Meta’s latest quarterly results put a concrete number on AI’s role in advertising: its third-quarter ad revenue reached $63.1 billion, up 26% year over year. A fresh CNBC analysis published November 1 connects that growth to Meta’s AI-driven ad systems and reports a more operational signal for marketers: advertisers adopting recommendations from Meta’s AI Business Assistant saw nearly 5% more conversions at the same cost per conversion than similar non-adopters, according to the company. Nearly 10 million advertisers used at least one of Meta’s generative AI creative tools during the quarter.

The takeaway is not that every business should hand campaign decisions to a platform. It is that AI is moving beyond copy and image generation into the systems that rank ads, find audiences, and optimize delivery. Businesses should test those systems against their own economics, not assume a platform-wide result will automatically transfer to their account.

Meta’s AI is working across the advertising funnel

Meta’s Q3 results show a substantial advertising business growing through both more delivery and higher prices. Its earnings release reports that ad impressions increased 10% year over year and average price per ad rose 14%. Those are overall business metrics, not a breakdown of how much growth AI caused.

The newer operational details are more useful for campaign planning. Meta said its expanded GEM ad-ranking model lifted conversions by 1.7% for off-site conversion campaigns and 9.6% for app-ad conversions. Its AI Business Assistant recommendations were associated with almost 5% more conversions at the same cost per conversion among adopting advertisers, compared with similar non-adopters. In addition, nearly 10 million advertisers used at least one generative AI creative tool in Q3, according to Adweek’s earnings coverage.

That combination matters: one set of tools helps produce campaign assets, while another changes which ads are shown to whom and how campaigns are managed. The competitive advantage may come less from generating more variants and more from pairing strong inputs with better targeting and delivery.

Promising platform numbers are not a business guarantee

Each reported gain has a defined scope. A 9.6% lift for app ads is not a forecast for every advertiser, objective, or industry; the 1.7% result covers off-site conversion campaigns. Meta’s comparison for Business Assistant users is also company-reported and observational: advertisers who adopt recommendations may differ from those who do not. Public summaries do not provide enough experimental detail to independently reproduce every result.

Nor should the 26% rise in ad revenue be presented as an AI-only effect. Meta’s filing attributes growth to higher impressions and average ad prices, while management says AI improvements helped performance. Seasonality, advertiser demand, pricing, audience engagement, and other product changes also matter. For a business deciding whether to use more automation, its own incremental conversions and cost per qualified outcome are more relevant than Meta’s aggregate growth.

What businesses should do before expanding AI-led campaigns

  • Establish a baseline. Record cost per qualified lead or sale, conversion rate, revenue per conversion, and refund or cancellation rates before changing automation settings. Compare similar audiences and time periods, and use a holdout or controlled split where available.
  • Improve the signals first. Check conversion tracking, event deduplication, product or service data, and lead-quality feedback. Automated systems can optimize toward the wrong goal when the conversion event is incomplete or weakly connected to revenue.
  • Keep creative review in the workflow. Human reviewers should verify claims, pricing, rights, disclosures, brand voice, and landing-page consistency before generated assets go live. Save approved versions and note which AI features altered them.
  • Separate creative automation from delivery automation. Test one change at a time, such as generated creative or expanded audience delivery. If both change together, it becomes harder to identify what improved or harmed performance.

Run a measured test, then scale what earns its place

A practical 30-day pilot can start with one campaign and one business outcome. Freeze the existing campaign as a comparison, enable one AI capability in a matched test, set a spend ceiling, and review results weekly. Scale only if the test produces a meaningful improvement in incremental conversions or qualified revenue without degrading lead quality, margin, or brand safety.

Meta’s results make AI-powered advertising harder to dismiss as a novelty, but they do not remove the need for measurement and judgment. Real Internet Sales helps businesses assess AI marketing opportunities, strengthen campaign data, and build a test plan tied to revenue. Call 803-708-5514 or visit realinternetsales.com.