The Meta Ads account structure that scales for ecommerce is small: one main Advantage+ sales campaign that gets most of the budget, one testing campaign that feeds it new creative, and a clean Conversions API setup underneath. Every ad set needs enough purchases to finish learning, so fewer ad sets with more budget each beats a sprawl of tiny ones.
Most ecommerce ad accounts we audit don't have a targeting problem. They have a structure problem. Fourteen campaigns, forty ad sets, each one starved for budget, half of them stuck in learning, and nobody can say which one is actually making money. When we take over a Meta account for a Shopify brand, the first big win is almost never a new audience. It's deleting things.
This guide covers how we structure Meta Ads for ecommerce in 2026, why Meta's own rules push you toward a simpler build, and what changed when Advantage+ shopping campaigns became Advantage+ sales campaigns.
What is the best Meta Ads account structure for ecommerce?
Two to three campaigns, not ten. Here is the build we start almost every DTC account on:
- A scaling campaign. An Advantage+ sales campaign optimized for purchases, with Advantage+ campaign budget on, broad Advantage+ audience, and automatic placements. This one gets 70 to 80 percent of spend once it's proven.
- A creative testing campaign. A smaller budget, its own campaign, where new concepts fight it out. Winners get moved into the scaling campaign. Losers get turned off without touching the main engine.
- An existing customer lane, only if the business needs one. For brands with real repeat purchase behavior (supplements, coffee, apparel drops), we split new and returning buyers so we know what we're paying to acquire a new customer. More on how that works now below.
That's it. No separate campaign per audience, per country, per product, per age band. The reason is not taste. It comes straight out of how Meta's delivery system learns.
Why does Meta reward fewer ad sets?
Because learning happens at the ad set level, and every ad set needs volume to learn. Meta's own documentation says learning stabilizes with a minimum of 50 events over a 7 day period, and that changes to an ad set can trigger a new learning period. Meta's Marketing API reference also notes that delivery performance may be unstable while an ad set is still in learning.
Do the math on your own account. Here is an illustrative example, not a client number: if your cost per purchase runs about $40, one ad set needs roughly $2,000 a week, or about $285 a day, to hit 50 purchases in 7 days. A brand spending $10,000 a month can fund maybe one ad set at that level. Split that same $10,000 across eight ad sets and none of them ever gets out of learning. You're paying full price for an algorithm that never finishes its homework.
This is the mistake we fix most often. Owners hear "test more audiences" and build a new ad set for every interest. Each one looks cheap on its own. Together they guarantee unstable delivery.
Should ecommerce brands use Advantage+ sales campaigns?
For most stores, yes, as the main scaling campaign. In Meta's current setup, a sales campaign counts as Advantage+ when three levers are on: Advantage+ campaign budget at the campaign level, Advantage+ audience in at least one ad set, and Advantage+ placements with no placement targeting or exclusions. Turn any one of them off and the campaign loses its Advantage+ status.
Meta describes Advantage+ sales campaigns as the updated, streamlined version of Advantage+ shopping campaigns (the old ASC). The practical differences matter:
- More than one ad set is now allowed. Old ASC campaigns were limited to a single ad set. Advantage+ sales campaigns can have more than one.
- The existing customer budget cap is gone for new campaigns. Old ASC let you set a maximum percentage of budget for existing customers. Meta's developer docs say that setting isn't available for new Advantage+ campaigns, and old ASC campaigns using it get paused at Marketing API v26.0.
- Moving an old campaign resets learning. Meta's FAQ says migrating will force the campaign into the learning stage. Plan the switch for a slow week, not the week before Black Friday.
Our take: Advantage+ is not magic, and it will happily spend money retargeting people who were going to buy anyway. But with good creative and clean tracking, the broad build beats hand built interest stacks on almost every DTC account we touch. The work moved from targeting to creative and data. Accept that and the account gets easier to run.
How do you separate new and existing customers now?
With two ad sets instead of one setting. Meta's own example for replacing the old existing customer cap uses one ad set that includes a custom audience of existing customers and one that excludes it, with a daily spend cap on the existing customer ad set to keep it in check.
Build the existing customer audience from real data: your Shopify customer list, past purchasers from the pixel, and email subscribers if they're mostly buyers. Then decide what you want returning customers to cost you. For a lot of brands the right answer is "very little," because email and SMS flows bring those people back for close to free. Paying Meta full price to reach someone who opens your emails every week is one of the quieter ways ROAS gets inflated while profit shrinks.
One warning. Splitting by customer status adds an ad set, and each ad set needs its own learning volume. If your budget can't feed both, skip the split until it can.
How should you test creative without breaking the scaling campaign?
Test in a separate campaign and only move proven winners into the main one. Every time you edit the scaling ad set, you risk kicking it back into learning. A testing campaign gives new concepts a place to fail without costing you the stable results you already have.
Some limits worth knowing. Meta's Marketing API reference caps an ad set at 50 non archived ads. The old ASC comparison table listed up to 150 creative combinations versus 50 for manual campaigns. You will not need anywhere near that many, but it tells you what Meta's system is built for: lots of creative variety, very little audience slicing.
Here's how we run the testing lane:
- Test concepts, not button colors. A new angle (price, proof, problem, founder story, UGC demo) is a test. A new font on the same image is not.
