What Advantage+ and Advantage Actually Mean

Meta uses two related names for two different tiers of automation. "Advantage+" (with the plus sign) names the fully automated campaign types — Advantage+ Sales and Advantage+ App — where the algorithm controls audience selection from a broad pool, placement across all Meta surfaces, budget distribution across ad sets, and creative selection together. "Advantage" (no plus) names the individual feature enhancements that attach to an otherwise manual campaign: Advantage campaign budget, Advantage+ audience, Advantage+ creative, and Advantage+ detailed targeting. These are independent toggles, not a unified system.

The distinction matters because conflating the two leads to a specific and expensive mistake: enabling an Advantage+ campaign type when the advertiser only wanted budget optimization across ad sets. Advantage campaign budget — the former CBO — gives the algorithm control of budget distribution. An Advantage+ Sales campaign gives the algorithm control of budget distribution and audience selection and placement and creative simultaneously, with demographic and interest targeting available only as suggestions, not hard constraints. Location, minimum age, language, and exclusions are the only true hard limits that remain. Those are not the same lever.

The most useful mental model: Advantage+ campaign types are the whole car driving itself. Advantage feature enhancements are cruise control on a car you are still steering. Meta's Advantage+ campaign experience flow, introduced in 2025, defaults new campaign creation to automation. Knowing which product is actually running requires reading past the interface nudge — and auditing which toggles are on is the prerequisite for any optimization work. Several enhancements, including text improvements and image expansion under Advantage+ creative, are on by default and require active opt-out at the ad level. Sources: Meta Business Help — Advantage+ campaign experience; Prism Digital, Advantage+ features explained 2026.

Key Takeaways

Three facts settle the majority of Advantage+ setup decisions before the first dollar is spent.

• Advantage+ Sales — renamed from Advantage+ Shopping in February 2025 — is a campaign type that removes demographic and interest targeting as hard constraints. Enabling it without intending to cede audience control is the most common Advantage+ mistake.
• Meta's delivery stack (Andromeda, GEM, Lattice) runs under every campaign, manual or automated. "Manual targeting" refers to the configuration layer — the delivery machinery has been automated since late 2024 regardless of campaign type.
• When the algorithm owns targeting from a broad pool, creative becomes the targeting signal. The value of differentiated creative rises as automation expands, not falls.
• Flexible ads were removed from Meta's ad-setup format picker in March 2026. The multi-asset delivery logic was absorbed into Format display options and Advantage+ creative — the capability survived; the label did not.

Advantage+ Feature Map: What the AI Automates

The table below maps each Advantage+ campaign type and Advantage feature enhancement to what the algorithm controls and what the manual equivalent was. The "Type" column is the deciding column — it distinguishes full-automation campaign types from individual feature bolt-ons.

Advantage+ Sales now runs three objectives: sales (e-commerce conversions), app installs, and lead generation. Before the February 2025 rename from Advantage+ Shopping, the format was e-commerce only. The expanded objective scope means Advantage+ Sales is now relevant for DTC verticals including fashion PPC, beauty products PPC, shoes PPC, and supplements and nutrition PPC, and also for lead-gen verticals such as legal PPC and dental PPC that are now within scope. The hard-constraint rule applies across all objectives: location, minimum age, language, and exclusions are enforced as limits; interest and demographic targeting are signals.

MB Adv Agency's audit of new Advantage+ Sales setups consistently surfaces the same misconfiguration: the campaign objective is correctly set, but the advertiser is unaware that interest targeting is functioning as a suggestion. The result is spend reaching audiences outside the intended scope before the algorithm has gathered enough signal to self-correct. Pre-flight review of which constraints are hard versus soft is the single most skipped step in Advantage+ onboarding. See Meta audience targeting for a full treatment of when manual targeting outperforms broad-pool expansion.

Advantage+ Sales: What Changed in the Rename

Advantage+ Shopping Campaigns (ASC) were renamed Advantage+ Sales in February 2025. The rename was structural: the campaign type now runs sales, lead generation, and app install objectives, where legacy ASC ran only e-commerce conversions. The legacy Advantage+ Shopping creation API was deprecated across all Marketing API versions in Q1 2026.

Three structural changes came with the rename. First, Advantage+ Sales removed legacy ASC's single-ad-set constraint — the campaign now supports unlimited ad sets. Second, the existing_customer_budget_percentage API field was deprecated at Marketing API v25.0; the two-ad-set method (one ad set for new customers, one for existing customers) replaced it for separating audience types. Third, audience suggestions and custom exclusions were added as optional advertiser inputs, giving the algorithm directional signals without converting them to hard limits. Sources: PPC.land, October 2025; Bir.ch, Advantage+ Sales guide.

