Retargeting

Retargeting means serving ads to people who already did something with your business, usually visiting your site, watching a video, or adding to cart. The word is used across paid social and ad tech, and it is the habit on Meta, where the building block is a custom audience. A website custom audience is made of people who visited your site, collected through the Meta Pixel, which Meta may supplement with on-platform events. A common setup is to retarget people who visited in the last 30 days and have not converted yet, which is the example Meta itself gives. Google calls the same practice remarketing, and the two terms describe one posture. The pixel must fire correctly for any of this to work, so verify events in Events Manager before building audiences from them.

Under the hood, the Meta Pixel sends events back to Meta when someone views a page or takes an action, and Meta adds that person to the audience if they match your rules. Rules can include which pages were visited and how recently. If you limit the audience to recent visitors, people age out as time passes, so the audience is a moving window rather than a permanent list. That is why a short window gives a small, hot audience and a long window gives a bigger, cooler one with weaker intent. You can combine rules, for example product viewers who did not add to cart, to give the system a clear picture of where each person stopped.

Settings interact in useful ways. You exclude purchasers so you do not pay to pitch customers on the thing they just bought, and you set a separate budget so retargeting does not drain prospecting. Creative should differ from cold campaigns: reminders of the exact product, social proof, shipping or guarantee details, or a time-limited offer. Frequency limits matter more here, since the audience is small and the same people see ads repeatedly. Bidding on cost cap or ROAS goals is common, but results come out looking better than they are. Many advertisers also add a cap on how long a creative runs, rotating new ads in every few weeks so the same face is not shown for months.

That last point is the main mistake. Retargeting reaches people already close to buying, many of whom would have returned and converted without an ad. Platform-reported ROAS therefore credits the ad with sales it did not cause, and a business can look profitable while the real driver is the cold traffic, email, and organic reputation that filled the audience. Overspending is the other trap: the pool is capped by traffic, so pushing budget raises frequency and annoys people without adding sales. It also pays to remember privacy rules and consent banners, which can shrink audiences in some regions and make lists smaller than your traffic suggests. It also helps to cap how long any single creative runs, since a small audience tires of the same ad within a few weeks.

To read it, compare retargeting results with a holdout when possible: exclude a random slice of the audience for a few weeks and see how many convert anyway. Without that, treat the reported return as an upper bound and judge the channel on incremental sales. This week, check audience size and frequency, confirm purchasers are excluded, and shorten the window for your most expensive offers where buying decisions happen within days. Document your holdout design so the result is repeatable, and rerun it when you change creative or budget significantly. If a holdout is not possible, compare new customers per month before and after pausing retargeting for two weeks on a low-risk product line.