Facebook Ads Not Performing? Fix It After the Andromeda Update
Want my team to build this for you? Work with Advergent. Rather build it yourself? Join the coaching waitlist.
If your Facebook ads are not performing the way they used to, the most likely cause is Meta's Andromeda update, which changed how the ad delivery algorithm ranks and matches creative. It now scans a far bigger pool of ads per auction and rewards accounts that feed it variety. If your creative library looks the same ad after ad, costs climb and results fade, no matter what you do to targeting or campaign structure.
Why did my Facebook ads performance suddenly drop?
Because the lever that used to move results, audience targeting and campaign structure, barely matters anymore. Meta's Andromeda system replaced the older ad delivery process that compared a few hundred ads per auction. Now it scans thousands, sometimes tens of thousands, of ads in seconds and matches each person to an ad based on their past behavior and interests, not just their stated interests or a lookalike audience.
Meta paired that change with Advantage+, which is on by default for most campaigns now. Advantage+ handles audiences, budgets and placements on autopilot, so the manual levers advertisers used to pull, tight interest stacks, CBO structuring, manual placement control, don't do what they used to. You can't really micromanage a campaign the way you could two or three years ago. The algorithm runs most of that process itself.
What is the Meta Andromeda update, and why did Meta build it?
Because AI-generated ad creative flooded the platform faster than the old ranking system could process it. Once everyone could produce hundreds of ad variations with AI tools, Meta's system got overloaded, and the old delivery process couldn't keep up. Meta rebuilt how it ranks, filters and recommends ads around creative diversity instead, specifically so your own ad variations stop competing against each other inside one account.
Andromeda started rolling out in mid-2025, and it rewards accounts that supply a wide range of ideas, messages and styles, because that variety gives the algorithm more options to match against each person it's deciding to show an ad to. Upload creative that all sounds and looks the same, and the system runs out of new material to learn from. That's the direct cause behind the CPM spikes and performance drops advertisers reported once the rollout hit, and it's still the operating reality now, not a temporary blip that's about to pass.
The takeaway: your targeting isn't the problem, and there's no meaningful CBO debate left to have. The conversation moved entirely to your creative inputs, because Andromeda wants more ideas, not more ad sets.
How do you fix Facebook ads performance after Andromeda?
You pick one of two paths and commit to it: flood the algorithm with volume, or feed it a deliberately diverse set of messages. Both work. The difference is speed versus longevity, and which one fits your team's actual capacity to produce creative without burning out.
| Path 1: Volume | Path 2: Message diversity | |
|---|---|---|
| The move | 30-50 new creatives a week | 15-20 unique creative concepts a month |
| Why it works | More variety gives Meta more combinations to test, so it learns faster and finds winners quicker | Each concept speaks to a different reason to buy, so it connects with a different pocket of your audience |
| Downside | Time-consuming, burns resources, creators burn out faster, shorter ad lifespan | Requires more strategic thinking upfront and a deeper read on customer psychology |
| Best for | Larger teams or agencies with a real production system | Smaller teams and solo operators |
Meta's own materials point at the volume path directly: they featured a case study of a brand that went from posting three to four ads a week to fifty a week, and performance returned to normal. If you have editors, creators and an AI-assisted workflow already in place, that's the faster route.
If you're a smaller team, message diversity gets you most of the same benefit without the production load. Let's say you sell an AI SDR tool for B2B sales teams. Instead of fifteen hooks about replacing human outbound with AI, you'd build one ad around AI and your sales team collaborating, one around fixing pipeline leakage, one around keeping your CRM updated automatically, and so on. Each ad tells a complete, different story about one selling point, not just a new opening line on the same pitch.
My honest answer, if you're not forced to pick one immediately: start with path two. Find the messaging that resonates first, then double down on what's working with volume once you know it converts. Testing intentionally before you scale is cheaper than scaling before you know what's worth scaling.
Whichever path you pick, the moment you stop feeding Meta new creative, performance drops again. This isn't a one-time fix, it's an ongoing production habit.
What does creative diversification actually mean?
It means variety across three dimensions at once, not just a bigger pile of ads. Different angles, different audience types, and different formats, mapped to where a prospect actually is in their decision.
