion history, pixel data, and past ad engagers.
While this sounds like a hands-off solution, Meta still gives advertisers the option to input "audience suggestions" (such as custom audiences or lookalikes) to guide the algorithm. This leaves many media buyers wondering: does the AI actually need these hints?
The A/B Test Setup
To determine whether audience suggestions actually impact budget distribution and targeting accuracy, we set up a clean A/B test:
- Ad Set A (With Suggestions): Advantage+ Audience with specific Custom Audience suggestions.
- Ad Set B (No Suggestions): Advantage+ Audience with absolutely no targeting suggestions.
To ensure a fair test, the custom audiences used as suggestions in Ad Set A matched our exact retargeting segments. Theoretically, Ad Set A should have allocated more budget, generated more impressions, and driven more conversions from this specific group.
Our Hypothesis: Without suggestions, the algorithm would rely on broad targeting, allocating perhaps only 25% of the budget to warm retargeting audiences. With suggestions, that allocation should rise to at least 32% or more.
The Results: Meta's AI Outsmarts the Suggestions
The actual data completely disproved our hypothesis:
- Ad Set A (With Suggestions): Approximately 32% of the budget was spent on the suggested audience segment, aligning with our expectations.
- Ad Set B (No Suggestions): Surprisingly, 35% of the budget naturally flowed to the retargeting audience.
Even without manual suggestions, Advantage+ Audience allocated the majority of its budget to users who had already interacted with the brand or visited the website. The system automatically identified and prioritized high-intent users, proving that the algorithm does not need manual reminders to find warm audiences.
Why Did the "No Suggestions" Campaign Spend More on Retargeting?
There are a few logical explanations for why the campaign without suggestions actually allocated slightly more budget to warm audiences:
- Algorithmic Autonomy: Meta's underlying model prioritizes real-time behavioral data over manual targeting inputs. It trusts actual user actions more than the tags you provide.
- Cold Start vs. Warm Audiences: Providing suggestions can sometimes introduce broader, lookalike-adjacent users into the mix, dispersing the budget. Without suggestions, the model may more aggressively cut budget from users who are too far from converting.
- Statistical Variance: Ad delivery is inherently dynamic. A 2% to 3% fluctuation in budget allocation is well within the normal margin of variance for Meta's auction system.
The Takeaway for Performance Marketers
Do not treat audience suggestions as a silver bullet for your Meta campaigns. They function more as a loose reference point for the algorithm rather than a strict targeting command. Ultimately, the success of your performance marketing campaigns still relies on the fundamentals: high-quality ad creatives, strategic bidding, proper budget allocation, and robust conversion tracking.