They already know you.
Now win them back.

In a regular search campaign, a keyword like "legal service" or "online course" costs a fortune. But when someone who already visited you searches that same keyword, the chance they'll close with you is four times higher. So why pay exactly the same price for it?

RLSA lets you run two bidding layers: one for an unfamiliar audience and one for an audience that already visited. For the familiar audience you raise the bid substantially and show a completely different message — "great to see you again," a specific offer based on what they saw, a message that assumes familiarity. In practice, this lowers the CPA on these keywords by 30%-50% compared to a generic search campaign aimed at the same audience.

Dynamic Remarketing is mainly known for stores that show "the product you saw yesterday." But the same principle works great for services, real estate, and education too — with a little creativity in building the Feed.

For a learning academy: a Feed containing the course name, the next start date, the number of available spots, and the price. Someone who viewed a Python course gets an ad showing exactly that course, with "3 spots left." For real estate: the project image, entry price, and handover date. The match is precise, so the CTR is far higher than a generic ad.

Money saved by avoiding poor targeting is money you can invest in an audience worth pursuing. Proper exclusion prevents three main wastes: showing an ad to someone who already became a customer, to someone who visited 18 months ago and has no chance of being a fit now, and to someone who only visited the "Terms of Use" page — traffic with no purchase intent whatsoever.

Exclusion also works the other way: you exclude the retargeting audience from the general acquisition campaign, so you don't pay twice for the same visitor. In campaigns we've reviewed, organized exclusion cut wasted spend by 15%-25%.

Regular retargeting says: "visited the site, show an ad." Trigger-based retargeting says: "visited, scrolled to 70% of the page, paused on the pricing section for 45 seconds — and now we show them an ad with a message about price and ROI."

It works through custom events in GA4 that stream into Google Ads as Audiences. You define: scroll_depth, time_on_section, video_progress, button_hover, and more — then build an audience around each one. This precision drives a CTR two to three times higher than generic retargeting.

Traditional retargeting works on an audience that already knows you. Lookalike expands that — Google finds people whose characteristics resemble those who already convert with you. It's a bridge between a cold acquisition campaign and warm retargeting.

Demand Gen (the successor to Discovery) is Google's best tool for this right now: it uses YouTube, Gmail, and Discover data to show ads to audiences resembling your converters — before they've even reached the site. Combined with Customer Match, you can build a complete loop: converters → Lookalike → Demand Gen → site → retargeting → conversion.

Under certain conditions, GA4 generates three built-in Predictive Audiences: Likely 7-day purchasers, Likely 7-day churning purchasers, and Predicted 28-day top spenders. These aren't audiences we built by hand — they're calculated by Google's ML model based on past behavior.

When you connect GA4 to Google Ads and use these audiences as Audience Signals within Performance Max or as an Observation Layer in a search campaign, the system knows to weight the clicks that are likely to convert. This is retargeting that the computer already predicted will buy, before the visitor even knows it.