Advantage+ is the brand Meta uses for its automation layer, and it operates at two levels: Advantage+ campaigns (a campaign with almost no structure, where the algorithm decides audience, placement and budget split) and individual Advantage+ features (audience, placements, budget and creative) that you can switch on inside a normal campaign. It is not a new objective and it is not a different auction: it is the same auction with fewer of your hands inside it.
Telling the two levels apart is what avoids most of the mess. "Let us test Advantage+" can mean building a fully automated sales campaign or ticking one audience box in the campaign you already have. Those are decisions of very different size, and only one of them can be reversed at no cost.
The two levels: Advantage+ campaign and Advantage+ features
Level 1: the Advantage+ campaign. This is the full package, designed mainly for ecommerce sales (with an app equivalent). A single campaign, without the classic structure of several targeted ad sets: you give it budget, a catalogue or a website, a decent stack of creatives, and the system splits spend between new audiences and existing customers, picks placements and decides which ad goes to whom. Learning concentrates in one place instead of being spread across ten ad sets competing with each other.
Level 2: the individual features. These are switched on inside normal campaigns and are independent of one another:
- Advantage+ audience. The biggest conceptual change. What you enter (interests, lookalikes, custom audiences) stops being a limit and becomes a suggestion: Meta is allowed to go outside it if it finds conversions elsewhere. If you were expecting a hard restriction, you are not going to get one.
- Advantage+ placements. The old "automatic placements": the system splits delivery across feed, stories, reels, in-stream and audience network.
- Advantage+ budget. Campaign budget optimisation as it has always been (CBO): it distributes across ad sets wherever it sees opportunity.
- Advantage+ creative. Automatic enhancements on top of your asset: visual tweaks, image expansion to fit each format, text variations, music. This is where control is lost most quietly.
Meta has spent a while merging the two levels in the interface, so the automated setup now shows up as the default when you create sales or lead campaigns and manual is the alternative you have to switch off. It is worth checking what is ticked on every new campaign rather than assuming last year's configuration.
What control you keep (and which control actually matters)
The list of real levers in an Advantage+ sales campaign is short, and that is precisely why it either works or fails:
- Budget and cost or return goal. The main lever.
- Existing customer budget cap. The share of spend allowed to go to people who have already bought from you. It is the most important control in the set and the one most people never touch.
- Creatives. You can load a lot and the system decides. Creative volume and variety become the main optimisation lever, ahead of targeting.
- Exclusions and customer lists, which define what counts as an existing customer.
- Placements and creative enhancements, which can be turned off one by one.
What you do not keep is fine targeting, separate ad sets to read results by segment, or control over which creative is shown to whom. If your operation depends on reading "which audience works", that reading disappears: what you will be able to read is which creative works.
The measurement problem: why panel ROAS looks so good
This is the pattern we see most often when auditing accounts that have just switched it on. The Advantage+ campaign starts, panel ROAS climbs spectacularly and total company revenue stays exactly where it was. There is no magic and no fraud: the system finds the cheapest route to an attributed conversion, and that route is always demand that already existed. People who already knew you, who already had a cart open, who were going to buy anyway.
That is why the existing customer cap is the critical lever: leave it high and you are paying to claim credit for sales you already had. With a low cap you force the campaign to hunt for new demand, and there panel ROAS drops while the actual business improves. The number that settles this is not campaign ROAS but MER: total business revenue against total spend. If campaign ROAS rises 60% and MER does not move, you have not gained efficiency, you have moved attribution around.
Add to that the fact that what you see depends on the attribution window, and that the Advantage+ audience tends to pick up brand traffic and disguised retargeting. Before declaring a win, the check takes two minutes: look at total revenue and new customer orders in your backend for the window before and the window after. If neither moves, the win belongs to the report.
When it fits and when it does not
It fits well when:
- You are ecommerce with a catalogue and enough event volume for the system to learn quickly; with a handful of weekly sales, learning never closes and performance is erratic.
- You have creative production to feed it. It is a machine that eats assets: if all you can hand it is three ads, you are removing its only real lever.
- The account is fragmented into too many small ad sets that cannibalise each other and never exit learning. Consolidating usually beats slicing further.
