<aside> 🧭
Research snapshot: 11 September 2026. Meta changes fast. Re-check anything marked "reported" before you act on it.
How much to trust each claim: Items tagged [Meta] come from Meta itself (engineering blog, earnings, product changes confirmed by several trade outlets). [Consensus] means experienced practitioners broadly agree. [Heuristic] means a single agency or vendor rule of thumb. It's directional, not proof.
</aside>
The system Meta runs now isn't the one most accounts were built for. Three things changed how it works:
Operating rule worth adopting (Loomer): every change has to fix a problem you've proven with data. If you can't say what problem an edit solves, don't make it.
| When | Change | So what |
|---|---|---|
| 12 Jan | 7-day view and 28-day view attribution windows removed. Remaining windows: 1d/7d/28d click, 1d view, 1d engaged view. [Meta] | Year-on-year comparisons that used view windows no longer match up. Rebuild baselines on click. |
| 2025 → Feb 2026 | Advantage+ Shopping replaced by Advantage+ Sales (announced Feb 2025, rolled out through 2025). Advantage+ audience, placements and budget are on by default for Sales, App and Leads, and turning one off moves you to a manual setup. Multiple ad sets are now allowed. [Meta] Creating legacy ASC through the API was reportedly disabled in v25.0 (Feb 2026). | Old "ASC" playbooks and naming are out of date. |
| Mar | Click-through now means link clicks only. Likes, comments, shares and saves move to a new engage-through column (1-day). The video engaged-view threshold drops from 10s to 5s. [Meta] | Reported click conversions dip, mostly for retargeting-heavy accounts. |
| 2026 | Customer lifecycle setting in Sales campaigns: "Reach new and existing customers" (default) or "Acquire new customers", which auto-excludes your defined Existing Customers. The old existing-customer budget cap is gone. [Meta, via Loomer] | Now the main lever against spend leaking to returning customers. |
| Jul (reported) | Advantage+ can rewrite headline text on the image itself, generating variants. Custom audiences are grouped under Customer/Engaged labels, which unlock value rules. | Premium brands should fill in the Branding settings or opt out of rewrites. |
| Q2 earnings | Average price per ad +12% YoY (US & Canada ~+20%). New ranking models: +8.3% ad clicks, +15.7% conversions on Facebook. Advantage+ at a $75B annualised run-rate. [Meta] | Plan for higher CPMs. Meta is betting everything on automation. |
| Ongoing | Data retention: frequency breakdowns kept for only 6 months, unique and hourly data for 13 months. [Meta] | Export frequency and hourly data every month if you want history. |
Default: one campaign, one ad set per business goal, optimised for the deepest conversion event you can feed. [Consensus] Add a campaign only when it has a genuinely different job:
Separate retargeting campaigns are rarely worth it. Meta already sends roughly 20–25% of budget to warm audiences on its own (Loomer's account data), and a separate campaign mainly adds overlap. Some agencies add one only past about $1K/day and cap it under 10% of spend. [Heuristic]
The main disagreement is where testing happens: