Meta Andromeda and Creative Volume: Why the Algorithm Now Wants More Ads, Not Better Targeting
TL;DR
Meta Andromeda rewards creative volume over targeting precision. Here is how many ad concepts you actually need per month and how to build that pipeline.
Clients often ask us to fix their targeting when their Meta account underperforms, and increasingly the honest answer is that targeting is not the bottleneck anymore - creative volume is. I am Manav Gupta, and since Andromeda became the default retrieval engine across most ad accounts, the accounts we see win are consistently the ones feeding the system more distinct creative, not the ones with the cleverest audience setup.
The citable answer: Meta's Andromeda retrieval engine can evaluate vastly more ad candidates per auction than the old system, which means accounts running 15-30 genuinely distinct creative concepts a month consistently outperform accounts running 3-5 - creative variety, not audience precision, is now the primary lever for delivery efficiency. Here is what that means operationally, how much volume you actually need, and how to build a pipeline that produces it without burning out your creative team.
Why Volume Became the Lever
Under the old delivery model, the auction mostly decided a winner from a small pool of ads you had manually targeted to a narrow audience. Andromeda inverts that: it retrieves from a much larger candidate pool and matches ads to micro-intents you would never think to build a targeting rule for. If you only give it 3 creative concepts, it can only ever match those 3 concepts to whatever intents it finds - no matter how good the retrieval engine is, it cannot manufacture variety you did not supply.
This is why two accounts with identical budgets and similar products can see very different delivery efficiency: the one feeding a richer candidate pool gives Andromeda more surface area to find profitable matches. We've audited accounts spending the same daily budget on nearly identical products, where the only structural difference was creative count - and the account with 4x the creative volume consistently showed 20-30% better cost-per-purchase, holding targeting and bid strategy constant.
How Many Concepts You Actually Need
| Account stage | Concepts per month | Why |
|---|---|---|
| New account / validation | 10-15 | Establishing which angles and hooks resonate at all |
| Scaling | 15-25 | Feeding a larger candidate pool as spend and audience size grow |
| Mature, high-spend account | 25-40+ | Fatigue sets in faster at volume; constant refresh needed to sustain reach |
A "concept" here means a genuinely different angle - different hook, different proof point, different format - not five colour variations of the same script. Andromeda's candidate pool benefits from real variety; five near-identical edits of one ad barely expand what the system can match against. We define "genuinely different" internally as passing a simple test: if you removed the branding, could someone tell these two ads apart based on the argument they're making, not just the visuals.
Building a Creative Engine, Not a Creative Department
The operational shift most agencies and in-house teams have not made yet is treating creative as a continuous production pipeline rather than a project. We run this in three parts for every creative strategy engagement:
- Angle bank. A running list of 20-30 distinct customer pain points, objections, and desires, refreshed monthly from actual customer reviews, support tickets, and comment sections - not brainstormed in a vacuum. The best angle banks are living documents, not a one-time brainstorm session that gets referenced for a year without updates.
- Format rotation. The same angle expressed as UGC testimonial, founder-led, demo, and static carousel. Format diversity within an angle multiplies the effective candidate pool without needing new ideas - a single strong angle can realistically produce 4-5 distinct pieces of creative just by varying who delivers it and in what format.
- Win-rate tracking. Every concept gets tagged by angle and format, and we track which combinations Andromeda keeps pulling into delivery weeks after launch. That data feeds directly back into the next month's angle bank, so the system compounds rather than resetting from zero each month.
The Trap: Volume Without Structure
Some accounts respond to "Andromeda wants volume" by producing 30 disconnected ads with no organizing logic, and burn budget without learning anything. Volume only compounds when it is structured - each concept tagged to an angle and format so you can see patterns in what wins, rather than 30 one-off bets you cannot learn from. The goal is a growing angle bank that gets smarter every month, not a pile of ads.
We've seen this trap specifically hurt agencies trying to satisfy a client's demand for "more creative" without a system behind it - the client sees a higher volume of ads going live, feels reassured, but nobody on the team can actually explain six months later which angles are working and why, because nothing was tagged or tracked in a way that let the account learn from its own history.
