What Is the Meta Andromeda Update, and How Should eCommerce Brands Adapt in 2026?

Meta changed how ads are selected before they reach the auction. Here’s what the technology actually does, what’s new in October 2026, and how to build a better Facebook and Instagram ad strategy.

A Meta campaign that worked for months suddenly slows down. You change the audience, adjust the budget, and try a slightly different headline. Nothing seems to stick.

It is tempting to blame the algorithm. But there is a more useful question: Are you giving Meta genuinely different reasons to show your products to different people?

Meta Andromeda is the AI-powered ad retrieval system that selects which eligible Facebook and Instagram ads move forward for more detailed ranking. It works before the final ad-selection stages, helping Meta identify relevant candidates from an enormous pool. As Meta’s advertising technology evolves, eCommerce brands need to rethink creative variety, conversion data, and how they evaluate results. Andromeda is not proof that targeting is dead, or that every advertiser needs a fixed number of ads.

The latest development is particularly important. On October 28, 2026, Meta described a new generation of Andromeda that evaluates vastly more potential ads while using richer signals to match people with relevant advertising. The implication for merchants is practical: Make your ads meaningfully different, give the system reliable signals, and let performance guide what you keep.

Key Takeaways#

• Andromeda handles retrieval, not the entire auction. It identifies promising ad candidates before later ranking stages.
• Meta upgraded Andromeda in October 2026. Meta says the new production system evaluates more than one billion eligible ads per second on a single GPU server.
• Creative diversity matters, but there is no official Andromeda creative quota. Test different customer motivations, not just cosmetic edits.
• AdScale’s analysis of 1,200+ eCommerce ad accounts found a useful four-to-six-creative range in the ad sets studied. It is an observed benchmark, not a universal platform rule.
• Strong creative needs accurate conversion data and sound economics. Judge results against purchases, customer acquisition cost, return on ad spend, and margin, not clicks alone.

What Is Meta Andromeda, Exactly?#

Andromeda is Meta’s personalized ads retrieval engine for Facebook and Instagram. Meta first explained the system in its December 2024 engineering announcement.

Think of Meta’s ad delivery as a sequence, not one mysterious algorithm:

1. Eligibility: Meta considers which ads can be shown under relevant targeting, policy, and delivery conditions.
2. Retrieval: Andromeda identifies a smaller set of promising ads from the eligible pool.
3. Ranking and delivery: Later models evaluate candidates in more detail to determine which ads are shown.

Meta’s original engineering explanation said retrieval narrowed tens of millions of candidate ads to a few thousand for further consideration. That distinction matters. An ad cannot win a later selection process if it never becomes a candidate.

But don’t confuse retrieval with a public checklist of creative rules. Meta has not published a rule saying that a particular image crop, headline change, or number of ads automatically wins or loses Andromeda retrieval.

Does Andromeda replace audience targeting?#

No. Meta’s October 2026 technical explanation explicitly describes index matching that enforces advertisers’ targeting rules. Audience choices still affect eligibility, while Meta’s automation and models influence matching and delivery within those boundaries.

For many eCommerce advertisers, this shifts the work away from endlessly subdividing audiences and toward a healthier combination of broad tests, customer insights, creative variety, and reliable measurement. But location restrictions, exclusions, remarketing goals, and other business requirements can still justify more specific audience settings.

What Changed in the October 2026 Andromeda Update?#

On October 28, Meta published a new engineering report on Andromeda. This was a technical infrastructure update, not a new Ads Manager setting that merchants need to switch on.

Three details stand out:

1. Andromeda can consider far more ads. Meta reports that its new production pipeline scores eligible ad candidates at more than one billion ads per second on a single GPU server, roughly 100 times the throughput of its 2024 system. That is an engineering throughput measure, not a promise that an individual campaign will reach 100 times as many shoppers.

2. Matching uses richer signals. The newer architecture incorporates signals such as sequences of ad engagement, user cohorts, contextual information, and semantic features. In plain English, Meta is developing more sophisticated ways to connect an ad with the person likely to find it relevant.

3. Meta reports a conversion gain. According to Meta, Andromeda improvements delivered a 2.4% increase in Facebook conversions since 2025. This is a Meta-reported platform-level result, not a guaranteed improvement for every advertiser or a result measured by AdScale.

The new architecture also draws on what Meta calls a mixture-of-experts model. You don’t need to redesign your campaigns around that term. What matters is that Meta is investing in more capable matching across a rapidly growing volume of ad creatives.

What the 2026 update does not prove#

The announcement does not establish that ads with similar visuals are literally merged into one ad. It does not prescribe four, six, ten, or twenty creatives per ad set. And it does not mean a business can ignore pricing, offers, landing pages, tracking, or the audience it is trying to reach.

