What Andromeda actually changed

Meta's rollout of Advantage+ over the last 18 months is powered by an algorithm internally codenamed Andromeda. This isn't marketing speak for a rebrand. It's a fundamental retraining of how the delivery system finds the right person for your ad.

Under the old model, your targeting setup was the dominant signal. Interest layers, custom audiences, lookalike percentages. They shaped who saw the ad. Creative mattered, but targeting mattered more.

Under Andromeda, the calculus has flipped. The algorithm is now confident enough in its ability to find your buyer from a broad pool that narrow targeting often underperforms broad audiences running the same creative. Your targeting work is, increasingly, noise. The algorithm knows who to show the ad to. What it doesn't know is whether your creative is any good.

As Thomas likes to put it: "Targeting doesn't matter anymore. Creatives all the way."

This has a specific operational consequence. The time your team spends narrowing audiences, layering exclusions, and testing lookalike percentages produces sharply diminishing returns. The time they spend briefing, iterating, and shipping creative is where performance actually lives now.

Set your audience controls at the account level, then step back

Here's the framework we've rolled out across our client accounts. Stop doing targeting work inside campaigns. Do it once, at the account level.

At the account level, set:

• Excluded audiences. Existing customers, people who've already converted, audiences you never want to reach.

• Returning visitor definitions. So retargeting and acquisition campaigns don't overlap.

• Geographic caps. Relevant for any business with a physical delivery radius or service area.

• Brand safety controls. Placements, topics, publisher blocklists.

Once these are set, the campaign level audience should be as broad as your business logic permits. Let the algorithm pick from that pool.

What this replaces: daily and weekly tinkering with interest layers, 1% versus 3% lookalike A/B tests, custom audience slicing. These were necessary in 2022. They're noise in 2026.

What this frees up: roughly 40% of a performance manager's week. That time should be reinvested into creative briefs, reviews, iteration, and hook testing.

Why authentic creative now outperforms polished visuals

When targeting isn't the lever, creative is. And the creative that wins has changed dramatically.

We run A/B tests continuously across our client accounts. The pattern repeats month after month. Authentic, UGC style content (camera on your head videos, creator led walkthroughs, unpolished product demonstrations) outperforms expensively produced studio creative.

Thomas describes it like this: "A camera on your head, telling your own experience. That works a hundred times better than polished visuals."

It's true across:

• Ecommerce (ICP: E-com Eddie). "How I use it" videos and unscripted product demos outperform studio product shots.

• B2B SaaS (ICP: B2B Bruno). A founder or CSM explaining a feature into a phone camera outperforms a polished product video.

• Services and real estate (ICP: Growth Greta). Agent led and operator led walkthroughs outperform brochure style visuals.

The economics are doubly painful for the old model. Authentic creative costs less to produce (often 5 to 10 times less) and performs better. One well executed authentic video can replace a quarter of polished production budget and deliver stronger ROAS. Thomas's rule of thumb: "With one good video, you can earn ten back."

Two reasons this is happening:

1. Feed training. Users are trained by social feeds to scroll past anything that looks "designed". The brain has learned that signal equals ad. Authentic content reads like a post from someone you know.

2. Creative velocity. Creative fatigue on Meta now hits faster than it used to, closer to 22 days than the 30 to 35 of two years ago. Polished production pipelines can't keep up. Authentic pipelines can.

Where brand discipline still matters: the exceptions

The rule isn't universal. In some categories, brand authority and trust signals do work that authenticity can't replace.

Banking and insurance. Regulated categories where users are evaluating institutional trust. Logos, brand colors, formal address (especially "u" in Belgian Dutch), typefaces. These carry real weight. Authentic UGC here can actually underperform because it signals informality in a category where formality is a trust cue.

Hanneke says it bluntly: "In banking and insurance you need the authority of the brand to come through. Logo, recognizable colors, formal tone. That's where branding still wins."

Luxury and regulated pharmaceuticals. Similar logic. Brand coherence carries more than creative velocity.

For everyone else (most B2B, most D2C, most services, most real estate), the performance data is clear. If you haven't tested authentic creative against polished creative in the last six months, you're flying blind on one of the two biggest levers in your account.

The internal tension: performance versus branding

This shift creates a political problem. Your brand and creative team spent years building visual guidelines. Your performance team now needs to ship content that doesn't fully respect them. Someone has to resolve this.

The wrong answer is to have one team win and one team lose. The right answer is to redefine the creative brief so both teams are working from the platform's reality.

Our recommended split:

• Brand guidelines govern owned channels. Website, email, above the fold hero content, brand films.

• Paid social creative briefs start from the platform, not the brand book. Platform conventions win over brand conventions inside the feed.

• Brand and performance share ownership of a "UGC brand frame". Which visual elements must appear (logo placement, one dominant color cue, a specific product angle) and which are optional. This gives performance room to ship while keeping enough coherence that the paid feed still reads as the brand.

This is a conversation most agencies dodge because it's politically expensive. It's also the conversation that most determines whether your 2026 paid social performance goes up or sideways.

How to restructure your paid social team

Four concrete moves, in priority order:

1. Audit your account level audience controls. Before any new campaign, verify your exclusions, returning visitor definitions, and geo caps are set at the account level. If your team is still setting these per campaign, consolidate this week.

2. Reassign your performance manager's week. The old split was roughly 60% targeting, 40% creative and reporting. The new split should be closer to 20% targeting, 60% creative, 20% reporting. If your agency's weekly report still leads with audience refinement updates, that's a signal they haven't made the shift.

3. Build a UGC pipeline. Identify two or three creators you can brief within a week. Build a backlog of 4 to 6 creative concepts per month. Accept that creative is now a consumable, not a permanent object. The budget question shifts from "how good is this ad?" to "how fast can we ship the next three?".

Run a polished versus authentic A/B test this month. Take one active campaign. Produce one authentic UGC style video against your best performing polished creative. Run them at equal budget for 14 days. Let the data drive the conversation with your creative team, not opinion.