What Happened
Social Media Examiner published an article on 17 September 2026 examining how far advertisers should let Meta’s AI take over campaign management. The piece covers recent updates to ad placements, creative AI and the algorithm, and asks which parts of a Facebook or Instagram campaign Meta AI now handles better than a human operator.
The framing matters because the direction of travel is one-way. Meta has spent several years moving advertisers from manual control towards automated products, first with campaign budget optimisation, then with automatic placements, and more recently with Advantage+ campaign types that consolidate targeting, placement and creative decisions into a single machine-led system. Each step has reduced the number of levers an advertiser pulls directly.
Social Media Examiner frames the question around two outcomes UK advertisers watch closely: lower ad costs and better lead quality. Those two do not always move together. Automation has a consistent record of reducing cost per result, because the system finds cheap inventory across Facebook, Instagram, Messenger and the Audience Network. Cheaper results are not automatically better results, particularly for B2B and high-value services where a form fill costs little but converts rarely.
This is general industry context rather than a claim from the article, but it explains why the question is live. For a UK business selling a £15,000 service, a cost per lead that drops 30 percent while lead quality halves is a net loss that standard in-platform reporting will not show you.
How to decide what to automate and what to hold
Start by separating decisions Meta can make from decisions only you can make. Meta’s systems are good at bidding, placement selection and serving the right creative variant to the right person at the right moment. They are working from signals you cannot see and processing volumes no account manager can match. Hand those over.
What Meta cannot judge is whether the lead was any good. It only knows what you tell it. If your conversion event is a form submission, the algorithm optimises towards people who fill in forms, which is a different population from people who buy. Feeding qualified-lead or closed-won data back through the Conversions API changes what the system chases. This single change usually does more for lead quality than any targeting adjustment, and it is the piece most UK accounts skip.
Set a testing window before you judge anything. Meta’s learning phase needs roughly 50 conversions per ad set per week to exit, so a campaign generating 10 leads a week never stabilises and its results are noise. If your volume is low, consolidate ad sets rather than splitting them. Our social media marketing work almost always starts with account consolidation for exactly this reason.
The takeaway on Meta automation
Automate the mechanics, own the definition of success. Meta’s AI optimises brilliantly towards whatever target you give it, which makes your conversion event and your feedback loop the most important settings in the account.
Getting the automation balance right for your account
We treat Meta’s automated products as useful defaults rather than a strategy. Before switching an account to Advantage+, we check whether the tracking can tell good leads from bad, because handing control to a system fed poor data produces cheap, useless volume at speed. That audit takes a week or two and tends to reveal tracking gaps that predate any campaign decision.
Expect a realistic timeline. A restructured Meta account typically needs 4 to 6 weeks of stable spend before the data means anything, and longer if lead volume is modest. We would rather tell you that upfront than report on a fortnight of learning-phase noise.
Questions about handing campaigns to Meta AI
Should I switch everything to Advantage+ campaigns?
Not without testing. Run Advantage+ against your existing structure with the same budget and compare on qualified leads rather than cost per lead, because the two metrics often disagree.
Will automation make my ads cheaper?
Automation frequently lowers cost per result by finding cheaper placements, but cheaper results only help if the leads still convert. Judge it on pipeline, not platform reporting.
How much control do I actually lose?
You keep budget, conversion event, creative assets and audience exclusions. You give up granular placement, bid and delivery decisions, which are the areas where the machine genuinely outperforms manual management.
Thinking through your Meta setup
If you are weighing up how much of your Meta advertising to automate, it usually helps to look at your tracking and account structure first. You are welcome to get in touch for a conversation about how your account is currently set up and what a sensible test would look like.
Written by the London Marketing Company team.
Source: https://www.socialmediaexaminer.com/meta-wants-to-run-more-of-your-marketing-but-should-they/
