ANALYTICS
Advantage+ vs manual: six A/B tests, one clear pattern.
Meta pushed Advantage+ Shopping Campaigns (ASC) hard through 2023 and made the "all Advantage+" defaults in Ads Manager progressively more aggressive through 2024. We ran six controlled A/B tests between January and August 2024 across four DTC brands. Here is the table.
Test setup
Same product, same landing pages, same daily budget, same audience regions. One ad set on ASC with minimal manual overrides (catalog + country targeting only), one on manual interest-based targeting with lookalike seeds. Measured over 14-day windows. Primary metric: CAC on shipped orders, measured server-side via warehouse join.
Results
Brand / category | ASC CAC | Manual CAC | Delta |
|---|---|---|---|
Apparel, $60 AOV | $19.40 | $22.80 | -15% |
Skincare, $110 AOV | $38.20 | $34.50 | +11% |
Home goods, $180 AOV | $61.30 | $72.10 | -15% |
Supplements, $45 AOV | $21.10 | $19.90 | +6% |
Footwear, $130 AOV | $44.80 | $53.20 | -16% |
Accessories, $35 AOV | $14.60 | $15.10 | -3% |
Four of six tests favoured ASC; two favoured manual. The pattern is not random.
The pattern
ASC won decisively on brands with (a) broad addressable audiences (apparel, home, footwear) and (b) strong creative volume (15+ active creatives per week). Manual held or won on brands with narrow audiences (skincare with specific skin concerns, supplements targeting specific health conditions) where the machine could not outperform humans who knew the segments.
Second-order finding
ASC incrementality tested significantly worse than its reported CAC. Geo-lift tests on two of the winning ASC accounts showed actual incremental CAC was 30 to 45% higher than Meta-reported CAC. Manual targeting showed a 15 to 20% gap. ASC cannibalises organic and branded more aggressively than manual.
What we actually recommend
Run ASC for top-of-funnel prospecting where your audience is broad and your creative volume is high.
Run manual for narrow, high-intent segments where you have structural knowledge Meta does not.
Always validate with incrementality, not platform-reported CAC.
Sources: six client A/B tests, Jan to Aug 2024; GeoLift incrementality tests on two accounts.