Retargeting is the easiest channel to make look good and one of the easiest to get genuinely wrong. Reported ROAS on retargeting campaigns is almost always the best-looking number in the account, because you are advertising to people who already know you, some of whom were going to convert regardless. Making retargeting actually work, rather than just look good in a dashboard, requires more deliberate setup than most accounts give it.
A single retargeting audience covering everyone who visited your site treats a person who bounced off the homepage in four seconds the same as someone who added a product to cart or filled in three of four steps of a signup form. Those are different audiences with different objections and different messages needed to bring them back. Segmenting by depth of engagement, pages viewed, time on site, specific high-intent actions, and matching creative to that segment is what separates retargeting that recovers real revenue from retargeting that just re-shows your homepage ad to everyone.
This sounds obvious and gets missed constantly, especially on platforms that do not automatically suppress converted users the way some do. Continuing to spend on people who already purchased or signed up, unless you have a specific cross-sell or upsell reason to target them, is pure waste. Set up exclusion audiences as part of initial campaign setup, not as a fix you get to eventually.
Retargeting audiences are small relative to prospecting audiences, which means the same people see your ads repeatedly if frequency is not managed. Beyond a certain point, repeated exposure stops reminding people and starts irritating them, and you are paying for impressions with negative marginal value. Set explicit frequency caps and rotate creative on a real schedule rather than letting the same ad run indefinitely to the same shrinking pool of people.
A 7-day retargeting window makes sense for an impulse consumer purchase. It makes very little sense for a B2B product with a six-week evaluation cycle, where you want to stay present through the whole consideration period, not just the first week. Set your retargeting windows deliberately based on how long people actually take to decide, not based on whatever the platform suggests by default.
Because retargeting captures people already in your funnel, its reported performance overstates its incremental value more than almost any other channel. A holdout test, withholding retargeting ads from a portion of your qualifying audience and comparing conversion rates against the group that sees them, is the clearest way to find out how much of that impressive ROAS is genuinely incremental versus how much would have converted anyway. Doing this periodically, especially before increasing retargeting budget, keeps you honest about what the channel is actually contributing.
If your retargeting setup has never been segmented or tested properly, it is usually a quick audit with a clear payoff.