Signal Still Matters, Just Differently
Creative volume works best paired with clean first-party signal via Conversions API. Andromeda uses that signal to decide which of your many concepts to match against which intent - so volume without signal quality is like giving the engine more options but worse information to choose between them. The two levers reinforce each other; neither substitutes for the other. An account with excellent creative volume but poor event match quality will still underperform an account with moderate volume and clean signal, because the retrieval engine simply has less to work with in deciding who to show which ad to.
A Worked Example: Sizing an Angle Bank From Scratch
When we start a new creative engine for a client with zero existing structure, the first step is pulling raw material, not writing ad copy. For a typical D2C brand, that means reading the last 100-150 customer reviews (both positive and negative), scanning support ticket themes from the last quarter, and checking comment sections on any existing organic or paid content for recurring objections or praise. This usually surfaces 15-20 genuinely distinct angles within a day or two of research - far more than most brands assume exists, because most brands have never actually mined their own customer feedback systematically for creative material.
From there, each angle gets tested in its cheapest-to-produce format first (usually a UGC-style video or simple static ad) before investing in a more expensive format for that same angle. This keeps the initial validation cost low: of 15-20 angles tested cheaply, typically 4-6 show real signal within the first two weeks, and those are the ones that earn a bigger production budget and format variety in the following month.
How Long Before You See Results From Higher Volume
Clients often expect an immediate CAC improvement the week creative volume increases, and the honest timeline is closer to 3-6 weeks. The first week or two is mostly the retrieval engine building up enough data on the new concepts to know which ones to scale - during this window, cost metrics can actually look worse than the old, smaller creative set, simply because the algorithm is exploring rather than exploiting known winners. Accounts that panic and cut back volume during this exploration phase usually never see the payoff, because they revert before the system has had time to learn.
By week four to six, if the angle bank and format rotation were built with real variety, the account typically settles into a meaningfully better cost structure than before - not because any single ad is a breakthrough, but because the retrieval engine now has enough genuine diversity to match against a wider range of buyer intents than the original small creative set ever could.
A Real Example
A D2C supplements brand came to us running 4 static ads for six months with a plateaued CAC. We rebuilt their creative pipeline to 18 concepts a month across 5 angles (label-reading skepticism, before/after routine, founder story, dosage-confusion objection, and gifting) each in 3-4 formats. Within eight weeks, blended CAC dropped 24%, driven almost entirely by Andromeda finding profitable micro-audiences the old 4-ad set never reached - no targeting change was made at all.
The founder's initial reaction to the plan was skepticism about production capacity - the brand had no in-house creative team and assumed 18 concepts a month meant an unaffordable production budget. In practice, most of the volume came from UGC-style content sourced from micro-creators at a fraction of studio production cost, which is exactly why format diversity within an angle matters as much as the number of angles itself.
FAQ
How much creative volume does Meta Andromeda actually need?
As a baseline, most accounts need 15-25 genuinely distinct creative concepts per month to give the retrieval engine enough variety to match against different micro-intents. Smaller, newer accounts can start around 10-15; larger accounts often need 25-40+ to sustain reach without fatigue.
Does more creative volume mean lower quality is acceptable?
No - volume and quality are not a tradeoff you have to accept. The goal is genuine variety in angle and format, not quantity for its own sake. Ten distinct, well-made concepts targeting real objections will outperform thirty near-identical variations of one ad.
Is creative volume more important than targeting now?
For most Andromeda-era accounts, yes. Broad targeting paired with high creative variety consistently outperforms narrow targeting with limited creative, because the retrieval engine needs a rich candidate pool more than it needs a narrow audience definition.
Where do I find enough distinct angles to hit these volume targets?
Customer reviews, support tickets, and comment sections on existing content are usually a far richer source of genuine angles than internal brainstorming. Most brands are sitting on 15-20 untapped angles in their own customer feedback that nobody has systematically mined yet.
How do I avoid producing volume without any real strategy behind it?
Tag every concept by angle and format from day one, and review win-rate data monthly to see which combinations Andromeda keeps scaling. Without this tracking, volume becomes a pile of disconnected bets instead of a system that gets smarter over time.
Build a Creative Pipeline That Feeds Andromeda
If your Meta account has been running the same handful of ads for months, the fastest fix is usually not a targeting change - it is a real creative production pipeline. Book a call with Balistro and we will build you an angle bank and format rotation that keeps the algorithm fed with genuine variety, not just more ads.