- The pixel and the conversions API are properly implemented. All automation rests on event quality: with broken measurement, automating only speeds up the error.
Tread carefully when:
- B2B lead generation. The system optimises towards the event you give it, and if that event is "form submitted" it will bring you cheap, bad forms. Without sending qualification back (valid lead, opportunity) as an event, the automation works against you.
- Strict brand control. Creative enhancements crop, expand and rewrite. On assets with closed brand rules you have to switch the visual ones off one at a time and review the preview for every placement.
- You need to read by segment (geography, customer type, product line) to make decisions outside Meta. That reading is not there.
- Very small budgets split across several campaigns on top of that: consolidation is exactly the advantage of the format, and slicing it up cancels it out.
How to test it without fooling yourself
A protocol that works and that avoids the false conclusions of the first few weeks:
- One campaign, not five. Duplicating Advantage+ campaigns to "test variants" splits the learning and puts them against each other in the auction.
- Low existing customer cap from the start if what you want is growth. That is the difference between buying new demand and repurchasing your own.
- Let it run untouched. Every budget or goal edit restarts learning. Minimum reading window: two weeks, plus your purchase cycle on top if your order value is high.
- Measure outside the panel. Before you start, write down MER, total orders and the new customer share. Those three numbers decide, not campaign ROAS.
- Switch the individual features on one at a time. If you turn on audience, placements and creative together and the result changes, you will not know which one did it.
- Watch the placement split in the first weeks. A CPC that collapses alongside conversion rate is usually cheap inventory absorbing the budget, not an efficiency gain.
Frequently asked questions
What is Meta Advantage+?
It is the name Meta gives its automation layer, and it covers two different things. On one side, Advantage+ campaigns: a campaign with almost no structure where the algorithm decides audience, placements and the split between new audiences and existing customers. On the other, the individual Advantage+ features (audience, placements, budget and creative), which are switched on independently inside normal campaigns.
What is the difference between an Advantage+ campaign and a normal campaign?
In a normal campaign you define ad sets with their targeting, their placements and their budget, and you can read results by segment. In an Advantage+ campaign there is no such structure: you provide budget, creatives and a cost or return goal, and the system decides the rest. In exchange for losing targeting and per-audience reading, you concentrate all the learning in one place instead of splitting it across ad sets that compete with each other.
Does Advantage+ work better than manual campaigns?
It depends on the account, and panel ROAS cannot answer the question. It works well in ecommerce with a catalogue, enough event volume and constant creative production, especially if the account was sliced into many small ad sets. It works badly when measurement is broken, when the optimised event is not the one that matters (typical in B2B lead gen), or when you need brand control or segment-level reading.
What is the existing customer budget cap?
It is the maximum share of campaign spend allowed to reach people who have already bought from you, defined by the customer lists you upload. It is the most important control in an Advantage+ sales campaign: leaving it high means the campaign takes credit for sales you already had and panel ROAS looks excellent without the business growing. Lowering it forces the system to hunt for new demand.
How much budget does an Advantage+ campaign need?
Rather than a fixed figure, what it needs is enough conversion event volume to exit learning within a reasonable timeframe, and enough creatives to have something to optimise with. The practical rule: if your budget does not produce a steady flow of weekly conversions, the system never closes learning and performance turns erratic. In that scenario, consolidating into a single campaign helps far more than splitting the budget across several.
Does the Advantage+ audience respect the targeting I set?
Not as a limit. The interests, lookalikes and custom audiences you enter work as a starting suggestion: the system is allowed to go outside them if it finds conversions elsewhere. The hard restrictions that are respected are exclusions and the legal location and age requirements in special categories.
Can Advantage+ be used for B2B lead generation?
It can, but only with measurement fixed first. Automation optimises towards the event you give it: if that event is the form submission, it will bring cheap, low quality forms. To make it work you have to send the real qualification event back to Meta (valid lead or opportunity) through the conversions API and optimise against that, not against the form.
Advantage+ showing an excellent ROAS while revenue stays flat?
The viability audit separates new demand from repurchased demand and tells you how much of that return is incremental. Free, no strings attached.