Meta's advertising ecosystem never stays still. Every quarter brings algorithm shifts, new ad formats, updated bidding logic, or creative requirements that quietly reshape what works. Most brands don't notice until their ROAS quietly deteriorates and they're left wondering why the same campaigns that worked six months ago are now bleeding budget with diminishing returns.

We manage paid media across dozens of active Meta accounts spanning e-commerce, fashion, SaaS, and professional services. What follows is an honest account of what our team is seeing right now — not theory, not recycled best-practice guides, but live observations from campaigns running today.

Broad Targeting Has Become the Default — and That's Not a Bad Thing

For years, the paid media playbook was built on tight audience segmentation. Stack your interest layers, exclude your customer lists, build lookalikes off your best buyers. That approach worked brilliantly — until Meta's machine learning became powerful enough to out-target any human-built audience.

In 2025, broad targeting (minimal restrictions, Advantage+ audiences, or simply targeting by country with no interest layers) is consistently outperforming heavily segmented approaches on accounts with sufficient conversion data. The algorithm has enough signal — from pixel history, purchase events, and behavioural patterns — to find the right buyer without us telling it where to look.

This doesn't mean structure doesn't matter. Campaign architecture still drives efficiency. But the levers have shifted from audience building to creative quality and bid strategy.

Creative Is Now the Primary Targeting Mechanism

This is the most important shift in Meta advertising over the past 18 months. Creative doesn't just determine whether someone clicks — it determines who the algorithm shows your ad to in the first place. Meta reads your creative and uses it to identify who is most likely to respond. A video that opens with "struggling with X?" will be served to people who struggle with X, even without a single interest layer specifying that audience.

"If your creative isn't doing the targeting, your budget is funding Meta's guesswork. In 2025, the ad itself is the audience brief."

The practical implication: creative testing is now your highest-leverage activity in paid media. Not bidding. Not budget allocation. Creative. Brands that run systematic creative testing — iterating on hooks, formats, and messaging angles across a structured testing framework — will see compounding improvements in cost per acquisition that no bidding tweak can replicate.

Short-Form Video Dominates, But Static Isn't Dead

Reels placements continue to drive strong performance for awareness and prospecting, particularly for visually-led brands. The first three seconds are everything — if the hook doesn't create immediate curiosity, pattern interruption, or emotional resonance, the algorithm reads the skip rate and restricts distribution fast.

What's interesting, though, is that static image ads and carousels are outperforming video for certain objectives — particularly retargeting and direct-response conversion. A clean product image with sharp copy and a strong offer often converts better in the lower funnel than a polished video, because the cognitive load is lower and the decision pathway is shorter.

The accounts performing best in our portfolio are running mixed creative libraries: short-form video for prospecting, UGC-style content for mid-funnel trust-building, and static or carousel ads for retargeting and closing.

Advantage+ Shopping Campaigns Are Worth Testing — With Caveats

Meta's Advantage+ Shopping Campaigns (ASC) have matured significantly and are delivering strong ROAS for e-commerce accounts with clean product catalogues and reliable pixel data. They remove most of the manual targeting decisions and let the algorithm manage placement, audience, and budget allocation autonomously.

The caveat: ASC tends to heavily favour retargeting audiences, which can inflate your reported ROAS while your new customer acquisition cost quietly climbs. We run ASC alongside manually structured prospecting campaigns and monitor the customer acquisition breakdown carefully. If new customer ROAS degrades without new customer volume increasing, ASC is doing the easy work and leaving the hard acquisition to bleed.

What We're Doing Differently Right Now

Across our client accounts, here are the moves we're actively making in Q2 2025:

  • Consolidating campaigns. Fewer campaigns, higher budgets per campaign — this gives the algorithm more signal per campaign and exits the learning phase faster. We've reduced campaign counts by 30–40% on most accounts without reducing results.
  • Running a dedicated hook-testing framework. We test 4–6 different video hooks monthly, isolating the first 3 seconds as the only variable. The winning hooks inform our organic content strategy too.
  • Separating new customer and retargeting reporting. Blended ROAS is a vanity metric. We track new customer CPA separately and optimise toward that, even when it means accepting a lower blended ROAS.
  • Using cost cap bidding selectively. Highest volume bidding remains our default, but for mature accounts with large conversion histories, cost cap bidding at 1.2–1.5× target CPA is delivering more efficient scale.

Meta's algorithm rewards accounts that feed it quality data and quality creative. Every other variable — bidding, structure, placements — operates in service of those two inputs. If your Meta performance has plateaued, the answer is almost certainly in your creative pipeline, not your campaign settings.