IntersectionObserver lazy vs native fetchMargin

What it does: two ways to lazy-load — your own observer hooks, or the ad manager's built-in lazy flag.

Setup effort: native 1/5 (one config line). Custom observer 4/5 (you manage thresholds, refresh interplay, race conditions).

RPM lift seen: roughly identical viewability gains (~20pts). The custom route only pays off when you need per-slot logic native can't express.

Pros
— Native: zero maintenance, vendor-tested
— Custom: fine control, can chain with refresh and bid-throttling

Cons
— Native: one global margin for all slots, no nuance
— Custom: every bug is now yours; easy to double-fire requests

Verdict: start native. Only build the observer if you've proven a per-slot need. Cheaper than custom, 90% of the lift. Lift: ★★★☆☆ — Risk: ★★☆☆☆
Best for: teams without a dedicated adops dev. Skip custom unless you can A/B it.