The legacy API deprecation has a direct operational implication for agencies. Meta announced on October 8, 2025 that creation of new Advantage+ Shopping and Advantage App Campaign objects via legacy API versions would be prohibited; the full deprecation took effect with Marketing API v25.0 in Q1 2026. Campaigns created through Meta Ads Manager UI are unaffected. Agencies running volume-scale Advantage+ Sales programs through API integrations or third-party tools built on API versions 23.0 or earlier needed to migrate before Q1 2026; after v25.0, legacy creation endpoints are disabled across all versions.

Advantage+ App campaigns (A+AC) succeed Automated App Ads (AAA) under the same full-automation model: targeting, placement, bid optimization, and audience expansion are machine-controlled; location, minimum age, language, and exclusions are the only hard limits. Meta's product marketing reports "up to 7% lower CPA" for Advantage+ App versus legacy AAA — this is a Meta-reported figure from internal testing, not independently audited. Source: Meta Business Help, Advantage+ App Campaigns.

What to Hand to the Machine vs. What to Keep Manual

The decision is not binary. Meta's automation has three layers: the delivery infrastructure (always automated, no opt-out), the campaign-configuration layer (partially automated through Advantage+ types and feature enhancements), and the creative and signal layer (always the advertiser's responsibility). The table maps the practical recommendation for each lever.

The audience automation question separates DTC from B2B performance. Advantage+ Sales with broad-pool targeting works when ad volume is high enough for the algorithm to find buyers through creative signal — DTC categories such as jewelry PPC and furniture PPC generate that impression volume. Tightly defined B2B audiences, restricted categories, or low-traffic accounts do not give the algorithm enough signal to expand accurately. Running broad audiences in those contexts is not "letting Meta optimize" — it is spending before the data exists to optimize anything.

MB Adv Agency's position on conversion signals: the Andromeda retrieval, GEM prediction, and Lattice ranking pipeline optimizes toward whatever conversion event is defined. A campaign optimized toward add-to-cart sends a different signal than one optimized toward purchase. Meta's automation amplifies misaligned event definitions; it does not correct them. The campaign objectives guide covers event selection in detail. Getting the signal right before enabling full Advantage+ automation is the single highest-leverage pre-launch action. For PPC management in Austin or PPC management in Chicago, where local service businesses run lead-gen objectives, the conversion event is almost always a form submission or phone call — and the default event Meta proposes is rarely either.

The Delivery Stack Underneath Every Meta Campaign

Andromeda, GEM, and Lattice are Meta's three-stage AI delivery infrastructure. They run under every campaign — Advantage+ and manual alike — and no campaign setting disables them. Understanding this stack reframes the "manual vs. automated" question: what advertisers control is the configuration layer on top of a delivery system that was already automated.

Andromeda is the retrieval engine, deployed across Facebook and Instagram and announced via Meta Engineering on December 2, 2024. Its job is the first stage of ad delivery: narrowing the eligible ad pool to a smaller candidate set for the ranking stage. Meta Engineering reports the following on selected segments: +6% recall improvement and +8% ads quality improvement, over 100× improvement in feature-extraction latency and throughput versus prior CPU-based components, and over 3× improvement in end-to-end model inference QPS. Hardware foundation: NVIDIA Grace Hopper Superchips. These are Meta Engineering's figures on selected segments, not independent advertiser-outcome audits.

GEM (Generative Ads Recommendation Model) was deployed across Facebook and Instagram in mid-2025 and described by Meta Engineering in November 2025 as "the largest foundation model for recommendation systems in the industry." Meta Engineering's first-party performance figures: +5% increase in ad conversions on Instagram and +3% on Facebook Feed in Q2 2025, 4× more efficient at driving performance gains per unit of data and compute versus prior models, and a 23× increase in effective training FLOPs from the re-engineered training stack. Attribution: Meta Engineering's own post, not independent audit.

Lattice handles prediction and final auction ranking after Andromeda's retrieval stage. Meta has released fewer technical details on Lattice than on Andromeda or GEM. According to Meta's Q3 2025 earnings materials as reported by Search Engine Land and ALM Corp, rolling Lattice out to app ads drove nearly 3% gain in conversions for that objective — a Meta-reported earnings figure, not independently audited.

