The one-sentence version
Search harvests demand that already exists. Display and its relatives create demand that does not exist yet. Neither is better; they sit at different points in the same process.
If you only ever harvest, you are limited by how many people happen to be looking this month.
When search wins
When the need is urgent and the search is explicit — emergency plumbing, a dentist accepting new patients, a locksmith. When the category is high consideration and people research before buying. And when your budget is small enough that you can only afford to talk to people already raising their hand.
Its ceiling is the volume of searches in your area, which is finite and known.
When display, social, and streaming win
When the product is discretionary or new, when you need to be remembered rather than found, and when you have already captured most available search demand and want to grow the pool.
They are also how you defend a market. If a competitor moves into your county, search alone is a reactive posture.
The blended reality
They interact. Households exposed to streaming or display convert on search at measurably better rates, and your brand search volume rises during upper-funnel flights. This means channel-level cost per lead is misleading: search will always look cheapest because it gets credit for demand something else created.
Judge on blended cost per outcome across the whole program.
A default split
For a local business with an established offer: roughly 60 percent to demand capture, 30 percent to demand creation, 10 percent to testing something new. Skew harder toward capture when cash is tight and toward creation when you are trying to grow rather than defend.
Revisit quarterly, not weekly.