Platform shifts, 2026

#Meta: Advantage+ absorbed execution, and the evidence on whether that's actually better is mixed

Meta has spent 2025–2026 consolidating legacy manual campaign controls into its Advantage+ automated system, moving toward a state where an advertiser supplies a product URL and a budget and Meta's AI generates the creative, selects audiences without demographic input, and optimises placement and creative in real time. [Established] as Meta's stated direction and shipped product behaviour through 2026; the specific end-state ("full automation") is closer to [Directional] since it describes where Meta says it's heading, not a completed migration.

Meta reports Advantage+ campaigns running at roughly $60 billion in annualised revenue and a 22% ROAS improvement over manual setups, by Meta's own benchmarks. [Directional] — both figures are Meta's own self-reported numbers relayed through secondary coverage rather than checked against Meta's investor filings directly, and a platform reporting how well its own product performs carries an obvious structural incentive to report favourably.

The more important finding is independent of Meta's own claim. Haus, a marketing-measurement firm with no obvious incentive to make Meta look bad, ran 640 incrementality tests and found Advantage+ campaigns underperform manual campaigns over time despite strong platform-reported ROAS. [Directional] — Haus does sell incrementality-measurement services, so it has its own commercial interest in a finding that makes platform-reported metrics look untrustworthy, but a study across 640 tests is a materially stronger evidence base than either a blog assertion or a platform's own self-reported benchmark, and the finding is directionally consistent with a well-understood mechanism: a platform's own attribution naturally tends to credit itself for conversions that would have happened anyway. This is the single most important fact in this lesson for the root-mechanism argument in the previous module — it means the "agency value shifts from execution to judgment" claim isn't just theoretical, it's the exact gap Advantage+'s self-reported numbers are missing and a measurement-literate agency can catch.

What this means for the actual work. Agencies selling pure campaign-setup execution are competing against a feature Meta ships for free inside the same platform. The agencies holding value are the ones doing the things Advantage+ doesn't do on its own: independent measurement that catches the gap the Haus study identifies, creative production (the system optimises what you feed it, but still needs real creative assets, not just a URL, to perform well), and cross-platform strategy Meta has no visibility into or incentive to optimise for. [Directional] consensus framing across current trade commentary, following directly from the mechanism argued in the previous module rather than an independently-measured claim.

Meta's average CPM in 2026 runs around $13.48, roughly 20% above 2025 levels by one ad-tech vendor's benchmark tracking. [Directional] — a single vendor's benchmark with undisclosed methodology; treat the direction (CPMs rising) as more reliable than the precise figure. Reels inventory reportedly runs a 30–40% CPM discount versus feed placements, from the same source tier. Rising CPMs matter directly to the unit-economics math in the previous module: a client's ad-spend efficiency is compressing even where an agency's fee stays flat, which is downward pressure on the value story an agency has to keep making to justify the retainer.

#TikTok: ownership changed, and the algorithm shifted toward search and longer watch time

TikTok's US ownership structure changed in 2025–2026 under the divestiture requirements that had been building since 2024, and the resulting new ownership introduced additional compliance review layers around ad targeting parameters and content categories that didn't exist under the platform's prior structure. [Directional] — the underlying ownership change is real and widely reported; the specific detail about added compliance-review processes for advertisers comes from a single trade-press source this research could not independently corroborate, so treat the general direction (more compliance friction under new ownership) as more solid than the specific mechanism described.

On the algorithm itself: TikTok increasingly rewards content built for its own search function — a reported 84% of TikTok searches happen during what the platform calls the "exploration phase," making keyword-optimised captions and on-screen text meaningfully more important for discoverability than in prior years. [Speculative] on the specific 84% figure, which this research could not trace to a named study; the broader direction — TikTok pushing search-and-discovery weighting alongside pure entertainment metrics — is corroborated by multiple independent sources and is consistent with TikTok's own public statements about expanding search features, so treat that broader direction as [Directional].

The algorithm also increasingly rewards longer content (1–3 minutes and beyond) with strong completion rates over short high-engagement clips, and watch-time/completion percentage now outweighs raw engagement rate in ranking — a video holding 80% of viewers for 20 seconds reportedly outperforms one holding 50% for 40 seconds. [Directional] — consistent across several independent creator- and marketer-facing sources, though none names TikTok's own published ranking documentation directly.

#What this means for an agency operating in 2026, concretely

Both platforms moved in the same direction for different reasons: less manual-execution value, more premium on things a platform's automation genuinely can't do — creative production that performs well when fed into an automated system, independent measurement that doesn't trust the platform's own attribution at face value, and content strategy built around each platform's actual current ranking logic rather than a 2020-era playbook. An agency whose entire pitch is "we'll run your Facebook and Instagram ads" is pitching a task the platform increasingly does by itself; an agency whose pitch is "we produce the creative and measure the result independently of what the platform tells you" is pitching something the platform structurally cannot replace, because it has no incentive to build a measurement layer that might show its own campaigns underperforming.