Those distinctions make your creative strategy more useful, not less. They keep you focused on levers you can actually test.

Why Does Creative Diversity Matter More Than Cosmetic Variations?#

Imagine you sell a travel bag. You publish six ads using the same product photo, changing the background color and headline each time. Technically, you have six assets. From a customer’s point of view, you may have only one idea.

Now imagine three ads instead:

• The practical angle: Fits under an airline seat without checking a bag.
• The organization angle: Keeps a laptop, charger, and essentials easy to reach.
• The gifting angle: A useful gift for someone who travels every month.

These are three different buying motivations. Each gives a shopper a different reason to care.

A creative is a file. A concept is a reason to buy.

That is the distinction worth building your testing around. Changing a crop can be useful for placement. Testing a new benefit, use case, proof point, or offer can teach you more about what customers respond to.

This is an AdScale creative-planning recommendation, not a disclosed Andromeda scoring formula. Meta’s technical writing describes more sophisticated matching; it does not publish an exhaustive list of how visual similarity affects candidate retrieval.

What Does AdScale’s eCommerce Data Say About Creative Volume?#

There is another trap: assuming that because creative diversity matters, adding more ads always improves results.

In AdScale’s published analysis of more than 1,200 eCommerce ad accounts, ad sets with four to six active creatives were associated with 22% higher click-through rates than those running one or two. The analysis also reported that beyond seven creatives, CTR plateaued while CPM rose 8% to 12%.

These are observational results from the accounts studied, not a controlled test proving that creative count alone caused the differences. They should not be presented as a Meta requirement or as a guarantee that the same range is optimal for every budget and campaign.

The practical lesson is more measured: a small advertiser should not feel obligated to upload dozens of ads just because Meta can evaluate an enormous pool. Test enough meaningful ideas to learn, but avoid building a creative library so large that you cannot judge which messages drive business results.

For the study’s methodology and the full breakdown, read How Many Ad Creatives Should You Run Per Ad Set on Meta?. This article is about adapting to Andromeda; that one is the deeper guide to creative count.

How Should eCommerce Brands Adapt Their Meta Ad Strategy?#

You do not need to start your ad account over. Start by making the relationship between your customers, creative, and measurement more deliberate.

1. Audit your ads by buying motivation#

Open your current ad library and group assets by the reason a customer would buy. Product benefits, social proof, price, convenience, gifting, and real-world demonstrations can all be distinct angles. If several ads deliver exactly the same promise, ask what new information each one is adding.

2. Build a concept-first creative testing plan#

Pick a product or collection and write three different customer motivations before making new assets. For example, a skincare product might be positioned around a simple routine, a specific ingredient, or customer testimonials. Produce assets that communicate those ideas clearly rather than defaulting to recolored templates.

3. Match the format to the story#

Use video when movement, demonstration, or a real customer story adds value. Choose a static image when one strong product benefit can be understood immediately. Use a carousel when comparison or a sequence helps explain the product. There is no evidence that Andromeda requires every brand to prioritize one format exclusively.

4. Test audience automation without abandoning business constraints#

Where appropriate, test broader audiences and Meta’s Advantage+ capabilities against your current setup. Keep relevant location restrictions, exclusions, and other guardrails. Avoid changing targeting and creative simultaneously if you need to understand which adjustment affected results.

For the wider relationship between Meta’s automation and Google Ads, see AdScale’s guide to Google and Meta automation for eCommerce.

5. Check conversion tracking before scaling#

Verify that your Meta Pixel and Conversions API setup records the correct events and purchase values, with deduplication where applicable. Check purchase events against your store’s orders. Better ad matching is not a substitute for trustworthy performance data.

6. Refresh when the evidence calls for it#

Watch creative-level spend, frequency, click-through rate, cost per purchase, and ROAS alongside actual store sales. A change in results could reflect creative fatigue, but it could also reflect seasonality, price, promotions, stock, or a landing-page problem. Review on a regular cadence, then replace or add concepts based on what you find. Avoid treating an arbitrary two-week expiration date as an Andromeda rule.

7. Judge creative by business outcomes#

A high CTR is a clue, not the finish line. One ad might attract plenty of inexpensive clicks yet generate fewer profitable orders. Compare spend, conversions, revenue, acquisition cost, and contribution margin where available. If a campaign performs well on Meta but the store’s overall efficiency is declining, investigate before increasing the budget.

How AdScale Helps eCommerce Brands Put This Into Practice#

The challenge is not just understanding Andromeda. It is finding the time to produce meaningful creative, launch campaigns, check performance, and make budget decisions while running an online store.

