The promise of Meta's Andromeda AI brain sorting through a billion users to find your customers on auto-pilot sounds enticing. But is it actually good for brands? I have spent 10 years tracking Meta campaign performance against actual orders, and four things have stayed true the entire time:
Meta overstates its ability to drive immediate bottom-of-funnel sales.
Meta understates its ability to drive top-of-funnel cold traffic that leads to new customers.
Meta shows a lot more ROAS when you widen the attribution window.
Meta's Conversion API is heavily optimized for immediate sales on the default setup.
With Meta now pushing advertisers hard toward Advantage+, and switching it on behind the scenes if you do not read the fine print, it is time to settle this with actual data instead of opinion. So I had the Wicked DB team pull 55,661 Meta campaigns and compare January through May 2024 against the same window in 2025. Here is what the data shows.
There is a 4% swing in budget moving toward Advantage+ over manual campaigns year over year. We are watching this closely through the summer, because there are rumors advertisers are being forced into Advantage+ with no ability to configure manually. If that holds, expect this number to jump sharply, and it means advertisers will need solid Meta CAPI integration to cope with the trends below.
Here is the headline. On Advantage+ campaigns, the cost to acquire a new customer has skyrocketed over the past three months. Comparing May 2024 to May 2025, average nCAC on Advantage+ went from $257 to $528 per customer. That is not a rounding error, that is nearly double.
Meanwhile, Meta manual campaigns are actually acquiring customers more cheaply in 2025 than they were in 2024. At the start of the year, Advantage+ was destroying manual campaign performance. Then it flipped hard, and by May, manual campaigns were crushing Advantage+ on nCAC.
This could be the start of model degradation, or it could be a temporary blip. Either way the danger is structural: when new customer costs spike, budget gets cut on those campaigns, which means less cold-traffic spend, which means less top-of-funnel volume, which means weaker bottom-of-funnel performance down the line. The whole funnel contracts.
Next we benchmarked first-click and last-click ROAS across all Meta campaigns, regardless of type. On average in 2025, first-click and last-click ROAS came out surprisingly close to each other.
Exclude Advantage+ and look at manual campaigns alone, and they track closely to that overall picture, first and last-click ROAS again fairly similar.
But isolate Advantage+ campaigns and the story changes. Over the last three months, last-click ROAS is significantly better than first-click, and first-click ROAS plummeted in May.
That gap is the tell. Advantage+ appears to be optimizing toward last-click ROAS, chasing the easy final-touch conversions while its ability to drive genuine new-customer, top-of-funnel results erodes. That is exactly what the nCAC trend above would predict.
May was a dramatically worse month for Advantage+ performance. If the nCAC and first-click ROAS trends hold for another month, the recommendation is straightforward: use Advantage+ for bottom-of-funnel, where its last-click optimization is a strength, and use manual campaigns for top-of-funnel cold traffic, where they are currently acquiring new customers far more cheaply.
The broader point is the one that has held for a decade. Meta's own reporting will tell you Advantage+ is working, because it grades itself on the last-click, immediate-sale terms it optimizes for. Only when you measure against actual orders, separating new customers from repeat buyers, does the real picture show up. That is the difference between running your budget on Meta's story and running it on the truth. See how it works on the platform overview, or book a demo to see this on your own campaigns.
Data from 55,661 Meta campaigns shows a sharp increase in new customer acquisition cost for Advantage+, with average nCAC roughly doubling from $257 to $528 per customer between May 2024 and May 2025. Over the same period, Meta manual campaigns acquired new customers more cheaply in 2025 than in 2024. The trend suggests Advantage+ is degrading at driving cost-effective new-customer sales, while manual campaigns are currently stronger for that job.
Based on the trends, rising nCAC and falling first-click ROAS on Advantage+, the current recommendation is to use Advantage+ for bottom-of-funnel retargeting, where its last-click optimization helps, and manual campaigns for top-of-funnel cold traffic. Manual campaigns are currently acquiring new customers at a lower nCAC, which makes them the better choice for genuine acquisition.
The data indicates Advantage+ is increasingly optimizing toward last-click ROAS, where the final ad clicked gets nearly all the credit. Last-click ROAS running well ahead of first-click for Advantage+ inflates immediate-sale metrics while top-of-funnel first-click ROAS drops, which is a direct driver of the higher new customer acquisition costs the study found. Measuring against real orders rather than Meta's own reporting is the only way to see it.