How remarketing works
A tag on your website — the Google tag, or Meta’s pixel — records that a browser was there and what it looked at. Those visits are gathered into lists defined by rules you write: everyone who opened a product page, everyone who started a form and abandoned it, everyone who reached the thank-you page, everyone who has not been back for a while. When someone on a list later browses another site, watches a video or searches again, your ad becomes eligible to reach them.
Every list has a membership window, and people fall out of it once that window passes without a new visit. Lists also have to reach a minimum size before a platform will serve to them, which is a real constraint on a low-traffic site or in a small market. Consent settings and browser restrictions on cross-site tracking shrink the lists further, so the audience you can actually reach is always smaller than your visitor count.
Why remarketing matters
Most first visits end without a purchase or an enquiry, and considered decisions — a course, a clinic, a property, a remittance service — usually involve several visits over days. Remarketing is how you stay reachable in that gap without paying cold prospecting prices again. Because the audience already knows who you are, the cost per conversion normally sits below prospecting, which is why the reports look so flattering.
It is also the cheapest place to answer objections. A cold ad has to introduce you. A remarketing ad can go straight to the thing that stopped the person: the price, the delivery time, the paperwork, the guarantee.
Where remarketing goes wrong
The most common failure is treating the whole audience as one group. Someone who glanced at the homepage and someone who abandoned a filled-in cart need different messages, and someone who already bought needs to be excluded entirely — nothing annoys a customer faster than being sold what they own. Exclusions matter as much as inclusions.
The second failure is believing the credit. Remarketing takes the last click from people who were coming back anyway, so its reported conversions overstate what it added. This is exactly what incrementality testing exists to check, and the honest way to size the channel is to turn it down and see what actually changes. Frequency is the third: without a cap, a small list sees the same creative until it stops working and starts irritating.
How to set it up sensibly
Build lists around depth of intent and recency rather than one catch-all visitor list, and set each window to match how long your buying decision genuinely takes — a same-day purchase and a six-month enrolment do not deserve the same one. Exclude converters, cap frequency, and write a different message for each stage instead of running one advert everywhere.
Then judge it properly. Hold the reported cost per conversion next to what the channel adds when it is paused, and treat the difference as the real number. The mechanics of building and segmenting these audiences sit in remarketing campaign management for anyone who wants the setup handled rather than explained.