- Give each round a fixed budget and a fixed window, usually 7 days, and judge on cost per purchase, not click through rate.
- Move winners into the scaling campaign in batches, once a week, instead of one at a time every day. Fewer edits, fewer learning resets.
- Retire ads in the scaling campaign only when they fade. Don't kill a winner because you're bored of looking at it. Your customers have seen it far less than you have.
What tracking does this structure need?
The Conversions API, set up properly, on top of the pixel. A broad, automated structure lives or dies on the purchase signal Meta gets back. Meta's developer docs describe the Conversions API as a connection between your server, website platform, or CRM and Meta's systems that optimize targeting, lower cost per result, and measure outcomes.
Three checks we run on every Shopify account:
- Both pixel and server events are firing, and they're deduplicated. Meta's setup steps call out confirming that events are deduplicated and matched correctly. If purchases are double counted, the algorithm optimizes toward a fantasy.
- Customer details are passed the same way on both sides. Meta's customer information docs say to apply the same parameters on the browser and server sides, and that sending client IP address and user agent may help improve event matching.
- The purchase value is real. Send revenue after discounts, in the right currency. A value bug quietly wrecks value based bidding and every ROAS number you report.
If your ROAS has been sliding and nobody has looked at tracking in a year, start there before touching structure. We wrote up the other usual causes in why your ROAS is dropping.
What would we actually do with a $10k to $50k a month account?
Here is the plan we would run on a typical Shopify brand coming to us with a messy account.
Week 1, audit and clean. Check Conversions API, deduplication, and purchase values first. Then list every active campaign and ad set with its spend, purchases, and learning status. Anything spending money without enough volume to learn gets paused.
Week 2, rebuild. Launch one Advantage+ sales campaign with the account's best 8 to 15 proven ads. Launch a testing campaign at roughly 15 to 25 percent of total spend. Add the existing customer split only if the budget clears the learning math for both ad sets.
Weeks 3 to 6, leave it alone. This is the hard part. No daily budget swings, no daily edits. Feed the testing campaign 3 to 5 new concepts a week and graduate winners in one weekly batch.
After that, scale. Raise budget on the scaling campaign in steady steps while cost per purchase holds, and judge results on blended numbers (total revenue against total ad spend, plus new customer counts from Shopify) instead of trusting in platform ROAS alone. Our post on what a good ROAS really is explains why.
None of this is flashy. It works because it gives Meta's system what it was built to use: enough budget per ad set, enough creative to choose from, and an honest purchase signal.
Want us to look at your Meta account structure?
Send us access and we will tell you which campaigns are starving, which ad sets are stuck in learning, and what we would consolidate first. Free, no pitch deck. See how we run Meta Ads for DTC brands, or read our NUDE Envie case study.
Get the free audit →FAQ
How many campaigns should an ecommerce brand run on Meta?
Most stores spending 5k to 50k a month do best with two or three: one main Advantage+ sales campaign for scaling, one creative testing campaign, and an optional split for existing customers. Every ad set needs enough purchases to finish learning, which Meta puts at a minimum of 50 events over 7 days, so more campaigns usually means worse results.
What replaced Advantage+ shopping campaigns?
Advantage+ sales campaigns. Meta calls them the updated, streamlined version of Advantage+ shopping campaigns. A sales campaign is Advantage+ when Advantage+ campaign budget, Advantage+ audience, and Advantage+ placements are all on, and unlike the old version it can have more than one ad set.
Can I still cap spend on existing customers?
Not with the old percentage setting on new campaigns. Meta's suggested replacement is two ad sets: one that includes a custom audience of existing customers and one that excludes it, with a daily spend cap on the existing customer ad set.
Does editing a Meta ad set reset learning?
It can. Meta says changes can trigger a new learning period, and migrating an old Advantage+ shopping campaign forces learning. That is why we test creative in a separate campaign and move winners into the main campaign in weekly batches.
Sources
- Meta for Developers, Advantage+ campaign experience for sales, app, and leads (three levers that make a campaign Advantage+; described as the streamlined version of Advantage+ shopping campaigns; existing customer budget percentage not available for new campaigns and paused at v26.0; two ad set include and exclude approach with a daily spend cap; migration forces learning; more than one ad set allowed)
- Meta for Developers blog, Advantage+ campaign experience for sales and app (June 3, 2025) (Advantage+ campaign budget manages budget across ad sets; Advantage+ shopping API deprecation timeline)
- Meta for Developers, Advantage+ shopping campaigns (one ad set per ASC campaign; up to 150 creative combinations versus 50 for manual; existing customer budget percentage definition)
- Meta for Developers, Understanding and optimizing your ad campaign (learning happens at the ad set level and stabilizes with a minimum of 50 events over a 7 day period; changes can trigger a new learning period)
- Meta Marketing API reference, Ad Set (learning stage info; performance may be unstable during learning; 50 non archived ads per ad set)
- Meta for Developers, Conversions API and customer information parameters (what the Conversions API connects; confirm events are deduplicated and matched; same parameters browser and server side; client IP and user agent may help matching)
- The $40 cost per purchase math is an illustrative example, not a client result. The structure, budget splits, and opinions are ours. Search volume data was not available for this post, so the target keyword comes from the calendar plus live search results.