Facebook advertising where the account structure matters more than the creative

What changed in Meta advertising and why does it matter?

Meta's ad system has consolidated toward Advantage+ campaign types that automate audience selection, placement and budget allocation across your assets. Detailed targeting has been progressively de-emphasised to the point where manual audience layering is largely legacy structure that costs you budget without giving you control.

This inverts the structure most accounts still run — dozens of ad sets built from an audience research meeting, creative as the slot to be filled. That structure now carries management overhead without proportional control, and it splits your budget across segments too small to learn from.

How do Advantage+ campaigns actually work?

Advantage+ campaigns ask the system to decide audience, placement and budget allocation rather than constraining it. In exchange for giving up that control, they need sufficient conversion volume to make good decisions — which is why consolidating campaigns and letting them accumulate events usually outperforms fragmenting them across dozens of ad sets.

The trade-off is real and worth stating plainly. You gain performance from better allocation and lose the ability to see performance by granular audience segment. For most businesses that is a good trade, because the audience segmentation you could report on was rarely actionable anyway. For a business with genuinely differentiated segments and the budget to test separately, some structure is still worth keeping.

1. Fix conversion tracking before consolidating anything. Automated bidding needs reliable events to optimize toward.
2. Consolidate ad sets that target the same objective with overlapping audiences. Budget split too thin across segments produces no learning in any of them.
3. Run one consolidated campaign for a meaningful test period before judging it. The system re-enters a learning phase after significant changes, and frequent edits reset that.
4. Keep creative variety inside the campaign rather than splitting campaigns by creative. That is where the system's optimization has room to work.
5. Use the platform's own A/B test tool when you need a genuine read on creative rather than relying on delivery fluctuations.

Why does event quality decide everything?

Meta's delivery depends on the events it receives. Missing events, duplicated events and events fired with poor data quality all degrade the system, and the failure is usually invisible in the interface — campaigns simply stop performing and the reason is not surfaced.

Check event match quality in Events Manager before changing anything about the campaign. It reports directly on how well your data is matching, and it is the fastest diagnostic for an account that was working and stopped.

What creative does Meta require in 2026?

Vertical short-form video as the primary format, with the first two seconds carrying the entire burden of stopping the scroll. Meta's inventory is predominantly Reels, and creative that ignores that — horizontal video, a logo card first, text-heavy static — is competing for the least valuable placements.

• Hook in the first two seconds. Motion, a spoken opening line, and no logo card.
• 9:16 vertical, designed for a phone held at arm's length rather than cropped from a wide original.
• Captions always. The majority of views begin muted.
• Sound-on creative and sound-off cuts of the same idea, rather than one asset hoping to work both ways.
• One idea per asset. Three messages in fifteen seconds communicates none of them.
• Show the product in use rather than describing it. Demonstration outperforms assertion in every feed.
• Enough variants to test properly — see paid social consulting for the testing discipline.

Are Instant Forms better than a landing page?

Instant Forms — the in-feed lead form — trade landing-page context for lower friction, and the right answer depends entirely on what happens after the submission.

What can you actually measure on Meta?

Platform-reported conversions on Meta are directional. Several effects make them structurally optimistic, and none of them are visible in the interface.

• View-through inflation. Some conversions are credited because an ad was viewable, not because it caused anything.
• Modelled conversions. Where signal is lost to privacy protections, Meta estimates. Modelled and observed conversions behave differently and should be separated.
• Ad blockers and ITP. A meaningful share of conversions never fire a client-side pixel, and they are not randomly distributed — they skew toward privacy-conscious audiences.
• Last-click bias. The final click takes credit for journeys that began elsewhere, including organic search and direct.
• Cross-device gaps. A phone click followed by a laptop purchase is frequently invisible.

Treat platform return on ad spend as a modelled estimate and label it as one wherever it appears in a report. The measurement detail is in the GA4 guide, and the incrementality methods are in paid social consulting.

On creative, note the sequencing: on Facebook the feed is a relevance surface, so a static or native-format post may outperform a video ad for cold audiences. On Instagram the same asset performs differently, because the discovery surface there is Reels. Same brand, same offer, different execution — which is why separate platform pages are worth having rather than treating Meta as one buy.