Different angles. Each ad should argue a different reason to buy: problem and solution, transformation or benefit, a founder's story or perspective, social proof or a testimonial, or an offer with urgency behind it. For B2B specifically, the emotion behind each angle is usually a logical one, saving time, saving money, or making more money, and each version should speak to a different one of those.
Different avatars. Not every customer thinks the same way about the same product. Pull insights from your reviews, comments, emails and DMs to identify three to five audience types, then write creative that talks directly to each one's pain or job to be done.
Different formats. Mix UGC clips, testimonials, demos, VSLs or talking-head video, and static or carousel images. High-production video and low-fi native clips filmed in your office both have a place, because each format taps a different behavioral response: testimonials read as trustworthy, demos show the product itself, polished video builds authority, and static images do the work of retargeting people who are already interested.
You can also map creative to awareness stage, using Eugene Schwartz's model as a guide:
| Awareness stage | What the ad should do |
|---|---|
| Unaware | Educate on the problem itself |
| Problem aware | Highlight the pain point directly |
| Solution aware | Introduce your method or mechanism |
| Product aware | Show proof and make the offer |
Covering all four stages in your creative mix gives Meta a match for wherever someone actually is in their buying journey, instead of forcing every viewer through the same pitch.
How should you structure your ad account for Andromeda?
Keep it simple: one campaign per core offer, broad targeting, and two lanes inside it. A testing campaign absorbs new concepts, and an evergreen campaign holds the winners that keep performing. That structure gives Meta enough data density in one place to learn fast and keep delivering cheaper traffic.
Before you build a single new ad, build a creative portfolio instead of a pile of one-off ads. Treat your account as a set of ideas, not isolated creative: once you have 10-15 strong concepts that each represent a different story or motivator, Meta can rotate them intelligently instead of burning through one idea in a week. Rotate them by producing drastically different versions of each concept, across the angles and formats above, rather than small variations on the same one.
Then protect a consistent cadence. Whether that's 30-50 ads a week on the volume path or 15-20 concepts a month on the message diversity path, set a production schedule and hold it. Scaling your Meta ads only really works once this creative pipeline exists underneath it, because you can't scale spend into an account that's starving for new material.
What backend metrics should you watch after the update?
Watch new customers against repeat buyers, because Andromeda optimizes for conversions, not specifically for new customer acquisition. If you notice new customer volume dropping while repeat buyers or existing site visitors keep converting, add exclusions for past purchasers and past site visitors so the algorithm has to go find new people again.
This matters more now because Andromeda and Advantage+ will happily keep your conversion numbers looking fine by re-converting people who were always going to buy, while your actual growth stalls. Always optimize for what your business needs, a new customer, a booked call, real revenue, not for whatever definition of "conversion" is easiest for Meta to hit. If cheap volume that doesn't turn into real pipeline is the pattern you're seeing under Advantage+, my post on why Meta leads look bad covers that problem specifically.
What to do first
Start here, in order:
- Audit how many genuinely different creative concepts, not just hook variations, are live in your account right now. If it's under ten, that's very likely your ceiling.
- Pick a path. Volume if you have production capacity, message diversity if you don't. Don't try to do both half-heartedly.
- Map your next batch of creative against angles, avatars and awareness stages, not just "let's try another UGC video."
- Restructure into one campaign per offer, with a testing lane and an evergreen lane.
- Check new-customer volume against repeat buyers monthly, and exclude past converters if the split is drifting the wrong way.
If your team is small and this feels like a lot to run alongside everything else, scaling Meta ads on a smaller budget walks through how to sequence it without overextending. And if you'd rather have me and my team build and run this creative system for you, book a call at advergent.co/apply.
FAQ
Why did my Facebook ads stop performing?
Because targeting and campaign structure barely move results anymore. Meta's Andromeda system matches each person to an ad based on the creative, so accounts running the same few ideas lose delivery to accounts with diverse creative.
How many ad concepts does Andromeda need?
Aim for 10 to 15 strong concepts that each represent a different story or motivator, either through volume at 30 to 50 ads a week or through diversity at 15 to 20 distinct concepts a month.
How should I structure my account after Andromeda?
One campaign per core offer on broad targeting, with a testing lane for new concepts and an evergreen lane for winners.