Advantage+ Shopping Won 42% of 640 Tests. Manual Won the Other 58%.

Haus measured 640 Meta incrementality experiments across advertisers spending an average of $14M a year and found that Advantage+ Shopping beat manual campaigns in only 42% of accounts. On average, Advantage+ delivered 12% lower incremental ROAS than manual, even though it spent 18% less per day. Meta's marketing materials promise a 17% lower CPA. The incrementality math does not promise the same thing.

That gap is the entire reason this decision still matters. In our pillar on Meta ads strategy in 2026, we argued that Advantage+ has eaten most of the manual workflow but not the strategy work. This piece is the zoom-in. When does Advantage+ Shopping actually win, when does manual win, and how do you tell the difference before you spend three months learning it the expensive way?

The 17% number you keep seeing comes from Meta's own report

Meta's frequently quoted internal data shows Advantage+ Shopping campaigns delivered roughly 17% lower cost per acquisition and 32% higher ROAS than manual campaigns. That number is real for the accounts in Meta's sample. It is also a platform-attributed number, which means it includes everyone the algorithm thinks it converted, including buyers who would have purchased without an ad at all.

The Haus study, which measures incremental conversions through inverse holdout tests rather than platform attribution, found that Advantage+ over-reports relative to manual by about 12 percentage points on average. That is not Meta lying. It is the algorithm doing exactly what it was trained to do, which is finding the people most likely to convert and serving them ads. The trouble is that those people were already going to convert. The platform takes credit. The brand pays for the impression. The incremental gain is much smaller than the dashboard suggests.

If you have never run a holdout against your Advantage+ campaign, your reported ROAS is doing some of this same lying to you. From what I have seen in agency reporting, the gap between platform ROAS and incremental ROAS is usually somewhere between 30% and 60%, and it is wider on remarketing-heavy accounts.

The threshold where Advantage+ probably wins

The Haus data suggests Advantage+ outperformed manual on about 70% of new-customer impact, while manual was at 65%. So the algorithm is genuinely better at finding incremental new buyers, just not by as much as Meta's marketing implies. The deciding factors look like this in most accounts I have seen referenced in practitioner write-ups:

• Monthly Meta spend above $10K. Multiple practitioner guides converge on this as the threshold where the algorithm has enough conversion data to escape the learning phase quickly. Below that, manual ad sets with tight targeting are usually more efficient because the AI is just guessing.
• 20+ active SKUs in your catalog. Advantage+ is built to dynamically pull from a product feed. If you sell three things, the algorithm has nothing to optimize across.
• Broad enough creative library that the system can rotate. Foxwell Digital and other practitioners now recommend at least 10 to 15 creative variants in rotation per Advantage+ campaign, because the system burns through fatigued creative faster than manual.
• 50+ purchases a week as a standing baseline. Below that, the conversion signal is too noisy and the algorithm thrashes.

If you clear all four of those, Advantage+ is probably your better default for prospecting. That said, defaulting is not the same as setting and forgetting. The 12% iROAS gap shows up most often in accounts that never tested.

Where manual still beats Advantage+, and why that gap widened in 2025

In February 2025, Meta removed the Existing Customer Budget Cap from Advantage+ Shopping. That cap was the lever that let advertisers force Advantage+ to spend mostly on prospecting and limit retargeting to a defined slice of budget. Without it, Advantage+ tends to drift toward warmer audiences, which is the easiest way for the algorithm to hit its conversion goal and the easiest way to inflate platform ROAS.

Meta's official workaround is to build a manual sales campaign that excludes a custom audience of existing customers. That is functionally the old prospecting-only setup, just rebuilt as a manual campaign. So one of the cleanest reasons to still run manual in 2026 is that Meta itself recommends it for prospecting control.

Manual also beats Advantage+ in a few other recurring scenarios. New product launches do not have enough conversion history for the algorithm to optimize toward, so a manual ad set with interest-based or lookalike targeting usually outperforms in the first 14 to 30 days. Niche audiences (high-AOV B2B, regulated categories, very small geographies) tend to under-deliver on Advantage+ because the algorithm prioritizes scale. And creative testing is structurally harder on Advantage+ because you cannot isolate a variable cleanly. The system mixes placements, audiences, and creatives at the same time.

