Automated Delivery ≠ Automated Strategy
Why It Matters (For Busy Founders & CEOs)
"Can’t I just let Meta's AI run this myself now?" It's a fair question to ask before you hire, renew, or cut a media partner—here's the honest answer.
Andromeda + GEM automated delivery: bidding, placement, sequencing. They did not automate strategy: personas, messaging, creative direction, data integrity, interpretation.
The algorithm didn't reduce how much creative you need. It increased it—and raised the bar on how targeted each piece has to be.
This changes what you should be evaluating in an agency or hire—not "can they run ads," but “are they directing the algorithm, or just watching it."
Getting this wrong shows up as a resourcing decision, not a tactics problem: you either overpay for babysitting, or underinvest in the strategy the algorithm actually needs.
The Question Behind the Question
“Can't I just let Meta's AI handle this myself now?”
It's usually really a budget question. Andromeda and GEM automated a huge chunk of what used to require someone in Ads Manager every day—so it's reasonable to wonder whether you still need to pay for that expertise, or whether the platform has made it redundant.
Here's the honest answer: Meta automated delivery. It did not automate strategy. Which means the question isn't "human or AI." It's what you're actually paying for now, and whether whoever's running your account (in-house or agency) understands the difference.
Credit Where It’s Due
Andromeda and GEM are genuinely powerful. The system now handles bid and budget allocation, placement optimization across Facebook, Instagram, and Messenger, and—with GEM—sequencing: figuring out what order of ads is most likely to move a specific person from first impression to purchase.
That's real, and it means the manual, hands-on-keyboard work of media buying is worth less than it used to be. If you're paying someone primarily for that, you're paying for something the platform now does on its own.
For a full breakdown of what Andromeda and GEM actually changed under the hood, Foxwell Digital's rundown is a great deep dive.
Where the Real Work Moved: Strategy, Not Delivery
Here's the part that surprises most founders: automating delivery didn't shrink the strategic workload. It shifted it upstream, and made it bigger. Andromeda and GEM are only as good as what they're fed. If the inputs are shallow, the algorithm optimizes a shallow account brilliantly. The work that used to happen inside Ads Manager now has to happen before anything gets uploaded.
That work breaks down into four areas, and all four have gotten more demanding, not less.
1. Personas, Built From Real Buyers—Not Assumptions
Andromeda treats creative as targeting. That only works if the personas behind the creative are real. "Women 25-45 interested in home decor" isn't a persona, it's a demographic, and the algorithm has no use for it.
A real persona is built around a specific buyer's situation, hesitation, and motivation: the first-time buyer comparing you to three competitors, the repeat customer who needs a reason to upgrade, the gift-buyer who's not even sure this is the right category. Each one needs different creative because each one is at a different point in the decision. Skip this step and you're back to "more creative" with nothing distinct about any of it—volume without direction.
2. Messaging Pulled From What Customers Actually Say
The fastest way to build a persona-driven angle that actually converts is to stop guessing at language and start mining it. Product reviews, PDP Q&As, customer service transcripts, even return reasons—this is where the real objections, hesitations, and trigger phrases live, in the customer's own words, not a copywriter's assumption of them.
This matters more under Andromeda because the algorithm is testing angles at a much larger scale, much faster. If your input angles are generic, you're not testing messaging, you're testing noise. Review-mined messaging gives the algorithm something real to differentiate between.
3. Format Diversity, Not Just Message Diversity
GEM orchestrates across placements and formats as one connected system, which means static, video, carousel, and UGC aren't interchangeable ways of saying the same thing anymore. They're different entry points into the same sequence, and the algorithm needs variety in how something is said, not just what's being said.
A brand running the same message as five near-identical video cuts isn't diversifying. A brand running that same message as a UGC testimonial, a static product-in-context shot, a founder-voice video, and a customer-review carousel is giving GEM actual material to sequence against different people at different points in their journey.
4. More Volume Than Ever—On Purpose
This is the one that catches people off guard: Andromeda rewards accounts that produce genuinely new creative concepts continuously, not accounts that occasionally refresh a few top performers. Brands seeing the strongest results under this system are producing dramatically more net-new concepts per month than they were pre-Andromeda—not incremental tweaks to existing ads, but new personas, new angles, new formats.
That's not a "nice to have" anymore. Asset development is now one of the most resource-intensive parts of running Meta Ads well, and it's the part most likely to get cut when a brand assumes "the AI has it covered."
💡 Tool we lean on: With this much creative volume in play, you need a fast way to see what's actually working, what competitors are running, and where to iterate next—not just a spend dashboard. Motion is one of our go-to platforms for exactly that: creative performance research, competitive monitoring, and fast iteration cycles, all in one place.
What You Should Actually Be Evaluating Now
If delivery is automated, the value of a person or agency shifts entirely to the four things above. When you're deciding whether to hire, keep, or replace your media management, these are the real questions:
Do they have a real persona and messaging strategy, or just "more creative"? Ask to see the personas. Ask where the messaging came from. If the answer is "our best guess," that's a gap.
Are they mining actual customer language, or writing from assumption? Reviews, Q&As, support tickets—this should be a documented input to creative briefs, not an occasional afterthought.
Is format diversity part of the plan, or is it all one format with different messages? Static, video, UGC, carousel—each should show up with intention, not by accident.
Is asset volume increasing to match what Andromeda rewards, or shrinking because "the algorithm handles it"? If creative output has gone down since Andromeda rolled out, that's backwards.
Do they own data integrity, or just report on it? GEM runs on Pixel and CAPI signal. If nobody's proactively auditing that, the algorithm optimizes toward broken information with total confidence.
Did they design your account structure, or inherit it? Consolidation, exclusions, campaign architecture—these decisions determine what the algorithm has to work with in the first place.
Founder takeaway: If you can't get a clear answer to these six questions from whoever manages your account, you're likely paying for delivery you don't need and missing the strategy you do.
Where This Goes Wrong on Both Sides
Getting the resourcing decision wrong tends to look like one of two things:
Overpaying for babysitting. Someone's still logging in daily, "monitoring" an account the algorithm is already handling, without adding real strategic direction upstream. You're paying for a job that mostly doesn't exist anymore. (This is also usually the root cause of another expensive habit: pausing ads too early.)
Underinvesting in strategy. You cut spend on expertise because "the AI handles it," and nobody's building personas, mining messaging, diversifying formats, or increasing asset volume. The account runs, technically—just with thinner and thinner material for the algorithm to work with over time.
Either way, the cost isn't just budget. It's the gap between what the algorithm could do with rich, well-researched input and what it's actually doing without it.
The Actual Division of Labor
It's simpler than it sounds: the algorithm is the delivery engine. The person or team you pay is the strategist feeding it—and feeding it well now means persona development, review-mined messaging, format diversity, and a real asset pipeline, not a handful of ads refreshed once a quarter.
When you're structuring a team, a contract, or a hire, that's the split to build around: less time and cost on manual execution, meaningfully more investment in the upstream work the algorithm actually depends on.
The Real Question
It was never "human or AI." It's "who's directing the AI, and are you paying for the right part of the job." An account with a smart algorithm and no strategic direction behind it isn't more efficient. It's just automated in the wrong place—and you're likely paying for the wrong thing too.
👋 Not sure if your current spend on media management is buying strategy or just babysitting delivery? Get a free assessment and we'll tell you straight, no fluff.