AdScale helps connect those tasks. Its advertising platform works with eCommerce store data and manages campaigns across Meta and Google. Instead of treating each ad as a disconnected design asset, store owners can develop creative around the products, promotions, and business priorities that matter to them.

Create ad images and videos around real products#

AdScale’s Agentic Ad Creatives gives merchants AI-assisted image creation and editing within their advertising workflow. Video ad creation also helps turn product assets into formats suitable for Meta placements. The strategic choice still belongs to you: which customer problem, selling point, or offer should the new ad communicate?

You can explore the creative workflow in AdScale’s guide to Agentic Ad Creatives.

Connect advertising decisions to first-party store data#

AdScale integrates store-level information such as products and orders into its advertising workflow. That gives merchants a more business-focused context for choosing what to promote, rather than assessing ads only through engagement metrics.

Manage Meta and Google in one place#

Andromeda affects Meta ad retrieval, but customers do not shop in one channel. Someone might discover a product on Instagram and later search for it on Google. AdScale gives merchants tools to manage both advertising channels and make budget decisions using performance signals across them.

AdScale does not control Andromeda or guarantee that Meta will select a particular creative. It helps eCommerce teams handle the work around the algorithm: developing creative, operating campaigns, and reading performance in the context of their store.

Explore AdScale’s AI advertising platform.

A Simple Example: Three Concepts, One Product#

Suppose an online homeware store wants to promote a cordless table lamp. Instead of making twelve nearly identical product shots, the team develops three testable concepts:

This is an illustrative example, not a claimed customer result. The advantage is that each concept tests a real purchase motivation. Once results arrive, the team has a better question to answer than “Which background color won?”

Frequently Asked Questions#

What is the Meta Andromeda update in simple terms?

Meta Andromeda is the AI-powered ads retrieval system used in Facebook and Instagram advertising. It narrows a large pool of eligible advertisements into a smaller set for detailed ranking and delivery. Meta introduced the system in 2024 and announced major technical improvements in October 2026. Advertisers do not activate Andromeda manually.

Does Meta Andromeda mean audience targeting is dead?

No. Andromeda improves automated ad matching, but Meta’s own technical explanation says the retrieval system still enforces advertiser targeting rules. Broad targeting and Advantage+ can be useful tests for eCommerce brands, while location, customer exclusions, remarketing, and other constraints can remain relevant. Creative quality does not eliminate the need for audience strategy.

How many creatives should I run per ad set after Andromeda?

Meta has not published an official Andromeda creative-count requirement. AdScale’s observational analysis of more than 1,200 eCommerce accounts found favorable results among ad sets running four to six active creatives. Use that finding as a testing reference, not a universal cap. The right number depends on your budget, campaign structure, and meaningful concept variety.

How often should I refresh Facebook and Instagram ad creatives?

There is no universal refresh deadline built into Andromeda. Review performance regularly and look for declining results, increasing frequency, weakening purchase efficiency, or new customer insights. Refresh when the evidence supports it. Some ads remain effective for longer than a few weeks, while others need earlier changes because of offers, seasons, or audience response.

Does Andromeda favor video ads over static images?

Meta has not published an Andromeda rule requiring video rather than static images. Different formats can help communicate different product stories and work in different placements. Test demonstrations, testimonials, comparisons, and clear product visuals based on your customers and offer. Decide which formats to scale using purchase performance, not assumptions about the retrieval algorithm.

The Takeaway: Better Ideas, Better Signals, Better Decisions#

Andromeda has changed the machinery behind Meta ad delivery, and the October 2026 update shows how quickly that machinery continues to improve. But it has not changed the basic job of a good ad: give the right customer a clear reason to buy.

If your team keeps producing minor variations of the same creative, more assets might not mean more learning. If tracking is unreliable, you may misread the results. And if you optimize only for clicks, you can win attention without growing the business.

Instead, start with three different customer motivations. Build creative that expresses each one. Check that purchase data is accurate. Watch results long enough to learn, and let those results decide what happens next.

Meta can get better at finding potential matches. Your job is to give it something worth matching.

More files aren’t the goal. Better reasons to buy are.

Keep Learning#

• How Many Ad Creatives Should You Run Per Ad Set on Meta? — AdScale’s account-level analysis of creative volume and performance.
• Agentic Ad Creatives for eCommerce — How merchants can develop image creative inside AdScale.
• The Full Guide to Google and Meta Automation for eCommerce — How automation and store data fit across both ad platforms.

About the author: Laili Shalom is Head of Content at AdScale, where she covers eCommerce advertising, first-party data, and practical applications of AI in Google and Meta campaigns.