One more wrinkle worth flagging. Advantage+ Shopping CPMs have climbed roughly 94% year-over-year since mid-2024 across many DTC accounts, according to aggregated agency reports. Some of that is auction inflation everyone is paying. Some of it is Advantage+ specifically biasing toward higher-value placements. If your category is CPM-sensitive, manual gives you the option to cap inventory choices in a way Advantage+ does not.

The 60-30-10 split that keeps showing up in practitioner write-ups

Among the agency guides published in late 2025 and 2026, a recurring budget structure is roughly:

• 60-70% of budget on Advantage+ Shopping for broad prospecting
• 20-30% on a manual prospecting campaign with custom audiences excluded for existing customers (replicating the old budget cap)
• 10-20% on manual retargeting with dynamic product ads against your warm pool

I think this is roughly right as a default for most DTC accounts above $10K monthly. The 20-30% manual prospecting slice is the part most teams skip, and it is the slice that most often catches incremental customers Advantage+ misses. The retargeting carve-out matters because Advantage+ does not give you per-audience reporting, so if you want to know what your warm pool is actually contributing, you need a separate manual campaign to measure it.

One important nuance. The 60-30-10 only works if you actually run the test. Meridian GeoX going open-source on Google Cloud earlier this year made geo-holdout incrementality testing a lot more accessible, and brands spending over $50K a month on Meta should be running one at least quarterly. If you cannot tell whether your Advantage+ campaign is buying incremental sales or just being credited for them, the budget split decision is just guesswork.

The 30-day plan if you are running Advantage+ blind today

If your account is currently 100% Advantage+ with no holdout, no manual control, and platform-reported ROAS as your only number, here is what I would actually do in the next four weeks:

Week 1. Pull a 90-day report on Advantage+ Shopping by audience type if it is available, or at minimum compare your reported ROAS to your blended business ROAS (revenue divided by total ad spend). The bigger the gap, the more retargeting your Advantage+ is doing.

Week 2. Build the manual prospecting campaign Meta itself recommends. One ad set, broad targeting, exclude a custom audience of past purchasers from the last 180 days. Allocate 20% of total Meta budget. Use the same creative variants you are running on Advantage+.

Week 3. Run for 14 days. Compare Advantage+ prospecting CPA to manual prospecting CPA against the new-customer column in your shop, not the platform attribution. If you sell on Shopify, the customer report is fine for this. If you sell on a custom stack, you may need to tag email subscribers as known.

Week 4. Decide. If manual prospecting CPA is within 15% of Advantage+, keep it running and start a real geo-holdout for incrementality. If it is more than 30% worse, kill it and move the budget back to Advantage+ but add a creative testing manual campaign instead. Either way, you have learned something the dashboard would never tell you.

The practitioner question that decides the whole thing

Most teams I see still treat the Advantage+ versus manual choice as a setting in the campaign builder. It is closer to a measurement question. Advantage+ optimizes against platform-attributed conversions. Manual lets you slice budget cleanly enough to run an incrementality test. If you do not measure incrementally, both approaches will tell you they are winning, and one of them will probably be lying by 30 to 60 cents on the dollar.

If you take one thing from the Haus data, take this. Advantage+ is the right default for accounts that clear the four thresholds above. It is not the right answer for accounts that have never tested. The 58% of brands where manual won were not running weird setups. They were running normal e-commerce accounts that took the time to look past the platform dashboard. That is most of the actual edge left in Meta in 2026.

A few practitioner questions worth answering directly

Should small DTC brands ($2K-$5K monthly Meta spend) use Advantage+ at all? Probably not as a primary campaign. Below the $10K monthly threshold, the algorithm does not have enough conversion volume to escape the learning phase quickly, and Advantage+ tends to spend most of its budget guessing. A manual lookalike or interest campaign with three to five tightly-targeted ad sets is usually more efficient at that scale, even if it requires more day-to-day attention.

Can you run Advantage+ Shopping and a manual sales campaign at the same time without auction overlap penalties? Yes, and Meta's own help center now recommends it for prospecting control. The auction does have an overlap deduplication mechanism, so you will not bid against yourself, but you will see some delivery shift between the two as Meta picks whichever ad it thinks will win the impression. Use distinct creative or distinct audiences if you want clean attribution.

Does the 60-30-10 split still apply if you sell only one or two products? No. Advantage+ Shopping is built around catalog rotation. With a single SKU you should be running a manual sales campaign with multiple creative angles, not Advantage+. The exception is if you are testing a high-LTV subscription product where the algorithm can optimize on subscription start events and you have hundreds of those a week.

Notice Me Senpai Editorial