The architecture implication for campaign management: the Andromeda/GEM/Lattice pipeline retrieves, predicts, and ranks before campaign configuration touches the outcome. Setting a manual audience defines the pool Andromeda retrieves from; the delivery machinery does the rest. "Manual targeting" is a campaign-configuration choice, not a delivery mechanism. Agencies presenting manual Meta campaigns as non-automated are describing the configuration layer while ignoring the delivery layer.

Dynamic Creative, Flexible Ads, and What Replaced Them

Dynamic Creative was Meta's original multi-asset testing format. Flexible ads replaced it as the labeled option for multi-asset, auto-format delivery in the ad-setup picker. In March 2026, Flexible ads were removed from that picker entirely — not sunset, but absorbed into Format display options and Advantage+ creative.

The removal was quiet. According to Campaign Builder's March 2026 coverage, the automatic format-selection and multi-creative delivery logic was not deleted: it moved into Format display options (where Meta selects format automatically) and Advantage+ creative (which handles per-asset optimization and generative enhancements). The multi-asset delivery capability persists; the standalone format label does not. See Meta ads formats and creative for current spec requirements per asset type.

The durable lesson: Meta keeps the capability and discards the label on a roughly annual cycle. Dynamic Creative → Flexible → absorbed into Advantage+ creative is the third rename of the same underlying job since 2020. Build creative processes around the underlying task — feeding the algorithm multiple distinct, differentiated assets — rather than around whichever branded surface currently exposes that task.

Advantage+ creative in 2026 includes rule-based enhancements (aspect-ratio crops, placement adaptation) and generative AI: background generation (contextual lifestyle backgrounds behind product shots), image expansion (generative fill for additional aspect ratios), text improvements (alternative headline and body variations generated from one description), image-to-video (static product photos converted to multi-scene video), and 3D animation. Sources: AdSUploader 2026; Meta, Advantage+ overview. Text improvements and image expansion are on by default in most placements and require opt-out at the ad level under "Optimize Media." They cannot be disabled globally at the campaign or account level. Source: Jon Loomer Digital.

MB Adv Agency's position: generative Advantage+ creative reduces production time for format variants; it does not produce brand voice, offer framing, or the message-to-audience match that separates a strong ad from a generic one. For verticals with tight visual brand standards — beauty products PPC, fashion PPC — AI background generation in particular requires a human review gate before any generated variant serves at scale.

Two Misconceptions About Meta's AI Automation

The two misconceptions below produce the highest-cost configuration errors. Both trace to the same root: conflating the campaign-configuration layer with the delivery layer.

Misconception 1: "Advantage+ is just a setting I toggle on, and all the automation is the same."
It is not one setting and it is not one product. Advantage+ campaign types — Advantage+ Sales, Advantage+ App — hand the algorithm simultaneous control of audience, placement, budget distribution, and creative selection. Advantage feature enhancements — Advantage campaign budget, Advantage+ audience, Advantage+ creative — are individually toggled layers on a campaign the advertiser still steers. Treating them as interchangeable leads advertisers to enable a full-automation campaign type when they only wanted budget optimization across ad sets, surrendering targeting and placement control they never meant to give up.

Misconception 2: "Meta's AI runs the whole thing now, so creative and structure do not matter."
The opposite is true. As targeting, placement, and bidding move under the algorithm — and as Andromeda, GEM, and Lattice handle retrieval, prediction, and ranking automatically — the advertiser's remaining levers concentrate into two: the creative fed in, and the conversion signal sent back. When Meta selects the audience from a broad pool and serves from a set of loaded creative assets, the creative is the targeting signal — the algorithm uses resonance patterns to find the audience. Thin, undifferentiated creative gives the machine nothing to discriminate against. Automation raises the value of good creative and clean conversion tracking; it does not remove the need for them. For further context on structuring creative inputs, see Meta ads formats and creative.

Frequently Asked Questions

Methodology

Data sources: Ahrefs keyword data (US, June 2026); Meta Engineering blog posts (Andromeda, December 2024; GEM, November 2025); Meta Business Help Center (verified June 2026); Meta for Developers, Marketing API v25.0 changelog; practitioner sources (Foxwell Digital, PPC.land, Campaign Builder, Jon Loomer Digital, AdSUploader, Vizup, Bir.ch) cross-checked against each other. All Meta-reported performance figures are attributed as such and are not presented as independently audited benchmarks. No MB Adv Agency client metrics are cited; agency perspective is qualitative throughout. Last updated: June 2026. Reviewed by MB Adv Agency, June 2026.